VU Research Portal

Extracting Erosion and Exhumation Patterns from Detrital Thermochronology Gemignani, L.

2018

document version Publisher's PDF, also known as Version of record

Link to publication in VU Research Portal

citation for published version (APA) Gemignani, L. (2018). Extracting Erosion and Exhumation Patterns from Detrital Thermochronology: an example from the eastern Himalaya.

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?

Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

E-mail address: [email protected]

Download date: 08. Oct. 2021 Rivers transport key information that can be used as a proxy to infer the evo- lution of the landscape of eroding mountain ranges. Detrital thermochronology aims at shedding new light about the timing of past tectonic events, fi xing their rates. The Himalaya is a natural laboratory where studying those processes and contemplate the magnifi cence of the Nature.

ISBN: 978-90-9030865-4 Extracting erosion and exhumation patterns from detrital thermochronology Lorenzo Gemignani Extracting erosion and exhumation patterns from detrital thermochronology an example from the eastern Himalaya

Lorenzo Gemignani Extracting erosion and exhumation patterns from detrital thermochronology an example from the eastern Himalaya

Lorenzo Gemignani This Doctorate project is part of an international multi-partner MSCA Initial Training Network project called iTECC (“Investigating Tectonic, Erosion and Climate Cou- plings”) and has been funded by the European Commission (FP7). The Early Stage Researcher grant n. 316966 had the duration of three years.

Layout and Cover by Lorenzo Gemignani Printed by: Ipskamp, printing ISBN: 978-90-9030865-4 Pictures by Lorenzo Gemignani Cover pictures: eastern Himalaya, China, ASTER GDEM, NASA. VRIJE UNIVERSITEIT

EXTRACTING EROSION AND EXHUMATION PATTERNS FROM DETRITAL THERMOCHRONOLOGY an example from the eastern Himalaya

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnifi cus prof.dr. V. Subramaniam, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Bètawetenschappen op woensdag 28 maart 2018 om 11.45 uur in de aula van de universiteit, De Boelelaan 1105

door Lorenzo Gemignani geboren te Massa, Italië

promotoren : prof.dr. J.R. Wijbrans prof.dr. P. van der Beek copromotor: dr. Y. Najman

Members of the reading committee:

Prof. dr. Ronald T. van Balen (chairman) Dr. Christiaan .J. (Kay) Beets Prof. dr. Barbara Carrapa Prof. dr. Klaudia F. Kuiper Prof. dr. Franz Neubauer Prof. dr. Massimiliano Zattin A Sofi a, il piu’bell’átto creativo...

Kali Gandaki river, Mustang, Nepal

“Considerate la vostra semenza: Fatti non foste a viver come bruti, Ma per seguir virtute e conoscenza”. “Consider your origins: you were not born to live as brutes, but to follow virtue and knowledge.”

Dante Alighieri, Inferno, Canto 26 Contents

Summary 9

1. Introduction 11

2. Principles of thermochronology 21

3. Extracting information on the spatial variability in exhumation 47 rate stored in river detrital age distributions.

Published as: Braun J., Gemignani L., van der Beek P., (2017). Extract- ing information on the spatial variability in exhumation rate stored in river detrital age distributions. Earth Surf. Dynam. Discuss., https://doi. org/10.5194/esurf-2017-42. 4. A new detrital mica 40Ar/39Ar dating approach for provenance and ex- 69 humation of the Eastern Alps.

Published as: Gemignani L., X. Sun, J. Braun, T.D. van Gerve, and J.R. Wijbrans (2017). A new detrital mica 40Ar/39Ar dating approach for provenance and exhumation of the Eastern Alps. Tectonics, 36, 1521– 1537, doi:10.1002/2017TC004483. 5. Present-day and long-term erosion of the eastern Hima- 97 laya as detected by detrital thermochronology.

Submitted as: Gemignani L., P. van der Beek, J. Braun,. Najman, Y., Bernet, M., Garzanti, E. Wijbrans, J.R. (Submitted), Present-day and long-term erosion of the eastern Himalaya as detected by detrital ther- mochronology. Earth Planet. Sci. Lett.

6. Improving the precision of mica Ar-dating on smaller and 129 younger muscovite grains: implication for provenance studies.

Prepared for submission as: Gemignani L., K. Kuiper, A. Santato, and J.R. Wijbrans. (Submitted), Improving the precision of mica Ar-dating on smaller and younger muscovite grains: implication for provenance studies. Chemical Geology.

7. Synthesis and conclusion 155

Bibliography 163

Acknowledgements 181 Summary The Himalayan is rapidly emerging as a natural laboratory for studying intra con- tinental deformation related to continent-continent collision coupled with climatic- driven erosion. The extreme topography of the Himalaya forms a barrier, differentiating climatic conditions and erosional patterns between the two sides of the belt. The interaction between crustal-tectonic and climate-erosional processes is borne out by the present day topography resulting in the bent patterns, and deep incised gorges of the major Himalayan rivers: the Indus and the Brahmaputra. In the Eastern Himalaya, the Namche Barwa syntaxis is exhuming and eroding faster when compared with the central sectors of the belt since Lete-Miocene Pliocene. In the Namche Barwa syntaxis, cooling ages record a major exhumation pulse at ~5 Ma as young as ~106 a depending on the applied thermochronometer. The sediment fl ux derived from the Namche Barwa area is estimated at ~70 % of the total sediment fl ux in the Brahmaputra when reaching the foreland. However, the young thermochro- nometric signature (< ~5 Ma) downstream the Namche Barwa syntaxis seems to be suppressed by older age peaks derived from sediment components from tributaries draining the more central Himalayan rock units. This discrepancy in the modern river sediment age distributions is refl ected in the syn-sedimentary basins where ages are as young as ~6-7 Ma and older. The difference between the modern rate of sediment eroded from the Namche Barwa syntaxis and its downstream evolution is not com- pletely understood. The implied question, then, concerns on how the detrital records can be used to assess the transient change in exhumation/erosion in a dynamic moun- tain belt. The present work is aimed at these outstanding questions. For shedding new light on these issues, we proposed the following approach: We fi rst studied the consistency of the detrital mica 40Ar/39Ar and zircon fi ssion tracks dating approach as tools to characterize the tectonic history of source rocks within the river network of an evolving mountain range. We than developed a numerical linear inversion of the age distributions, the “mixing model” method (Chapter 3). The method was tested on available literature data from the eastern Himalaya. Our results show how we can get averaged present-day erosion estimates and the exhumation sig- nature at the superfi cial rocks from detrital age distributions. This approach was tested for modern river sand sediments obtained from 19 river’s catchments in the Eastern Alps (Chapter 4). In the second part of this thesis, we present the outcomes of studies where we analyzed the modern river sediments of the eastern Himalayan (Chapter 5) using two different thermochronometers (mica 40Ar/39Ar and zircon fi ssion-tracks).

9 Summary Understanding the tectonic evolution of the eastern Himalayan syntaxis is key to differentiating different models of coupling between tectonics and erosion. The multi-proxy approach allowed to produce a synoptic cooling-age map of the eastern Himalaya that highlighted the spatial variation in exhumation rates of the contribut- ing sources to the fl uvial system. The relative present-day erosion estimates were then compared with a quantitative estimate of steady-state exhumation rates required to produce major age components observed in the detrital samples. We noticed that whilst the young age peak is distinctive for the studied minerals and endures many kilometers downstream, the young mica population is much more suppressed, both in proximal and distal samples. The potential effect of dilution of the analysed target minerals has been addressed by looking at different grain-size frac- tions in the last part of this work (Chapter 6). We show that grain-size variability can bias age distributions when studying large catchment areas, such as the Brahmaputra foreland. The most important fi nding is that multiple grain-size analysis allows hav- ing a better resolution of the sources drained in the catchment area. This thesis has explored the exhumation/erosion patterns of two dynamically evolving mountain ranges characterized by two distinct spatio-temporal evolutions. The analysis of multi-proxy thermochronology shed lights on the dilution processes governing the Himalayan foreland for different target minerals. We have demonstrat- ed that a combination of multi-proxy thermochronology, numerical modeling, and analytical technique improvement provides new opportunities to study the evolution of the transient response of mountain belts to changes in boundary conditions on geo- logical (Ma) time scales.

10

Chapter 1 Introduction 1.1 Introduction 1.1.1 Earth surface processes and tectonics: a dynamic interaction 1 The idea that atmospheric, surfi cial and deep Earth processes may be linked in a complex interplay in active orogenic belts to build topography is a leading topic in Earth Science. Tectonic (rheologic and deformational), surface erosional and climate processes compete together to shape Earth’s surface. The surfi cial mass removal by erosional processes is coupled with the uprise of material from beneath following mass balance laws. Central questions that are posed to the researcher focus on the issue of leads and lags: is convergence in active mountain belts pushing up matter, or is removal from the exposed surface by erosion triggering re-equilibration of the isostatic balance, and hence exhumation? In the last decades, examples of surface evolution linked to tectonic regime and climate changes have attracted the attention in a series of paradigm (Molnar and Eng- land, 1990; Burbank, 1992; Zeitler et al., 2001; Wang et al., 2014). In the Himalayas for example, the interplay between tectonics, erosion, and climate has been used to explain the Cenozoic evolution of the belt (Beaumont et al., 2001; Koons et al., 2013). For its various length- and time scales, this phenomenon needs multiple approaches of study, ranging from crustal rheology, heat transfer and fracture mechanics, to land- scape evolution and atmospheric circulation patterns. Surface uplift (changes in mean elevation with respect to a fi xed horizon or refer- ence frame, following Molnar and England, 1990) and erosion are responsible for the exposure of fresh unweathered rocks to the surface. Climate variability can trigger changes in weathering and erosion. The Cenozoic, for example, is characterized by a monotonic cooling of the Earth’s surface on a global scale (Zachos et al., 2001), and the Plio-Pleistocene period represents one of the best-studied examples of a climate shift documented on Earth. It has been suggested that this period is associated with a global increase of erosion rates that is more pronounced in the glaciated mountain ranges of the mid to higher latitudes than at lower latitudes (Zhang et al., 2001; Her- man et al., 2013), although this hypothesis remains contentious. Moreover, climate- controlled chemical weathering of rocks, which are subsequently eroded and trans- ported via fl uvial systems downstream into the foreland basins, has the potential to alter globally the atmospheric geochemical composition. As an example, these coupled processes associated with the formation of topo- graphic relief caused by the Indian-Asian collision since the early Tertiary led to in- tensifi cation of the Indian-ocean monsoon, causing increased silicate weathering and organic carbon burial, which account for the substantial decrease in the CO2 atmos-

13 Introduction pheric concentration that has had an impact on Cenozoic global cooling (Raymo and Ruddiman, 1992; Derry and France-Lanord, 1996). Thermochonometric techniques can be used to obtain information about the cool- ing/exhumation history of rocks exposed at the surface. Surface bedrocks are limited by the possibility of their access. Limiting factors are the asperity of the mountain region or areas covered by ice or with geo-political instability. The analysis of detri- tal sediments eroded and exported from an individual river catchment area (known as detrital thermochronology) provides a potentially surface-integrated signal of the exposed rock in that catchment. One of the most remarkable advantages of detrital thermochronology stems from the fact that it is possible to obtain information from inaccessible remote areas by analysis of modern river sands (Enkelmann and Ehlers, 2015). This thesis is devoted to exploring how multi-proxy detrital thermochronology can be used as a tool to infer estimates of spatial-temporal variation in catchment- averaged erosion rates in mountain ranges. Previous studies have applied detrital ther- mochronology to foreland-basin sedimentary sequences, as well as to modern river and glacial deposits (Von Eynatten and Wijbrans, 2003; Bernet and Spiegel, 2004; Brewer et al., 2006; Carrapa et al., 2014; Enkelmann et al., 2015). These previ- ous work shed light on the ease and applicability of this technique for provenance analysis and to derive exhumation rates. However, the evolution of signals obtained from the detritus of modern river sediment can be affected by several processes dur- ing transport from source to sink in a river. Mineral comminution due to mechanical hydraulic sorting can affect the detrital signal of the sediments deposited in foreland basins, which in turn will lead to increased uncertainty in the interpretation that can be based on such signals. The potential differential downstream dilution of such signals can also strongly affect interpretation and is as yet not fully understood. I address this issue in the present thesis by systematically analyzing such processes, using a multi- mineral detrital thermochronology approach focused on two active mountain ranges; the Alps and the Himalaya. In this thesis, I explore how different thermochronometric sources mix in riv- ers and which information we can extract for constraining provenance and deriving exhumation histories. The fundamental underlying concept from which we try to un- derstand processes is the cooling age; the age recorded by the mineral or the time of its cooling through the closure temperature (Dodson, 1973). Furthermore, from the analysis of present-day mixing of different sources, we can infer estimates of present- day erosion rates. Present-day erosion and long-term exhumation will provide a com- mon thread through all the chapters. After initial testing of the method on the well-

14 Chapter 1 studied Eastern Alps mountain range, I focus attention on the downstream evolution of the signal from one of the most rapidly uplifting and exhuming mountain ranges on Earth: the Namche Barwa syntaxis of the Eastern Himalaya. In the fi nal part of this 1 thesis, I shed new light on how the improvement of analytical sensitivity of noble gas mass spectrometers used for the 40Ar/39Ar dating technique helps to measure small ion beam signals from small or young samples. The capability of measuring smaller and younger grains can be used to infer more precise constraints on exhumation rates using detrital thermochronology. Is the detrital age-distribution in a sample a function of the grain size? I explore this important topic using fi ve samples from the Siang-Brahmaputra River, showing how grain-size variability can affect observed sample age distribu- tions and how the choice of the grain size should be leveled to the distance in the river from the source of the analyzed grain-size fraction.

1.1.2 The eastern Himalaya: a dynamic laboratory on Earth Many lines of evidence show that the evolution of the Himalayan belt has been characterized by a complex interplay between tectonic and climate-driven forces. Fig- ure 1.1 displays the Himalaya and the Tibetan plateau in map view. It also shows a schematic geological map of the Eastern Himalaya, indicating the areal distribution of the characteristic young thermochonometric signal of the eastern syntaxis. In the east- ern Himalayan syntaxis, thermochronological data suggest that a dramatic increase in exhumation rates occurred in the Namche Barwa massif starting about 4-7 million years ago, and accompanied by high erosion rates of about 10-13 mm/yr (Singh and France-Lanord, 2002; Stewart et al., 2008; Lang et al., 2016). The extreme exhuma- tion rates of these two corner indenters are refl ected in the modern-day topography by the peculiar bending paths of the two major rivers draining southern Tibet: the Indus and the Brahmaputra Rivers (Figure 1.1a). Figure 1.1c shows a schematic cross-sec- tion of the crustal-scale antiform of the Namche Barwa, indicating the rapid advection of middle and lower crust and the exhumation and sediment-transport pathways (after Seward et al., 2009). The “tectonic aneurysm” model (Zeitler et al., 2001a, b; 2014; Koons et al., 2013) for the evolution of the syntaxes argues for the coeval develop- ment of high relief and deeply incised gorges characterized by exceptionally young, high-temperature metamorphic massifs developed in an area of weak hot crust. How this young signal is recorded in the Himalayan foreland basins is not clear, leading to disparate estimates of the precise timing of onset of rapid tectonic uplift and of the evolution of major river (Lang et al., 2016). In fi gure 1.2, I show the schematic evolution of the Eastern Himalayan syntaxis

15 Introduction from the Plio-Pleistocene to the present-day, explaining the possible evolution of the Yarlung-Brahmaputra River since the Miocene (Modifi ed after Seward and Burg, 2008). The evolution of the crustal-scale antiform at the Namche Barwa and the drainage evolution of the Yarlung-Tsangpo-Siang-Brahmaputra river is a debated topic (Zeitler et al., 2001; Bracciali et al., 2015; Lang et al., 2016). In particular, the timing of the capture event of the Yarlung-Tsangpo River by the Brahmaputra River is unclear; various hypotheses postulate this event at Middle- or Late-Miocene times (Lang et al., 2016). This onset has important implications for the thermo-mechanical feed- backs leading to the exhumation of deep crustal material: are these increased by the river incision (Zeitler et al., 2001; Koons et al., 2013)? Syn-tectonic sediments in the foreland basin present a temporal discordance in the Eastern Himalaya. Distal (>1000 km) sedimentary records of the Brahmaputra do not record rapid exhumation of rocks in the Namche Barwa syntaxis (NB) before1-2 Ma (Bracciali et al., 2016), whereas more proximal deposits show this characteristic signature as early as >6 Ma (Govin et al., 2016; Lang et al., 2016). The reason for this discrepancy is not clear: is the earlier signal diluted downstream leading to a bias in the more distal records? In a more general view, measured exhumation rates can be compared with mod- el predictions and the timing of climate, tectonic or drainage pattern change to as- sess possible feedback relationships. The Himalaya is, therefore, a natural laboratory where it is possible to explore the interplay of those processes.

16 Chapter 1 a. 1

35oN

Indus Tibetan plateau

30o Yarlung-Tsangpo

Ganga Brahmaputra 75o 80o 85o 90o 95oE

b. 92o 94o 96oE Greater Himalayan sequence Thetys 50 km Lesser Himalayan ~3-5 Ma 30oN sequence Lhasa block sediments and metasediments inrouded by Gangdese plutons ~1-3 Ma NBNB Nam la thrust Jiali Fault zone Suture zone Suture zone o 28 MCT STDS

c. NNW SSE river evacuation out of the sytem Namche Barwa 7000 7000 6000 6000 5000 Rapid erosion 5000 4000 4000 3000 3000 m 5 km m Rapid advection focused strain 0 T~350 C

Figure 1.1. a) Generalized topographic map of the Himalaya and Tibetan plateau, modifi ed af- ter Hodges (2000). b) Simplifi ed geologic map of the eastern Himalaya indicating the regions of very young thermochronology (mica 40Ar/39Ar and zircon (U-Th)/He) ages. c) Schematic cross-section (location in fi g. 1.1b; dotted black line) of the Namche Barwa antiform; the colored arrows summarize the trajectories associated with a fully evolved tectonic aneurysm.

17 Introduction a.

Yarlung suture zone

folded suture

Initial dome

~20 km ~ 4-7 Ma ~ 2Ma Today b. Jiali-Parlung Jiali-Parlung Jiali-Parlung

?? Paleo Tsangpo-Brahmaputra ?? Tsangpo Tsangpo suture zone Siang Siang ??

~20 Ma ~4-7 Ma Today

Figure 1.2. a) The growth of the Namche Barwa since ~4-7 Ma (cartoon not to scale). b) Tsangpo-Siang River evolution in relation to the Namche Barwa domal pop-up. The dotted line indicates the progressive folding of the Himalayan suture zone.

1.2 A multi-proxy detrital approach Thermochronology on detrital minerals is a key tool to inform about cooling/ex- humation history of the catchment area of rivers to a foreland basin. The use of mod- ern river sediments is particularly advantageous as it provides a convenient way to assess actively exhuming and remotely accessible belts. Multi-proxy detrital thermo- chronology provides information on past cooling of the mountain range, on sediment provenance and on possible dilution of sources downstream in the river system. The latter effect has an important implication on the correct interpretation of syn-orogenic sedimentary sequences that may be biased by downstream modifi cation of detrital sig- nals. I devoted part of this project to study the effect of this variability of the detrital signal in function of the grain size. The full dataset of this thesis comprises 34 samples collected from the Eastern Alps and Eastern Himalaya drainage systems. Detrital muscovite, biotite, and zircons were separated using standard mineral separation techniques at the Vrije Universiteit Amsterdam. 40Ar/39Ar analyses were performed on selected crystals of biotite and white micas of different grain sizes with a window of closure temperatures ranging

18 Chapter 1 from ~480°C to ~300°C. Zircon fi ssion-track analyses (ZFT) were performed on key samples providing closure/cooling information through the ~350°C to ~140°C tem- perature interval. 40Ar/39Ar age distributions were obtained by single-grain total fusion 1 analysis, using the latest-generation mass spectrometer Helix MC plus. In total, I gen- erated about 1300 new single-grain total fusion analyses on white mica and about 600 zircon fi ssion-track ages from the Eastern Himalaya. Additionally, I performed 342 single-grain muscovite and 258 biotite analyses for the Alps. I used standard statistical techniques such as the Probability Density Plot (PDP) and Kernel Density Estimator (KDE) to plot muscovite and biotite 40Ar/39Ar and zircon fi ssion-track ages using a Java-based Density Plotter program (Vermeesch, 2012). We have designed and tested a new method that can be used as a tool to infer estimates of long-term and present-day erosion of multiple sub-catchments along a sampled river. The method explores the advantage of using age distributions as a “markers” or “tracers” that can inform us on the amount of mixing acting today, which is directly proportional to the present-day erosion rates. This method differs from the previous quantitative analyses that used detrital age distribution, hypsometry and thermal models to predict erosion rates and their spatial distribution assuming steady- state between uplift and erosion (e.g., Brewer et al., 2003; Brewer and Burbank, 2006; Ruhl and Hodges, 2005; Huntington et al., 2006). The steady-state hypothesis does not account for the likely transient response of topography to changes in tectonic or climatic forcing. We will show that although the method predicts only fi rst-order erosion patterns, we are able to infer the lateral variation of erosion between adjacent catchments areas.

1.3 Thesis outline In this thesis, I explore the advantages and limitations of the multi-proxy detri- tal thermochronology approach. In particular, I make use of 40Ar/39Ar radio-isotopic techniques on biotite and white mica and of the lower-temperature zircon fi ssion- track thermochronometer. In Chapter 2, I introduce the fundaments of radio-isotopic chronology techniques. In this chapter, I discuss the advantages and limitations of the applied methods. Furthermore, I explore the applicability and future perspectives. In Chapter 3, we provide the full analytical explanation of the linear inversion and mixing model of the age distribution applied to infer estimates of present-day erosion from rivers sediments. The method is tested on available literature data from catch- ment draining the Eastern Himalaya. At the end of the chapter, we provide a simple implementation of the method in R.code within a Jupyter notebook (http://jupyter. org/) with the example tested in the chapter.

19 Introduction In Chapter 4, I explore the 40Ar/39Ar dating technique on muscovite and biotite as a proxy to constrain sediment provenance in a river system. For this purpose, we used well-constrained drainage catchments of limited areal extent in the Eastern Alps with respect to the geochronological ages of its bedrock. The 40Ar/39Ar muscovite and bio- tite ages of modern river sands are generally consistent and can be used as a powerful proxy to identify sediment provenance and the extent of cooling events. In the following Chapter 5, we apply the detrital thermochronology approach to infer the spatial variability of exhumation rates from the Eastern Himalaya. This area is characterized by extremely rapid and recent exhumation of the Namche Barwa syn- taxis, coupled with a signifi cant material fl ux into the major modern fl uvial system, the Yarlung-Siang-Brahmaputra. Using a multi-proxy approach (muscovite 40Ar/30Ar and zircon fi ssion-track) on the modern river detritus, we study the downstream evo- lution of the characteristic young signal with respect to the contribution of tributaries adding additional age signals and thus diluting the Namche Barwa signal in the main river trunk. We show how the young mica population becomes heavily diluted. To estimate present-day erosion rates in the catchments, we apply a mixing model based on the linear inversion of the binned age distributions. Chapter 6 focuses on the 40Ar/39Ar dating technique of smaller and younger sam- ples of muscovite using the new generation mass spectrometer (Helix MC plus). This chapter sheds new light on the variability of the age distribution of a sample in func- tion of the mineral grain size. Five detrital modern river samples from the Brahma- putra river were analyzed using fi ve different grain fractions ranging from 125 to 1000 μm. Analyses on the Helix MC plus mass spectrometer were performed using the latest generation 1013 Ohm amplifi er installed on the H2 and H1 Faraday cups of the multi-collector array that are used to measure the 40Ar and 39Ar ion beams respec- tively. Our new data show how grain size variability can bias the age distribution of a sample and subsequently the interpretation of the spatial distribution of exhumation.

20 Chapter 1 Chapter 2 2 Principles of Thermochronology

21 Principles of Thermochronology Abstract Detrital thermochronology, the core method explored in this thesis, is a relative- ly recent technique that can be applied to synorogenic sediments (both sedimentary basins and modern river detritus) to constrain the spatial time evolution of an “ac- tive” tectonic hinterlands. Detrital thermochronology can combine a large number of thermochronometric systems each with a different closure temperature. Some of the most commonly used thermochronometers are muscovite, hornblende, biotite, and microcline 40Ar/39Ar and zircon, apatite, and titanite U-Th/He and fi ssion-tracks. Each of these systems records different aspects of a geological history due to differences in their closure temperatures. The closure temperature of a geochronological system, as fi rst defi ned by Dodson (1973), is the temperature of the rock at the time of its calculated isotopic age. Isotopic dating developed in the 1950’s as the direct result of the discovery of radioactivity at the end of the 19th century. In this chapter, I will il- lustrate the analytical techniques that I used during this Ph.D. project to constrain the thermochronological history of selected sectors of two collisional belt: the Alps and the Himalayas. First, I will briefl y describe the historical background of the methods followed by the principles, applications, and limitations.

2.1 Introduction The precise assessment of the time of events is crucial for the understanding of the temporal evolution of planetary bodies, and of the processes that regulated the shaping of the present confi guration of the Earth. Isotope geochronology is the sci- ence of determining the age of rocks and allows placing observations in a temporal framework. Geochronology can be coupled with geologic analysis to inform about plate tectonic processes that take place today and in the past. Recently, the integration of absolute dating with sedimentary sequences interpre- tation has sparkled critical relationship among the tectonic, biological and climatic evolution of the Earth system. Geochronology methods are routinely used to constrain the exhumation of vast and complex geological settings. In recent years, the analysis of the samples that can be obtained radioisotopic dating methods became relatively fast and cheaper. Recent studies focussed to increase the precision of the analytical error using the 40Ar/39Ar technique (Kuiper et al., 2008a; Schmitz and Kuiper, 2013). The remainder of this section presents the fundamentals of radio-isotopic geochem- istry and its application to geological and planetary problems. In this chapter we will describe in detail the radioisotopic techniques that we applied to different geological settings, starting from an introduction of the historical milestones that characterized

23 Principles of Thermochronology the development of those techniques, their advantages, and their limitations. Isotope geochronology is the science of determining the absolute age of the Earth and planetary materials. Geochronology is based on the measurement of the accu- mulated amount of radiogenic isotopes over a determinable time period. Most of the isotopic dating techniques are based on mass spectrometry method. Isotopes of the same element, e.g. 40Ar and 39Ar, have the same number of protons (same charge) but a different number of neutrons (different mass number). Radiogenic isotopes have an unstable nucleus that releases energetic radiation (decay). Thermochronology is a branch of geochronology that records the thermal evolu- tion of rocks through the cooling ages obtained from a defi ned thermochronometer. A number of nomenclature conventions are associated with this term and some of the most common are (Reiners, 2005): a) Thermochronometer: a radio isotopic system that presents radioactive feature and the mineral in which it is found. b) Thermochro- nology: the technique, the method of application of a known thermochronometer to infer the thermal history of rocks or minerals. In the last decades, thermochronology has been applied to sedimentary sequences and modern river detritus to infer averaged basin exhumation rates, cooling history and provenance of detrital minerals (Wagner et al., 1979; Von Eynatten et al., 1996; Hodges, 2005). Detrital thermochronology depends on dating techniques which allow the dating of single grains (Hodges, 2005). Petrochronology is a complementary branch of the thermochronology that re- fers to the specifi c use of geochronologic age in petrology. As argued by Engi et al., (2017): “petrologist are not simply interested in thermal histories, but in the chemical and baric evolution of magma crystallization or metamorphic pressure-temperature evolution”. The need of understanding the rocks formation processes and not only their cooling evolution in a temporal framework has led to coining this new discipline. Isotope provenance is a branch of geochronology in which isotopic systems are used as a proxy to infer the origin of eroded sediments. For provenance purpose, single-grain age dating is routinely used for assessing the provenance using multi- minerals dating approaches. Some of the most applied techniques are U-Pb dating of zircons, fi ssion tracks analysis of zircon and apatite, 40Ar/39Ar dating of white mica, biotite, and(U-Th)/He on apatite and zircon. Generally, U-Pb dating is considered to constrain crystallization ages whereas the other techniques yield in most cases the cooling ages (Bernet and Spiegel, 2004).

2.2 Historical background The fi rst observations of radioactivity were reported in 1896 by the physicist

24 Chapter 2 Henry Becquerel while performing experiments with a phosphorescent uranium-rich salt. Marie Skłodowská-Curie fi rst coined the word “radioactivity” on the basis of the discovery of the activity of the newly discovered element radium. At the turn of the 20th century, the formulation of the radioactive decay law was proposed by Ruther- ford and Soddy (1902). A few years later, using Rutherford’s predictions, geologist Arthur Holmes performed the fi rst radiometric dating of rock minerals by using the 2 association of lead with uranium (Holmes, 1911). Despite the relative accuracy of Holmes analytical chemical lead procedure, the discovery of isotopes was formalized by Soddy (1913), leading to a potential step improvement in accuracy. Geochronology received a formal recognition in 1923 when the National Research Council of the U.S. Academy of Sciences formed the committee on the Measurement of Geologic Time by Atomic Disintegration (Faure, 1986). Radioactivity is a natural and spontaneous process that occurs at the scale of an atomic nucleus. This process occurs when an unstable atom of an element emits or ra- diates excess energy in the form of particles or an energy quantum. Some of the most important emissions are α, β, and γ-rays (fi g. 2.1), which differ in their penetrating powers through matter. The emission of such rays is observed in materials that are ra- dioactive. A radioactive nucleus will decay, i.e. emit particles of energy quantum until the product of decay reaches stability. This process may involve several steps before stability is reached. The most common radiation forms were classifi ed as: α decay. A group of radiogenic elements, mostly very heavy elements such as uranium, thorium, samarium and radium, decay with the spontaneous emission of a α particle from their nucleus (fi g. 2.1a). Alpha particles are composed of two protons and two neutrons and have a charge of +2 (Faure, 1986). Alpha particles are identi- cal to the nucleus of a helium atom 4He. The emission of an alpha particle from an element reduces both the atomic and neutron number by two and the mass number by four (Faure, 1986). The α ray is a highly ionizing form of particle radiation with a low penetration depth in most materials. Alpha particle will travel no more than few centimeters in air and can be stopped by a sheet of paper. β decay. A large number of unstable atoms decay by emitting negatively charged β particles (β- negatron) and neutrinos from the nucleus. The process follows the reac- tion n → p + e νe and is often associated with the emission of radiant energy (gamma ray) (Fig. 2.1b). The electron is then expelled from the nucleus as a negative high energy particle resulting in an increase of one of the atomic number and a reduction of one of the neutron number. E. Fermi fi rst named those particles with no charge but varying kinetic energy “neutrino”. The neutrinos emitted during β- decay is called

25 Principles of Thermochronology antineutrinos and are different from neutrinos formed during β+ decay. Beta particles interact less with the standard material than alpha particles and thus can penetrate the human body with serious health effects. γ decay. After a decay reaction, the nucleus of an element it is often in an “ex- cited” state (fi g. 2.1c). This energy excess is compensated by emitting a quantum of electromagnetic radiation called gamma ray. Gamma decay is a radioactive decay process that does not produce any nuclear transmutation. In gamma decay, a nucleus changes from a higher energy state to a lower energy state through the emission of electromagnetic radiation (photons). Gamma rays have the highest penetrative power of all forms of radiation. Spontaneous nuclear fi ssion is the fi ssion of heavy unstable nuclides (in an excited state) that can split into two or occasionally three smaller daughter nuclei (fi g. 2.1d). In addition, fi ssion events can lead to the emission of alpha particles, neutrons and a high amount of energy released by the emission of other particles (Faure, 1986). All elements have known radioactive isotopes, and a substantial number of radio- active isotopes can be found in the natural environment. Geochronological techniques are based on the instability of isotopes occurring in nature. An increasing range of isotopic dating techniques allow the scientifi c community to address geological ques- tions including, for example, the origin and evolution of the Earth, and other planetary bodies and to date the time scales of human evolution (Deino et al., 1998).

26 Chapter 2 Figure 2.1. a) Alpha decay for a) the 238U/234Th with emission an alpha particle (helium nucle- us), the next steps of the U-Th parent daughter α decay are not illustrated here. nucleus nucleus particle b) beta decay of the 14 C to 14N 238 234 4 92U 90Th 2He by converting a neutron (light b) grey) to a proton (dark grey+) an electron and an antineutri- 2 + no. c) example of gamma ray emitted from an excited nu- parent daughter electron anti cleus. d) nuclear fi ssion real- 0 nucleus nucleus -1e neutrino 440 ized by a parent nucleus in an 6C 7N 0v excited state.

c)

parent nucleus daughter γ ray (excited state) nucleus

d)

parent neutrons (excited state) daughter

2.3 Geochronology and Thermochronology: a key difference The analytical formulation of the closure temperature theory was formalized by Dodson (1973) and defi ned as the temperature of a given system at the time of its ap- parent age. The products of a radioactive decay within a crystal are highly mobile and are retained only if a sample is cold enough to retain the energy through the crystal lattice. Each termochronometer has a defi ned closure temperature (or range of tem- peratures) Berger and York (1981) applied 40Ar/39Ar dating experiments to constrain time and temperature using different K-bearing minerals. In their work Berger and York, (1981) refer to geothermometry by means of the use of kinetic data to interpret different isotopic systems in terms of thermal histories. Commonly used thermochro- nometers and their nominal closure temperatures are summarized in table 2.1. Rocks have a characteristic behavior in function of changes in temperature and tectonic forces the recognition of thermal histories of rocks can provide critical con- straints on geological processes in active orogens. The exact distinction of geochro- nology and thermochronology is sharp. Thermochronology focuses on the processes and rates of a tectonic event across an orogen and can be applied to solve both time and rates of the tectonic pulses (Reiners, 2005), whereas geochronology relies on the

27 Principles of Thermochronology time of crystallization of a volume of rock. Geochronology has been applied to rocks since the beginning of the 19th century to date rocks formation (e.g. Holmes, 1913). The thermal effects on radioisotopic ages were recognized a few decades later (Hurley, 1954). Using this fundamental discov- ery, the concept of comparing dates constrained with different radiometric techniques the cooling history of rocks was intuited starting from the early 1960s in few inter- esting pioneering work both in the Alps (Amstrong et al., 1966) and the Himalaya (Krummenacher, 1961). Armstrong et al., (1966), used Rb-Sr and K-Ar on biotite, white mica, and sanidine minerals to defi ne their ages as the time of the cooling of the rocks during Alpine metamorphism. They also predicted, using different isotopic systems with two different closure temperatures, how the two system can be different and can be used to infer the rate of cooling of a certain region. Later, a work by Wag- ner et al.(1979), used apatite fi ssion tracks data to constrain the exhumation rate of the Bergell intrusive body (Central Alps) and to recursively restore the original position of a set of boulders found in the Po plain (Italy), eroded during late Oligocene. Table 1. Summary of the most common thermochronometers.

Decay Closure Mineral References system temperature (oC)

40Ar/39Ar Hornblende 500±50 Harrison (1981)

40Ar/39Ar k-feldspar 150-350 Lovera et al., (1989)

40Ar/39Ar biotite 350-400 Grove and Harrison (1996); Harrison (1985)

40Ar/39Ar muscovite 300-350 Robbins (1972); Hames and Bowring (1994) Fission- b) zero (b) Tagami et al., (1998); zircon (b) 330-350 Tracks damage Rahn et al., (2004) (c) Brandon and Vance c) natural (c) 230±20 (1992); Brandon et al., (1998) Fission- Titanite 265-310 Coyle and Wagner (1998) Tracks Fission- Apatite 110 10 Gleadow and Duddy (1981) Tracks ± (U-Th)/He Zircon 200-230 Reiners et al., (2002) (U-Th)/He Titanite 150-200 Reiners and Farley (1999) (U-Th)/He Apatite 75±5 Wolf et al., (1998)

28 Chapter 2 As for K-Ar and Ar/Ar techniques, fi ssion track dating systems are also dependent on temperature variation (see section 2.7). The radiogenic lattice damage (track) is unstable at high temperatures. Solid state diffusion of electrons can shorten or even- tually erase the tracks, and was fi rst observed with the “closure” of fi ssion tracks systems and defi ned as the partial annealing zone (PAZ) (Wagner, 1989). The PAZ is defi ned as the temperature range where fi ssion tracks are only partially retained over 2 geologic timescales (Bernet et al., 2009 and reference therein), and has been extended to other thermochronological systems as partial retention zone (PRZ). The PRZ is a range of temperature in which some of the decay product of radioisotopic systems are retained while others are lost. Tectonic exhumation (e.g. normal faulting or erosion) can exhume fossil PRZ (Stockli et al., 2000). The relationship among fossil PRZ and the present-day eleva- tion profi le can give key-temporal information about the time of tectonic exhumation (uplift) and denudation (erosion) events as summarized in Figure 2.2. This concept has been successfully applied to low-temperature thermochronometers such apatite fi ssion-track because the partial annealing effect can be tested using confi ned track length distributions (Fitzgerald and Gleadow, 1990). However, age-elevation rela- tionship extrapolated from an ideal one-dimensional profi le should be used as esti- mates of the rates occurring in a naturally evolving geological setting. In fact, active orogeny, generally present more complex thermo-kinematic evolution where more complex processes, as for example lateral advection of material, can lead to biased interpretations of the dates (Batt and Brandon, 2002). To summarize, isotopic systems with different closure temperatures can be used as tools to infer process rates of a geological event. Systems with higher closure tem- perature (e.g., U-Pb, 40Ar/39Ar) retain information about the exhumation and erosion and thermal structure across a region, whereas lower closure system (e.g., (U-Th)/He, fi ssion-tracks) are more sensitive to locally induced variations as for instance the ef- fect of the regression of a glacier in a valley (Batt and Brandon, 2002).

29 Principles of Thermochronology Figure 2.2. Cooling age profi le relationship and types of uplift and denudation for a crustal section of depth z. a) Cooling age-depth (z) profi le with no tectonic activity during time t. The cooling ages decrease rapidly in the partial retention zone (PRZ). The upper limit of the PRZ depends on the geothermal gradient. b) At time t1, the crustal section starts to be exhumed (uplifted) and partially eroded. The grey hexagon marks the ideal thermochronometers that recorded the onset of the denudation t1 and that is preserved in the exhumed PRZ. Its “present- day” elevation with respect to its original depth give an estimate of the amount of rock uplift. To constrain the amount of tectonic (surface and rock uplift) the initial surface elevation at t1 need to be reconstructed, alternatively only the amount of the denudation rates can be con- strained from the base of the exhumed PRZ. Figure modifi ed after Braun et al. (2006).

2.4 In-situ versus detrital thermochronology In-situ thermochronology techniques can infer the thermal and time evolution of rocks as well as their averaged denudation rates. In fast exhuming orogen (i.e the Alps and Himalaya) where denudation rates are high, the topographic relief can be used as a proxy to approximate crustal depth using cooling ages measured from sam- ples collected along a vertical elevation profi le (R) (Fig. 2.3a) (Reiners and Brandon, 2006). The elevation profi le method yields an age-elevation relationship that can be used to infer estimates of the time and of the vertical velocity of the rocks, relative to their closure temperature. A detailed review of the method can be found in Rein- ers and Brandon, (2006). The method is based on three critical assumptions: 1) the

30 Chapter 2 closure isotherm was fl at over the period of closure of the system; 2) erosion rates are uniform over space. Interpretation of exhumation rates and their variability over time, constrained by this approach, can give an approximation of the long-term evolution of an orogen. However, the steady-state assumption applied on this method does not ac- count for the transient response of the landscape to changes in both erosion and exhu- mation. Another limitation of the method arises from the fact that isotherm is unlikely 2 to be fl at in active geodynamic settings (Braun, 2002, 2016; Van der Beek et al., 2002) Generally, where erosional processes are active surface samples do not retain in- formation about previous cooling events that have been removed by the erosion of the rocks that contained that record. This is the case, for example, of rapidly eroding orogenic settings, such as the Himalayan syntaxes or St. Elias mountain of Alaska, where in-situ thermochronology may infer only the most recent exhumation phases and thermal evolution of the last few millions (for middle-to-high systems) or even few thousands of years (for low-temperature system) (Zeitler et al., 2001; Enkelmann et al., 2009). The sediments eroded from such mountain belts can be used to infer denudation histories and to restore the evolution of an evolving hinterland using de- trital thermochronology. Detrital thermochronology using different thermochronometric systems has been applied to obtain information about sediment provenance, exhumation and thermal history of sediment source area and landscape evolution (Cerveny et al., 1988; Naj- man et al., 1997; 2001; Brewer et al., 2003; Von Eynatten and Wijbrans, 2003; Bernet et al., 2004; Carrapa et al., 2004) the sediments in the study area were derived from two different sources, one characterised by white micas with Si <6 .5 pfu and Permian 40Ar/39Ar ages (270 Ma. An important application of detrital thermochronology is the assessment of the difference between the depositional age of a sedimentary sequence and the cooling age of the detrital mineral. This difference is called lag time (Fig. 2.3a). In particularly active orogenic settings the lag-time can be very short (<2 Ma) (Copeland and Harrison, 1990; Bernet et al., 2004). If the erosion was the primary mechanism of cooling the lag time can be used to infer an estimate of the erosion rates in the source region of the sediments.

31 Principles of Thermochronology a

t range b 4 erosion 3 2 3 4 1 te Surface T t R 2 d 1 2 4 3 1

TE4 Ar-Ar TE3 white mica TE2 closure isotherm TE1 detrital sample R = E trange

Figure 2.3. Simplifi ed illustration showing how a range of cooling ages for a set of four sam- ples along a transect profi le can be used to estimate source-region erosion rate. a) Two dimen- sion profi le showing the positions of a set of in situ samples at an elevation varying from sam- ple 1 (at the valley bottom) to 4 at the highest elevation. If exhumation of the rocks is along a vertical path (shown by dashed lines), sample 4 should yield the oldest cooling ages retained from the applied thermochronometric system, assuming a TE4 the time to reach the surface, whereas sample 1 should be the youngest having TE1 as time to reach the surface since its clo- sure temperature. R is the total relief and range is the range of age of the distributions. Samples 1-4 should display a linear relationship between their cooling age and their elevation along the profi le. The source area is denuded by erosion E. If erosion rate was constant over the erosion interval E is equal to R/trange. Td is the depositional age of the cooling ages and is supposed to be the same of the time of exposure at surface Te in a steady-state model. This implies that the time of transport is negligible over geological time. b) after that the samples have been eroded and transported in the river detritus, the range of the cooling ages can be used to estimate E if we assume that the modern relief did not change during the cooling interval. Figure modifi ed after Hodges, 2005 and Braun et al., 2006.

Detrital cooling ages have been used to infer time-averaged erosion rates in drain- age with known relief (Brewer et al., 2003; Hodges et al., 2005; Enkelmann and Ehlers, 2015). An important advantage is that rivers provide an averaged means over a large area, especially in a region where access to bedrock exposure is diffi cult. Riv- ers sediments, in a well-mixed fl uvial system, provide a nearly uniform contribu- tion of the drainage area surfi cial rocks. However, it is important to achieve a robust statistical distribution of ages representative of the sampled drainage. From a purely statistical point of view, Vermeesch et al. (2004) indicated that in a multi-component age distribution of ages, in order to get a 95% confi dent of not missing any peak in the population at least ~100 ages should have been dated.

32 Chapter 2 2.5 K/Ar dating method Potassium has three natural radioactive isotopes that occur in nature: 39K, 40K, and 41K (Nier, 1935). 39K and 41K are stable, whereas 40K is radioactive and decays with a half-life of 1.25 Ga. The branched decay scheme of 40K is shown in fi g. 2.4. In most cases, 40K decays to 40Ca by electron β- emission (~89.5 %). Calcium is an abundant element in earth’s crust (5 %) and 40Ca is the main isotope. Therefore the accumula- 2 tion of radiogenic 40Ca from the decay of 40K can only be measured in rare Ca-poor and K-rich environments.

40K e.c. 0.05 MeV

1.5 10.32%

β+ 0.49 MeV -0.0001% β- 1.0 e.c. 1.33 MeV Ȗ Ȗ 1.51 MeV 89.52% 0.16%

1.48MeV 1.02MeV 0.5 energy release (MeV) release energy

40Ca ground state 40Ar 0 ground state

18 19 20 atomic number (z)

Figure 2.4. Branching diagram displaying the decay scheme for 40K, the energy released by the decay to 40Ar and 40Ca (modifi ed after McDougall and Harrison, 1999).

After the discovery of the occurrence of one isotope of 39K and 41K by Nier (1939), in the late 1940s, the decay of the radioactive parent 40K to the stable daughter 40Ar was successfully used as geochronometer for dating rocks (Aldrich and Nier, 1948). The innovation of the measurement, the high abundance of potassium in the conti- nental crust, and its long half-life of 1250 million years made it particularly suitable to measure geological events in a wide range of applications (Kelley, 2002). Isotopic dating was an essential contribution to the development of fi rst-order steps forward in our understanding of the geological time scale, including the geomagnetic polar- ity timescale and the numerical calibration of the Phanerozoic geological time scale (Harland et al., 1964, Gradstein et al., 2012).

33 Principles of Thermochronology Argon is a noble gas and is produced by the remaining 10.5% of the decay of 40K that yields 40Ar by electron capture. The process of electron capture produces 40Ar in an excited state and with the emission of a γ-quantum,40Ar may reach the ground state. The smallest proportion (-0.0001%) of the 40K decay to 40Ar produces the emission of a positron (β+) although this decay path is uncertain. The rate of isotopic decay follows an exponential law. Therefore, by measuring the absolute amount of radiogenic 40Ar and the absolute amount of 40K in a sample, the age can be determined by the equation: כ ݎܣߣ ସ଴ 1 (ݐ = ݈݊ ቆ 1 + ᇱ ସ଴ ቇ (2.1 ߣ ߣ௘ + ߣ௘ ܭ where t is time since the closure, λ is the decay constant of 40K and are the par- tial decay constants of 40Ar, 40Ar* is the amount of the radiogenic daughter, which is divided by the amount of parent 40K present in the analyzed sample (Kelley, 2002). The amount of both K and Ar can be measured in the sample with an isotope dilution technique using a mass spectrometer. For the K/Ar method, the parent and daughter isotopes have to be measured with two different approaches on two separate aliquots of the sample: because K is a metal-cation and Ar is a noble gas they have a different mobility within the mineral’s structure. The potassium cations form part of the crystal lattice, whereas, argon is trapped as a neutral gas molecule. Both can migrate follow- ing diffusion transport laws inside the crystal lattice (McDougall and Harrison, 1999). Argon can be released while heating the sample in a vacuum with a laser or a furnace and can be measured after purifi cation from other ambient gasses. The con- centration of potassium can be measured with different techniques. Some of them are wet-chemical atomic absorption spectrometry (AAS) or fl ame photometry methods, or by X-ray fl uorescence (XRF) on glass beads. The measurement of the isotopic abundance is calibrated using a known amount of 38Ar to the Ar from the sample and measuring the ratio of 40Ar to 38Ar (McDougall and Harrison, 1999). After few analyses (3-5 depending on laboratory routines) a set of blank analyses is performed. The blank correction analysis allows to detect the presence of trace amounts of hy- drocarbons and on m/e: 36 and 38 hydrochloric acid molecules on the surface of the extraction line of the mass spectrometer and to correct the amount of 40Ar measured. Contamination with atmospheric argon is corrected by measuring non-radiogenic sta- ble isotope of 36Ar relative to the 40Ar released. The ratio of this isotope is constant in the atmosphere (40Ar/36Ar = 298.56 ± 0.31; (Lee et al., 2006)). The isotopic age measured (date) by the K/Ar (and also 40Ar/39Ar) technique relies on fundamental assumption that needs to be validated in order to obtain a realistic age

34 Chapter 2 ( McDougall and Harrison, 1999; Faure, 1989; Kelley, 2002); The decay of 40Ar is independent of its physical state and it is not dependent on variations in temperature and pressure. If there are any changes in electron capture, partial constant decay may occur for 40K, these effects are negligible for the size of the Earth. The ratio between 40K and 39K is constant in nature. This assumption is well established using many experiments on terrestrial and extraterrestrial samples (i.e. 2 Turner, 1968). All the measured radiogenic argon in a sample is produced by 40K decay in the in- terval since the rock crystallized or recrystallized. There are examples in nature where this assumption is violated: i.e. glassy deep-sea basalts that may not have outgassed preexisting radiogenic argon. This argon amount is defi ned as extraneous argon. Terrestrial rocks can also contain non-radiogenic 40Ar and corrections can be made. The measured samples, either mineral or whole rock, must have behaved as a closed system since the event being dated. This implies no loss or gains for both Ar and K. This assumption cannot always be fulfi lled especially in rocks that expe- rienced particularly complex geological and thermal histories (i.e. reheating due to metamorphic events).

2.6 40Ar/39Ar method The K/Ar method was used extensively to provide geological age constraints in many geological settings over the second half of the last century. However, the dis- advantage of the method, as mentioned before, rises from the necessity of measuring the amount of potassium (a solid) and argon (a gas) in two different ways from dif- ferent splits of the same sample and thus may introduce an additional uncertainty to the measured ages. Using the 40Ar/39Ar method by previous irradiation of the sample a small proportion of 39K atoms are transformed into 39Ar (McDougall and Harrison, 1999). Some K/Ar experiments were conducted by Wänke and Konig, (1959) using samples that were previously irradiated, they attempted to measure the K content with a standard neutron activation technique. The 40Ar/39Ar technique was fi rst applied by Merrihue and Turner (1966). In their work, they tested the consistency of the 40Ar/39Ar method by comparing the measure- ment of some meteorites and lunar samples and compared these data with the K/Ar dating exist. The 40Ar/39Ar method is based on the production of 39Ar by the irradia- tion of K-bearing minerals by neutron bombardment in a nuclear reactor. The desired reaction is:

35 Principles of Thermochronology ͗͝K n ͗͝ + ݌ ͕͝ + ՜ ͕͜Ar (2.2) During irradiation of a K-bearing sample in a nuclear reactor, several interfering reactions that include potassium, calcium, and chlorine isotopes occur that also form isotopes of Ar. The amount of 39Ar produced during irradiation from 39K can be deter- mined following Mitchel, (1968):

ଷଽܣݎ = ଷଽܭ οන߮(ߝ)ߪ(ߝ)݀ߝ (2.3)

where ∆ is the irradiation time, φ(ε) is the fl ux density of neutrons with energy (ε), 39K is the number of atoms and σ(ε) is the neutron capture cross section of 39K neutrons of energy (ε). The K-Ar age equation can be rearranged in order to calculate the amount of radiogenic 40Ar* as follow: ߣ + ߣᇱ (௘ ௘ ((݁ఒ௧) െ 1) (2.4 ܭସ଴ = כ ݎܣସ଴ ߣ

When 2.3 and 2.4 are combined this yields:

ସ଴ ସ଴ ᇱ ఒ௧ (ߣ௘ + ߣ௘ 1 ((݁ ) െ 1 ܭ כ ݎܣ = (2.5) ߣ οܶ ׬ ߮(ߝ)ߪ(ߝ)݀ߝ ܭଷଽ ݎܣଷଽ the dimensionless irradiation parameter J can be derived as follows: The obtained J value is substituted in Eq. 2.5: ଷଽܭ ߣ ܬ = ସ଴ ᇱ οܶ න ߮(ߝ)ߪ(ߝ)݀ߝ (2.6) ܭ ߣ௘ + ߣ௘ the J value can be determined by irradiating a sample of known age, the fl ux moni- tor, together with the samples of unknown age and be computed from Eq. (2.7) after the 40Ar/39Ar ratio is determined by mass spectrometric analysis: ఒ௧) െ 1݁) כ ݎܣସ଴ = (2.7) ଷଽܣݎ ܬ where the 40Ar*/39Ar ratio is measured in the fl ux monitor and tm is the indepen- dently derived age of the sample. The intensity of the neutron fl ux during irradiation in the nuclear reactor depends on the position of the samples in the holder (stacking order). A known number of fl ux monitor standards is therefore positioned with the samples of unknown ages and sent for the irradiation in the nuclear reactor (see next section for technical details).

36 Chapter 2 The irradiation should be enough long (hours) to allow suffi cient 39Ar to be produced. The length of the irradiation depends on the average expected the age of the samples i.e samples >50 Ma will need a longer irradiation than samples <10 Ma, or <0.5 Ma. After that the 40Ar*/39Ar ratio is measured in the fl ux monitors, the J-values are com- puted using Equation (2.8) and plotted as a function of the position in the samples holder. The respective J-value for the unknown samples can recursively be obtained 2 by interpolating the fl ux monitor value and their position in the sequence. The measured 40Ar/39Ar* ratios are used to calculate mineral sample ages t using the age equation:

כ ݎܣସ଴ 1 (቉ (2.9 ܬݐ = ln ቈ1+ ߣ ଷଽܣݎ With the 40Ar/39Ar method, one cannot avoid the production of argon by interfer- ing isotopes and thus correction for these amounts of argon is needed. Argon isotopes in any sample indeed are not only produced by radioactive decay or admixture of atmospheric argon but also from interfering reactions with the isotopes of calcium, potassium, and chlorine (Faure, 1989). One of the most important interference is interferences is produced by neutron reactions on isotopes of calcium. Dalrymple and Lanphere, (1971), starting from the equation that relates the measured 40Ar*/39Ar and corrects it for all the interfering reactions by Brereton (1970), developed an expres- sion for their F (= 40Ar*/39Ar): ܣെܥ ܤ + ܥ ܥ ܦെ ܥ ܨ = ଵ ଵ ଶ ଷ (2.10) 1 െ ܥସܦ Where; A = measured 40Ar/39Ar, B = measured 36Ar/39Ar, C1 = atmospheric 40Ar/39Ar (= ~295.5), C2 = 40Ar/39Ar produced by interfering reaction with calcium, C3 = 40Ar/39Ar produced by interfering isotopes of potassium, C4 = 39Ar/37Ar by in- terfering of calcium, D = 37Ar /39Ar the value in the sample after correcting for decay of 37Ar (Faure, 1989).

2.7 Zircon Fission-Tracks dating

Zircon (ZrSiO4) (Fig. 2.5) is a common accessory in most metamorphic, igneous and sedimentary rocks and it is resistant to weathering, transport, and abrasion. Zircon can be dated with various isotopic methods because of its high concentration of ura- nium and thorium. Some of the most applied dating techniques are U/Pb, (U-Th)/He and fi ssion track (FT) analysis. The latter will be the focus of this paragraph.

37 Principles of Thermochronology 10 mm 2 mm

Figure 2.5 Optical picture of detrital zircons from the Brahmaputra rivers modern detritus.

The FT method is based on the damage produced by fi ssion decay of 238U. A small portion of the 238U decays by splitting or fi ssion of the atom. Inside a solid medium, the atomic reaction produces two positively charged high-energy nuclei that are prop- agated away from each other, producing a single linear array of ionization damage in the lattice that is known as a fi ssion track (Fleischer et al., 1975). The resultant path in the crystalline lattice is thought to be a high defect density are produced by the change in charge of the atoms along the ejected fi ssion fragment paths. Fission track dating techniques use the produced lattice damage (as daughter product) that can be revealed by chemical etching and counted under an optical microscope (Gleadow et al., 1981). In other words, the ionization damage produces widespread dislocation of atoms from their crystalline position along the fi ssion track (Fleischer et al., 1975). Fission tracks in their “natural” state (generally with a width of 3-12 nm) are de- tectable only using electron transmission microscopy. To reveal the tracks using opti- cal microscopy, samples need to be treated by polishing and chemically etching the crystal surface. The etching preparation of the sample will determine the length and width of fi ssion tracks and depend on the mineral and on the chemical composition of the etchant (Gleadow and Brooks, 1979) 18 zircons and 25 apatites separated largely from Lower Tertiary magmatic rocks of East Greenland, with a few examples from Caledonian rocks. The sphene and zircon ages of Caledonian rocks agree with other radiometric ages but apatite is strongly discordant indicating that these rocks cooled very slowly over a 200 m.y. period. It was not until the Permian/Lower Jurassic that they fi nally cooled below 100 ~ C, possibly as a consequence of uplift and erosion at this time in connection with extensive rifting. No evidence of a Tertiary imprint has been found in these rocks. Layered gabbros, such as Skaergaard, were em-placed at

38 Chapter 2 about the same time (ca. 54 ma). Because etchant properties and timing slightly vary from laboratories it is important to add an adequate description of the etching proce- dure used for the samples while comparing dates from different datasets (Gallagher et al., 1998).The decay scheme of fi ssion track follows the same principles of any other isotopic technique: the relative abundance of the daughter product released from the parent atom can be measured. The concentration of fi ssion tracks is determined by 2 counting the spontaneous tracks on a given surface of a mineral crystal under an opti- cal microscope (Gallagher et al., 1998). To compute the 238U content before counting, the sample needs to be irradiated with low-energy thermal neutrons to induce fi ssion of 235U. A high energy neutron fl ux has the disadvantage of producing undesired fi s- sion from 235Th and 238U (for example) that needs to be avoided (Gallagher et al., 1998). Assuming that we monitor the neutron fl ux in the irradiation process, using the number of “induced” tracks we can defi ne the amount of 235U and from the fi xed 235U/238U ratio, we are able to quantify the amount of 238U (Gallagher et al., 1998; Reiners and Brandon, 2006). The external detector method (Fig. 2.6) is one of the most applied zircon fi ssion- tracks technique and was applied also for the ZFT dates of chapter 5. Using this meth- od, the polished and etched samples where spontaneous fi ssion tracks are revealed are mounted together with an external mica with low U-content and irradiated. After irradiation, the external detector is etched to reveal the spontaneous fi ssion-tracks in the sample and in the mount (Bernet and Garver, 2005). The advantage of the exter- nal detector method is that single grains can be dated. Using the external detector ζ method, the equation to determinate the age t from fi ssion tracks is:

ଵ దೞ t = ln (ߣௗ ߷ௗߞ݃ +1) (2.10) ఒವ ద೔

ρs are the spontaneous tracks and ρi the induced tracks per unit area; λd is the total decay constant; ρd is the density of tracks in a glass with known amount of ura- nium (dosimeter) used as monitor for the neutron fl ux in the reactor: g geometry factor that for the external detector method represents the factor from twice the effective volume than the induced tracks represent (Gallagher et al., 1998); ζ is a calibration factor that is determined from a sample of known age (Hurford, 1990b).

39 Principles of Thermochronology Spontaneous fission tracks accumulation

Polished section through crystal

Spountaneous tracks surface confined after etching processes

External mica mounted on the polished zircon

Irradiation by thermal neutrons: induced fission tracks recorded in the detector

Induced fission tracks etched only in the detector

B B’ CC’C’ Mirror image

A A’ Spontaneus tracks in the External detector mount etched zircons mount with induced tracks Figure 2.6. The external detector method. Modifi ed after Hurford and Carter (1991), Gal- lagher et al., (1998).

2.8 Sample preparation and irradiation

Detrital 40Ar/39Ar and fi ssion track studies have been applied to a wide range of clastic sedimentary records, but most studies focused on clastic sedimentary sections (i.e. sandstones) and only recently to modern river detritus. Sampling strategies are different for the sampling environments (rocks or sediments) and for the nature of the intended study. Field collection strategy should be carefully planned while using river sediments as a proxy for provenance and long-term and present-day exhumation. Furthermore, a suffi cient number of grains should result from the mineral separation and the amount of sampled sands must be suffi cient to reproduce a robust number of analysis per samples (i.e 40-120 single grains analysis).

40 Chapter 2 All samples analyzed for this project were collected from modern river sand- bars of the fl uvial network in the eastern Alps and eastern Himalaya (chapter 3-4). The sampling sites were located at least 1 km away from tributary junctions to the river and from landslides to avoid bias toward one particular source in the main river streams. Sampling was done in the eastern Himalaya after the monsoon season in or- der to get relatively “fresh” eroded material and, in the summer season, in the eastern 2 Alps. Approximately 2 kg of medium-grained sand was collected from the top 10 cm of sediment at each sampling location from the edge of the active channel. Sample preparation was done at the mineral separation laboratory of the Vrije Universiteit, Amsterdam. Standard mineral separation procedures were used to obtain at least 200-250 target minerals per sample or size fraction. The samples were fi rst washed to remove the fi ne fraction (<30 μm) using an in-house built “desliming ap- paratus”. This device is based on Stoke’s Law: the sample was mixed in a 70 cm high water column, settles during 10 min and the suspended material in the water column was pumped away from the top of the cylinder. This process was repeated up to 10 times until samples were clean. The sample was then washed with purifi ed water and dried in an oven overnight at 60ºC. We sieved the obtained sands in different size fractions. For the Himalayan river samples, aliquots of 125-180 μm, 180-250 μm, 250-400 μm, 400-500 μm, 500-1000 μm were separated i. For the Alps rivers, we prepared two-grain fractions of 500-1000 μm and 250-500 μm. The samples from the Himalaya were used as well for testing the sensitivity of the new installed Helix+ on smaller mica grain fractions. The selected fractions were treated to obtain single grain minerals of muscovite, biotite, and zircon. Firstly, the fractions were separated in 4 different aliquots se- lected by their shape properties using a Faul-table. In the next step, the obtained sub- fractions were treated using heavy liquids density separation techniques. Different procedures were adopted to obtain muscovite (~ρ = 2.82 g/cm3), biotite (~ρ = 3.09 g/cm3) and zircon (~ρ = 4.65 g/cm3). If the mineral fraction was not “clean” enough from undesired mineral species, a Frantz magnetic separator was used to remove the magnetic minerals non-magnetic micas and zircon. Thereafter, grains with undesired physical alteration were removed (negative picking) or the best ones were collected (positive picking) by hand picking under a binocular optical microscope. For 40Ar/39Ar analysis, the samples were wrapped in 9 mm ID quartz tubes to- gether with the monitor standard Drachenfels sanidine dated at 25.43 ± 0.03 Ma. This value is compatible with the set of (Kuiper et al., 2008b). Samples and standards were irradiated in the cadmium-lined CLICIT facility of the TRIGA reactor of the Oregon State University Radiation Center, USA. The samples were irradiated for 12h

41 Principles of Thermochronology in irradiation VU100, VU101, and VU109 and 7 hours in VU108. Nuclear interfer- ence corrections factors were 0.1869 for 38Ar/36Ar, 0.00673 for 39Ar/37Ar, 0.000264 for 36Ar/37Ar and 0.00086 for 40Ar/39Ar.

2.9 40Ar/39Ar measurement and data analysis For 40Ar/39Ar analysis, single grains of standards and samples were loaded on a 185 hole copper disk, placed in a prebake vacuum unit and baked overnight at 250ºC to remove atmospheric argon. Next, the sample tray was placed in ultra-high vacuum purifi cation line of either the AGES or Helix system and baked overnight at 120ºC for detailed set up of the Ages we refer to Wijbrans et al., (1995) and for the Helix MC+ to the chapter 5 of this thesis. Five isotopes of argon from 36Ar to 40Ar (fi gure 2.7) were measured from each single mineral samples using 40Ar/39Ar total fusion of single-grain crystals of white mica and biotite.

The samples were fused using a 25W Synrad CO2 laser instrument. The released

gas was fi rst purifi ed by a cold trap (-70°C) to catch volatiles (such as CO, CO2, SO2,

H2O, and H2). The gas was further cleaned in an ultra-high vacuum gas purifi cation line by exposure to SAES NP10 (Fe-V-Zr alloy) getters. After purifi cation, generally, 5 min for muscovite and biotite samples, the gas is expanded into the mass spectrom- eter.

Figure 2.7. Measured argon isotopes with their sources in standard terrestrial conditions. The measured argon isotopes using 40Ar/39Ar experiments are fi ve. Trapped argon is assumed to be of atmospheric condition (298.56, Lee et al., 2006). Figure adapted from Deino et al., (1998).

42 Chapter 2 At the VU Argon laboratory, in the AGES fi ve isotopes of argon (36-40) are meas- ured consecutively in a peak jumping mode in 60 cycles. On the multi-collector Helix MC the fi ve isotopes of argon are measured simultaneously on 5 different collectors: 40Ar on H2-Faraday; 39Ar on H1-Faraday or H1-CDD; 38Ar on AX-CDD; 37Ar on L1- CDD and 36Ar on L2-CDD (CDD = compact discrete dynodes) for 15 cycles with 33 seconds integration time. 2 After data collection on the mass spectrometers, the raw data need to be processed and corrected for blanks, mass discrimination, nuclear interferences and, in the case of the Helix, for different gains of the collectors. We convert raw mass spectrometer data output using an in-house designed Excel macro script to make it compatible with the ArArCALC data reduction software (Koppers et al., 2002). Further information about ArArCALC software by Koppers et al. (2002) can be found at https://earthref. org/ArArCALC/. Data were plotted in age-probability-density plots using the Probability Density Function (PDFs) and Kernel Density Estimator (KDE) of Radial Plotter program (Vermeesch, 2012) as shown in fi gure 2.8. PDF and KDE are statistical techniques that aim to estimate the relative trend of age populations by summing a set of the Gaussian distribution. This tool is particularly useful for analyzing a detrital distri- bution of ages. The Probability Density plots (PDPs) produced by using the PDFs estimator is, however, highly dependent from age uncertainties and can bias the total distribution by privileging more accurate ages compared to ages yielding a bigger er- ror. This can produce particularly bias in the visualization of age distributions from datasets with higher analytical errors (i.e. Zircon Fission Tracks). PDPs can produce counter-intuitive results when a) the number of analysis is low, b) the analytical error is high (Vermeesch, 2012). Because of the aforementioned problems, Radial Plotter software propose the Kernel Density Estimator (KDE) option while dealing with the dataset with high uncertainties.

43 Principles of Thermochronology (n=57) Relative probability Relative

0 40 80 120 160 200 240 280 320 360 Age (Ma) Figure 2.8. Probability Density plot (grey area), Kernel Density estimator (black line) of 102 single grains analysis from the Brahmaputra modern river sand sample. The plot was obtained using the DensityPlotter, a Java application for Kernel Density Estimators (Vermeesch., 2012). The Y-axis represents the relative density while the X-axis express the ages in Ma. Due to the high analytical precision of 40Ar/39Ar ages, the two statistical methods produce two similar curves.

2.10 ZFT samples preparation and data presentation Aliquots of grain fraction <250 μm were taken from the “R” and “P/R” fraction (where minerals with rounded shape are accumulated) of the Faul table. Samples were then processed using a heavy liquid technique. To separate zircons (average density 4.65) we used Methylene iodide (Diiodomethane) that has a relatively high specifi c gravity of ~3.3 g/ml. Diiodomethane is highly soluble in acetone that is used to clean the sample. Because of the high costs (120 eu / 100 g), the heavy liquids are then separated from acetone and recollected for subsequent usage or recycling. The sink >3.3 g/ml fractions collected is dried in an oven overnight at 60ºC and the zircons can additionally be separated from magnetic or paramagnetic minerals us- ing the Frantz magnetic separator. Because different minerals have different magnetic susceptibilities, the sample needs to be passed through the machine different times at increasing amperage and at a different slope of the magnet. These procedures were done at the mineral separation laboratory of the Vrije University Amsterdam and al- lowed to obtain homogeneous fractions of zircon per samples. Zircon aliquots (200-1000 grains) were mounted in PFA Tefl on, polished and

44 Chapter 2 etched in a NaOH-KOH melt at 225-230 °C between 50-70 hours. Two mounts per sample were prepared and etching was done in a stepwise manner to establish a rea- sonable number of fi ssion tracks for the majority of the grains, focusing on younger grains in one mount and older ones in the other (Bernet and Garver, 2005). The sam- ples were covered with muscovite sheets as external detectors and sent for irradiation to the FRM II Research Reactor at the Technische Universität München, Germany, to- 2 gether with FCT (Fish Canyon Tuff) and BLK (Bulk) standards. The dosimeter glass employed was an IRMM541. Standards and samples were counted at 1250 magnifi ca- tion on an Olympus BH2 microscope, using the FT Stage 4.04 program. We aimed at obtaining at least 100 single-grain ages per sample where possible. After counting of the tracks, the data handling involves the use of different open source programs such as Zetaage, Zetafactor, Binomfi t etc. by M.T. Brandon that are available at ftp://love.geology.yale.edu/pub/brandon /FT_PROGRAMS/FT_Peaks/. A detailed explanation on how effectively fi ll the counting data with the zeta fac- tor method with program Zetaage is described in Bernet et al., (2004), where use- ful information is listed. Detrital fi ssion-track crystal grain age distributions can be decomposed into several age components using binomial peak fi tting (Galbraith and Green, 1990). Results can be evaluated after correctly fi lling the fi les using Zetaage program and a detailed method procedure is explained in Bernet et al., (2004). Using Zetaage output it is possible to proceed to the peak-fi tting procedure using BinomFit. BinomFit (Brandon, 1996; 2002), follow the binomial approach of the zeta method of Hurford and Green (1983). BinomFit it is a software for estimating components in a mixed distribution of fi ssion-tracks ages up to 1000 grain analysis. The program, using the Z method, calculates the grain-age in Ma with their 95% relative error. The ages are plotted as summarized in Fig. 2.9 and consists in unsorted Grain Ages Plot (Fig. 2.9a), Probability Density Plots (PDP) with best-fi t peaks option (Fig. 2.9b), and Radial plots (Fig. 2.9c). The number of peaks can be chosen arbitrarily but it is good to proceed by incremental steps. F test is applied in BinomFit to fi nd signifi cant peaks (Fig. 2.9d). The F test it is used at each step to increase the number of peaks of a representative population of ages and to fi nd the best-fi t solution.

45 Principles of Thermochronology a) b)

c) d)

Figure 2.9. Graphics output from BinomFit program by Brandon, 2002. Data are from Brah- maputra sample B4 from chapter 5. a) unsorted grain-age distributions. b) Probability density plot (continuous line) with best-fi t peaks represented on the single curves line. The histograms in the background are scaled such as the density is the same of the PDP. c) data showed as radial plot. d) F test used for signifi cant peaks distribution.

46 Chapter 2

Chapter 3 Extracting information on the spatial variability in ero- sion rate stored in river de- trital age distributions.

L. Gemignani J. Braun P. van der Beek

Note. The idea behind this chapter have been discussed be- tween Lorenzo Gemignani, Jean Braun, and Peter van der Beek. The code and math have been developed by Jean Braun. The method has been tested by Lorenzo Gemignani.

Citation: Extracting information on the spatial variability in erosion rate stored in detrital cooling age distributions in river sands. (2017). Earth Surf. Dynam. Discuss., https://doi. org/10.5194/esurf-2017-42.

Abstract The purpose of detrital thermochronology is to provide constraints on regional scale exhumation rate and its spatial variability in actively eroding mountain ranges. Procedures that use cooling age distributions coupled with hypsometry and thermal models have been developed in order to extract quantitative estimates of erosion rate and its spatial distribution, assuming steady state between tectonic uplift and erosion. This hypothesis precludes the use of these procedures to assess the likely transient response of mountain belts to changes in tectonic or climatic forcing. In this paper, we describe a simple method that, using the observed detrital mineral age distribu- 3 tions collected in a system of river catchments, allows to extract information about the relative distribution of erosion rates in an eroding hinterland without relying on a steady-state assumption or the value of thermal parameters. The model is based on a relatively low number of parameters describing lithological variability among the various catchments and their sizes, and only uses the raw binned ages. In order to illustrate the method, we invert age distributions collected in the Eastern Himalaya, one of the most tectonically active places on Earth. From the inversion of the cooling age distributions, we predict present-day erosion rates of the catchments along the Siang-Tsangpo-Brahmaputra river system, as well as smaller tributaries. We show that detrital age distributions contain dual information about present-day erosion rate, i.e. from the predicted distribution of surface ages within each catchment and from the relative contribution of any given catchment to the river distribution. The inversion additionally allows comparing modern erosion rates to long-term exhumation rates. We provide a simple implementation of the method in R.code within a Jupyter Note- book that includes the data used in this paper for illustration purposes.

3.1 Introduction Thermochronometric methods provide us with estimates of the cooling age of a rock, i.e. the time in the past when the rock cooled through a so-called closure temper- ature (Dodson, 1973), which varies between systems and minerals. One of the main geological processes through which rocks experience cooling is exhumation towards the cold, quasi-isothermal surface (Brown, 1991). Young ages are commonly inter- preted to indicate rapid exhumation and old ages should correspond to slow exhuma- tion. Cooling ages can also record more discrete cooling events such as the nearby emplacement of hot intrusions (Gleadow and Brooks, 1979) or the rapid relaxation of isotherms at the end of an episode of rapid erosion (Braun, 2016). Datasets are now routinely assembled by collecting and dating a large number of mineral grains from

49 Extracting erosion rate from rivers detrital age distributions a sand sample collected at a given location in a river draining an actively eroding area. Such detrital thermochronology datasets provide a proxy for the distribution of surface rock ages in a given catchment (Bernet et al., 2004; Brandon, 1992). By repeating this operation at different sites along a river stream, one obtains redundant information that can be used to document more precisely the spatial variability of in-situ thermochronological ages in a river catchment (Bernet et al., 2004; Brewer et al., 2006). Methods have been devised to extract quantitative information from such detrital datasets concerning the erosion history of a tectonically active area, as well as esti- mates of its spatial variability. Ruhl and Hodges (2005) convolved their detrital age datasets with the hypsometry of the catchment to test the assumption of topographic steady-state in a rapidly eroding catchment of Nepalese Himalaya. Similarly, Stock et al. (2006) and Vermeesch, (2007) combined detrital apatite (U-Th)/He age datasets with an age-elevation relationship established from in-situ samples to predict the dis- tribution of present-day erosion rates in the eastern Sierra Nevada and White Moun- tains of California, respectively. Whipp et al. (2009) used simulations from a ther- mo-kinematic model to defi ne the limits of applicability of such a technique, while Enkelmann and Ehlers (2015) used it in a glaciated landscape. Wobus et al. (2003, 2006) collected samples from tributaries of the Burhi Gandaki and Trisuli rivers to document the strong transition in erosion rate across a major topographic transition. By limiting their sampling to tributaries, they circumvented the need to develop and use a mixing model for the interpretation of their data. Brewer et al. (2006) derived optimal values for erosion rate in neighboring catchments by comparing and mixing theoretical probability density distributions with detrital age data from the Marsyandi River in Nepal. More recently, McPhillips and Brandon (2010) used detrital cooling ages combined with in-situ age measurements to infer a recent increase in relief in the Sierra Nevada, California. However, these methods have not taken advantage of the fact that detrital age dis- tributions contain two separate pieces of information concerning the spatial patterns of present and past rates of erosion. The fi rst piece of information comes from the ages themselves: catchments or sub-catchments where the proportion of grains with young ages dominates are likely to experience rapid exhumation today or in the recent past; whereas catchments or sub-catchments where the proportion of grains with old ages dominates are more likely to have experienced rapid erosion in a more distant past. However, there does not need to be a one-to-one correlation between young ages and fast present-day erosion rate or old ages and low present-day erosion rate, as a rapidly exhuming catchment may not have experienced suffi cient total erosion to

50 Chapter 3 exhume rocks bearing reset ages. 3.2 The method

3.2.1 Basic assumptions We assume that we have collected a series of age distributions measured at M specifi c points (or sites) along a river that drains a tectonically active region where exhumation rate is likely to vary spatially. We also assume that the distributions have been decomposed into N age bins that may, for example, correspond to given, known geological events or, alternatively, have been selected without prior knowledge, usu- 3 ally uniformly distributed and of equal age width over a given age range, i.e. the range of observed ages (see Figure 3.1). Although each bin corresponds to an age range, it might be easier to refer to it as representative of an event of a given “age” which can be taken as the mean age of the range, for example. We will call for K=1,…, N and i=1,…,M the relative height of bin k in distribution i . Because these are relative contributions, we have:

N k (3.1)  Hi = 1, for all i = 1, L ,M k =1

Example of measured age distribution, N = 4

Bin 2 h = 14 Corresponding Bin 1 Bin 3 Relative bin heights h = 10 h = 14 H1 = 10/41 = 0.238 Bin 4 H2 = 4/41 = 0. 341 h = 7 H3 = 14/41 = 0.268 H4 = 7/41 = 0.171 Number of grains in bin range, h in bin range, Number of grains

Age Bins (Ma)

k Figure 3.1. Example of a measured age distribution and the relative heights Hi of the cor- responding bins ( N = 4 in this example).

The landscape is divided into exclusive contributing areas for each of the points along the main river where we have measured a distribution. We take the convention

51 Extracting erosion rate from rivers detrital age distributions that Area 1 (of surface area A1 ) is the area contributing to site 1, whereas Area 2 (of

surface area A2 ) is the area contributing to site 2 but not to site 1. Area i (of surface

area Ai ) therefore contributes to site i but not to the previous i 1 sites (see Figure

3.2). In each Area i , we will assume that i is the relative surface rock density of the

mineral used to estimate the age distribution. We take the convention that 0 < i < 1,

with i = 1 corresponding to an area i with surface rocks that contain the mineral in

abundance (for example granite for muscovite) and i = 0 corresponding to an area i with surface rocks that do not contain the mineral (for example carbonates for mus- covite). If, for example, the area is made of 60% granite and 40% carbonates, and we have measured ages using a mineral that is abundant in granites (like muscovite) but

absent in carbonates, then  = 0.6 . We also call i the unknown present-day mean exhumation rate in Area i .

River main trunk Direction of flow

Figure 3.2. Schematic representation of how the landscape is divided into exclusive contrib-

uting areas Ai (different shades of grey) for each of the points (here the circles labeled i=1,…,3) along the main river where we have age distributions. εi and αi are the assumed mean exhuma- tion rate and surface rock mineral density of area i, respectively.

The surface areas, Ai, can be easily computed from a DEM. A good approxima-

tion of the concentration factors, αi can be derived from a geological map or from the relative concentration of given minerals representative of a given lithology in each of the samples used to derive the age distributions as explained in further detail in the work of Malusà et al., (2016b) their effects in sediments can be easily detected and modeled mathematically. By contrast, mineral fertility in parent rocks depends on their full geological history. As a consequence, the relationships between bedrock geology and mineral fertility are hardly predictable, and a direct measurement of this latter parameter is thus required. In this review article, we describe the basic prin- ciples of hydraulic sorting, and illustrate a quantitative approach for mineral fertility determination that applies these basic principles to the analysis of modern sediments.

52 Chapter 3 Its application to the European Alps shows that apatite and zircon fertility values may range over three orders of magnitude. Variable mineral fertility in parent rocks thus represents, by far, the largest source of bias in detrital geochronology studies. Our study highlights an evident relationship between bedrock geology and mineral fertil- ity, which confi rms that the mineral concentration in modern sediments, in the lack of hydraulic sorting effects, is a good proxy of the mineral abundance in bedrock. Min- eral fertility maps of the European Alps unravel that metamorphic and plutonic rocks generally have higher apatite and zircon fertility values than sedimentary rocks, but major variations are also observed between different tectonic units within the same 3 paleogeographic domain. The impact of mineral fertility in detrital studies is eventu- ally illustrated with examples from the Alpine region, based on alternative sampling strategies (i.e., the confl uence sampling and the along-trunk sampling approaches. From these simple assumptions, we can then write that the number of grains of age k coming out of catchment i is given by:

k k (3.2) D i = A i  i i C i

k where Ci is the unknown relative concentration of grains of age k in surfi cial rocks in Area i. We also have: N k for all i (3.3)  Ci = 1, k = 1 k because the Ci are also a relative or normalized concentrations. The relative k concentrations Ci tells us if the event corresponding to age k has affected Area i (or, more correctly, if it has been preserved in its superfi cial rocks) whereas i is a meas- ure of present-day exhumation rate in Area i.

3.2.2 Downstream bin summation along the main trunk We can now write that the predicted height of bin k in the distribution observed at site i should be equal to the total number of grains of age bin k coming from all upstream areas divided by the total number of grains of all ages coming from all up- stream areas:

i N i i N i k k k k k Hi = (D j )/(D j ) = (Aj j jC j )/(Aj j jC j ) (3.4) j=1 k=1 j=1 j=1 k=1 j=1 We can slightly re-arrange this to obtain:

53 Extracting erosion rate from rivers detrital age distributions i i N i i k k k k Hi = (Aj j jC j )/(Aj j j C j ) = (Aj j jC j )/(Aj j j ) (3.5) j=1 j=1 k=1 j=1 j=1

If we divide the numerator and denominator of this expression by A1ε1α1, we obtain: i i k k (3.6) Hi =  jC j / j j=1 j=1

where:

Aj j j  j = (3.7) A111 is the contribution from Area j relative to Area 1. Note that, if we assume that we

can confi dently estimate Aj and αj from a topographic map and a geological map, ρj

becomes a measure of the unknown exhumation rate, εj, in Area j relative to the un-

known exhumation rate, εj, in Area 1.

3.2.3 Incremental formulation k We now try to express Equation (3.6) as an incremental relationship between Hi k and Hi1 only, i.e. Between the relative bin heights between distributions measured at two successive points along the main trunk. From Equation (3.6), we can write:

i i i1 i1 k k k k (3.8) Hi =  jC j / j = ( jC j  iCi )/( j  i ) j=1 j=1 j=1 j=1

i1 and by dividing numerator and denominator by  , we obtain:  j=1 j

k k k Hi = (Hi1  iCi )/(1 i ) (3.9)

where: i1 (3.10) i = i/ j j=1 We can fi nally write: k k k k (3.11) Hi  Hi1 = (Ci  Hi )i

From this relationship, we see that the relative changes in bin height between two successive sites along the mainstream tell us something about the present- day erosion rate in the intervening catchment. If the relative bin height doesn’t

54 Chapter 3 k k change between two successive sites ( Hi = Hi1), then we cannot tell if it is be- cause the exhumation rate in catchment i is nil (i = 0  i = 0  i = 0 ), or because the signature of the source in catchment i , i.e. the distribution of ages k k k at the surface, is identical to that of the previous catchment (Ci = Hi = Hi1 ).

k 3.2.4 Inverting for Ci and i k Using Equation (3.11), we can now obtain the unknown Ci recursively using: H k  H k C k = i i1  H k (3.12) i i 3 i

by making fi rst the simplest assumption that  =  for all i , which leads to: i 1

Aii i = (3.13) A11 and, i1 i = Aii/Aj j (3.14) j=1

Assuming a uniform exhumation rate (εi = ε1) should be regarded as the zeroth- order scenario that should fi rst be considered to explain the data; it may, however, lead k k to unrealistic solutions for any of the Ci , i.e. values of Ci that are not in the range [0,1]. To avoid this we must add two conditions that affect the values for the unknown δ . C k > 0 implies that: i i H k  H k i1 i for all k = 1,…, N (3.15) i > k Hi

k and Ci < 1 implies, in turn, that:

H k  H k  > i i1 for all k = 1,…, N (3.16) i 1 H k i The fi rst condition applies where there is a decrease in any relative bin height k between locations i - 1 and location i, i.e. , whereas the second condition applies where there is an increase in any relative bin height k between locations i - 1 and i, i.e. k k . We make the further (and trivial) assumption that the true exhumation Hi > Hi1 rate must satisfy both conditions.

55 Extracting erosion rate from rivers detrital age distributions 3.2.5 Procedure summary for main trunk distributions To obtain estimates of erosion rate in each catchment, we then proceed sequen- tially for i = 1,…, M, where M is the number of locations within the river where we

have an age distribution. For each site i, we fi rst compute δi according to: i1 H k  H k H k  H k i1 i i i1 (3.17) i = max (Aii/ Aj j , k , k ) k=1, L ,N  j=1 Hi 1 Hi

From this value of δi, we can deduce an erosion rate (relative to the exhumation

rate in the fi rst catchment, ε1) from: i1 i i = Aj j j (3.18) Aii j=1

as well as the relative concentration of grain of age in bin k in Area Ai , using:

k k k Hi  Hi1 k Ci =  Hi (3.19) i

for all k = 1, L , N . For the fi rst catchment, i.e. i = 1 , we assume that 1 = 1 and k k Ci = Hi .

3.2.6 Using age distributions from tributaries Age distributions from tri butaries can be included to improve the solution locally,

i.e. in the catchment that includes the tributary. Let’s call At, αt and εt the catchment area, the surface rock density and mean exhumation rate of the catchment of the

tributary and AM, αM and εM the catchment area, the surface rock density and mean

exhumation rate of the rest of catchment Ai. For each bin k in the catchment i , we can write:

k k k (3.20) AiiiCi = ATTT CT  AM  MM CM By conservation of eroded rock mass, we have:

AM  MM = Aiii  ATTT (3.21) which we can use to transform Equation (3.20) into:

k k k (3.22) AiiiCi = ATTT CT  (Aiii  ATTT )CM

56 Chapter 3 to obtain: k k k AiiiCi  ATTT CT CM = (3.23) Aiii  ATTT Using the method for the main trunk data described in the previous sections, we k k k k know ε and C . The tributary data (age distributions) gives us the ( ) i i CT CT = HT and we can solve for the assuming fi rst that the exhumation rate is uniform in th catchment i, i.e. εT = εM = εi, to give: A C k  A  C k C k = i i i T T T (3.24) M 3 Aii  ATT

k However, this may lead to unrealistic values of the CT , i.e. not comprised be- k tween 0 and 1. Consequently, two conditions need to be added so that 0 < CT < 1 for all k which might yield to exhumation rate estimate in the tributary catchment, εt, dif- ferent from ε obtained for A . The fi rst condition (C k > 0 ) yields: i i M k AiiCi T < k i (3.25) ATT CT

while the second condition (C k < 1) yields: M A (1  C k )  < i i i  (3.26) T A  (1  C k ) i T T T The true exhumation rate must satisfy both conditions and we, therefore, select the smallest value of εT obtained by considering any relative surface concentration difference between the tributary sub-catchment concentration ( k k CT ) and that of the entire catchment (Ci ).

3.2.7 Procedure summary for tributary distributions In summary, we fi rst compute the exhumation rate in the sub-catchment of the tributary, according to: k k AiiCi Aii (1  Ci ) T = min (i , k i , k i ) (3.27) k=1, L ,N ATT CT ATT (1  CT )

k and we then use it to compute the CM according to:

57 Extracting erosion rate from rivers detrital age distributions k k k AiiiCi  ATTT CT (3.28) CM = Aiii  ATTT from the values of k and ε obtained from the trunk data analysis and the Ci i obtained form the measured distribution in the tributary. If there is more than one tributary in a catchment, we repeat the operation for k each tributary, using the previously computed εi and Ci from the main trunk analy- sis, under the assumption that the tributaries have disconnected drainage areas in the catchment i.

3.2.8 Uncertainty estimates by bootstrapping

We assess the uncertainty of our estimate s of erosion rate εi and relative concen- trations by bootstrapping. For this, we simply use the method described above on a large number of sub-samples of the observed distributions constructed by arbitrarily and randomly removing 25% of the observed age estimates. This yields distribution of exhumation rate and relative concentrations that can be used to estimate the uncer- tainty arising from the fi nite sample size. These distributions are usually not normal and we use their modal value, rather than their mean, as the most likely estimate of exhumation rate and their standard deviation to represent uncertainty. The code is provided as a Jupyter Notebook containing R-code and explanatory notes that refer to the equations given in this manuscript. The user must provide a se- ries of input fi les containing (a) the description of the sites, i.e. the order in which the sites are located along the river, whether they drain into the main river stream or into a tributary, the drainage area A , the lithological factor α, (b) the bin sizes and (c) the observed age data at each site. The code produces estimates of exhumation rate and relative surface rock concentration of grains of ages in each range, their mean value, standard deviation and modal values (from the bootstrapping).

3.3 Applications to detrital age distributions To illustrate the method, we now apply it to a detrital datasets from the Eastern Himalaya. The ages correspond to cooling ages, i.e. the time in the past when the rocks cooled through a given closure temperature. The datasets that we use contain ages that were obtained using the muscovite 40Ar/39Ar thermochronometer, which has ◦ a closure temperature Tc ≈ 385 ±70 C (Hames and Bowring, 1994), depending on grain size, chemistry, and cooling rate.

58 Chapter 3 Table 3.1. Age bins used to construct age distributions shown in Figure 3.4 and used in our example. Bin 1 Bin2 Bin 3 Bin 4 0-5 Ma 5-10 Ma 10-20 Ma 20-50 Ma

Age distributions were constructed from published age datasets collected along the main trunk of the Tsangpo-Siang-Brahmaputra river system, as well as along some of its tributaries (Figure 3.3), using age bins given in Table 3.1. Samples A, B, C (composite sample), X and Y are from Lang et al. (2016) and samples Z, T-40a and 3 T-41a are from Bracciali et al. (2016). The complete age datasets are given in the Data Repository, Table S1. In Table 3.2, we give the relative position of the successive sam- ples along the main trunk of the river, i, the respective exclusive contributing areas, Ai and the lithological factor, αi, (or abundances of target mineral in surface rocks). Table 3.2. Relative position along the main trunk of the Tsangpo-Siang-Brahmaputra river system. Negative numbers indicate samples collected along a tributary. Catchment areas and lithological factors used to compute the erosion rate reported in Table 3.3. Site names and catchments area refer to locations shown in Figure 3.3.

Site Position Catchment Area (km2) α parameter Reference TG-40a 1 55395 0.01 Bracciali et al., 2016 TG-41a 2 13265 0.03 Bracciali et al., 2016 A 3 41374 0.0135 Lang et al., 2016 Y -4 1250 0.013 Lang et al., 2016 B 5 2092 0.013 Lang et al., 2016 X -6 2135 0.011 Lang et al., 2016 C 7 1451 0.0123 Lang et al., 2016 Z 8 111706 0.0131 Bracciali et al., 2016

Results are shown in Table 3.3 as computed relative erosion rates (i.e. normalized such that the mean erosion rate is 1), standard deviations and modal values. Figure 3.5 contains maps of the various catchments shaded according to their predicted modal erosion rate and concentrations of grains of age within each range, obtained from the bootstrapping and mixing algorithms described above. Predicted concentrations are scaled such that the sum of the fi ve age bin concentrations is 1 in each catchment. We see that predicted erosion rates increase with distance along the main river trunk from its source area along the southern margin of the Tibetan Plateau. Maximum erosion rates are observed in catchment C that is closest to the eastern Himalayan syntaxis. Further downstream (catchment Z), the predicted erosion rate remains high but lower than observed near the syntaxis.

59 Extracting erosion rate from rivers detrital age distributions Figure 3.3. Location of the study area and b) location and name of sampling sites and geom- etry of the drainage basins contributing to each site.The orange shading represents catchments draining directly into the main trunk; pale blue shading represents the tributary catchments or sub-catchments.

60 Chapter 3 3

Figure 3.4. Observed distributions of ages (light grey bars) in samples collected at sites shown in Figure 3.3. and predicted surface age distributions (dark grey bars) in corresponding catch- ment areas. Data and results are shown for the sites along the main trunk only.

61 Extracting erosion rate from rivers detrital age distributions Table 3.3. Computed relative erosion rates, variance and modal values obtained from the mix- ing model and the bootstrapping procedure. Values are normalized such that the mean is 1. Site names refer to locations shown in Figure 3.3. Site Mean erosion1 SD1 Modal erosion1 TG-40a 0.0122 0.0000 0.0122 TG-41a 0.0166 0.0105 0.0118 A 0.0588 0.0307 0.0398 Y 0.9209 0.5634 0.6021 B 0.9236 0.5618 0.5994 X 0.4758 0.2698 0.3100 C 4.5130 2.9283 2.8189 Z 0.4758 0.2698 0.3100 1 The erosion rates exstimates are expressed as relative numbers. They do not refer to any unit of measure.

Interestingly, there is a good correspondence between present-day erosion rate and where the youngest ages are being generated (compare upper left panel showing relative concentration of youngest age bin, to central panel showing predicted pre- sent-day erosion rate), with the notable exception of the most downstream catchment (Z). In other words, where the mixing analysis predicts high erosion rate to account for a substantial change in the age distribution between two adjacent catchments, is also where it predicts the highest concentration of young ages in the surface rocks. At the downstream end of the river (Catchment Z), we predict a relatively high erosion rate from the mixing model but a relatively low concentration of young ages in com- parison to the other catchments. This could mean that, in catchment Z, the present-day high erosion rate is relatively recent and has not led yet to a complete resetting of cooling ages which were set during earlier events. We also note (Table 3.3 and Figure 3.6) that all erosion rate distributions predicted by the bootstrapping method are highly asymmetrical, as the median value is always signifi cantly smaller than the mean. The standard deviation is large, of the order of 30- 50% of the mean value or 50-100% of the modal value of predicted erosion rate val- ues. Interestingly, the standard deviation does not increase downstream which dem- onstrates that the uncertainty introduced by using incomplete or non-representative sub-samples of the true distributions at each of the station does not accumulate as our algorithm proceeds from station to station. This results from the incremental nature of our algorithm, as shown in Equation 3.11. One of the main sources of error/uncertainty in our estimates of the erosion rate

comes from the assumed value of the lithological factors, αi, which might be diffi cult

62 Chapter 3 to estimate in many situations. We can compute the uncertainty on the erosion rates,

∆ϵi arising from the uncertainty on the lithological factors, ∆αi, from:

௜ ߲ߝ ଶ ௜ ଶ (3.29) οߝ௜ = ඩ෍൬ ൰ οߙ௞ ߲ߙ௞ ௞ୀଵ

Where: 0 ݂݅ ݇ >1 3 ߝۓ ௜ ۖ ߲ߝ௜ ݂݅ ݇ =1 = ߙ௜ (3.30) ߲ߙ௞ ௜ିଵ ߜ௜ ߲߳௜ ۔ ݅ > ݇ ݂݅ ௝ ߙ௝ ቇܣ௞߳௞ + ෍ܣቆ ۖ ௜ߙ௜ ௝ୀଵ ߲߳ఈ௞ܣ ە

63 Extracting erosion rate from rivers detrital age distributions Bin 1: 0-5 Ma Bin 2: 5-10 Ma

1 0.5 0.2 0.1 0.05 0.02 0.01 >0.01

T-41a A T-40a B X Y C Modal value >1 1 0.5 0.2 0.1 0.05 0.02 Z 0.01 >0.01 Bin 3: 10-20 Ma Bin 4: 20-50 Ma

1 0.5 0.2 0.1 0.05 0.02 0.01 >0.01 Bin 5: 50-500 Ma

Figure 3.5. Predicted modal erosion rates (central panel) and relative surface age concentra- tion from the Muscovite detrital data from Eastern Himalaya. See Figure (3.4) for data distri- bution.

64 Chapter 3 3

Figure 3.6. Distributions of predicted present-day erosion rate in mm/yr as derived by boot- strapping. See Figure (3.3) for sites locations.

The results are shown in Figure 3.7 as a plot of the ratio between the relative un- certainty in estimates of erosion rate ∆ϵi/ϵi and the relative uncertainty in lithological factors, ∆αi/αi, for the six stations located along the main river trunk. We see that the relative uncertainty in erosion rate is approximately proportional to the relative uncer- tainty in lithological factor (i.e. all values are close to 1) and that there is only a minor downstream propagation of the uncertainty. This is also a simple consequence of the incremental nature of our algorithm, as explained by Equation 3.11.

65 Extracting erosion rate from rivers detrital age distributions Figure 3.7. Relative uncertainty in erosion rate scaled by the relative uncertainty in the litho- logical factor for the estimates obtained at each of the six sites along the main river trunk. The fi rst site has a fi xed erosion rate and therefore no uncertainty.

3.3 Con clusions We have developed a simple method to extract spatially variable erosion rates and surface age distributions from detrital cooling age datasets from modern river sands. The method is based on what we believe are the simplest assumptions necessary to interpret such data. In describing the method we demonstrate that it is well suited to extract from detrital cooling age datasets two seemingly independent sources of in- formation pertaining to the spatial distribution of present-day erosion rate along the river. By applying the method to an existing dataset from the eastern Himalaya, we show that the method provides estimates of present-day erosion rate patterns in the area, potentially evidencing that the fast present-day erosion rates in some parts of the study area are relatively young. Importantly, the method is limited to providing the spatial distribution of erosion rate; independent information is necessary to transform those into absolute estimates of erosion rates.

66 Chapter 3 Supplementary information The data used for the inversion are summarized in Table S1 and can be down- loaded at: https://doi.org/10.6084/m9.fi gshare.5594218

3

67 Extracting erosion rate from rivers detrital age distributions Chapter 4

A new detrital mica 40Ar/39Ar dating approach for provenance and exhuma- tion of the Eastern Alps

L. Gemignani Xilin Sun J. Braun T.D. van Geerve J.R. Wijbrans

This chapter is based on Gemignani L., Sun X.L., Braun J., van Gerve T.D., Wijbrans J.R. (2017). A new detrital mica 40Ar/39Ar dating approach for provenance and exhumation of the Eastern Alps. Tectonics, 36, doi: 10.1002/2017TC004483. Abstract

Detrital thermochronology can be used as a tool to quantitatively constrain exhu- mation rates and its spatial variability from active mountain belts. Commonly used methods for this purpose assume a steady-state relationship between tectonic uplift and erosion. However, this assumption does not account for the transitory response of a dynamic orogenic system to changes in the boundary conditions. We propose a different approach that uses the observed detrital age distributions as “markers” of the past exhumation, and of the present-day erosion and mixing occurring in a river system. In this paper, we present new 40Ar/39Ar biotite and white mica age distributions for nineteen modern river sands from the Eastern Alps north of the Periadriatic line. The results present three main clusters of ages at ~0.5-50, ~60-120, ~250-350 Ma that 4 record the main orogenic phases in this sector of the Alps. We have applied two numerical methods to the cooling ages to a) linearly com- pute the spatial variability of the relative present-day erosion of a set of 4 detrital mineral samples from drainage basins along the Inn river, and b) quantify the rates of the cooling and erosion in the Tauern Window during Paleocene-Miocene time of the Alpine orogeny. Our results suggest a 0.34-0.84 mm/yr range of exhumation rates for the Tauern Window since the Miocene. Our estimates of exhumation rates of the western Tauern Window are higher than those for the eastern Tauern Window, which is consistent with the previous studies. 4.1 Introduction

Topography in orogenic belts is the result of surface uplift of the rocks caused by the competition of internal tectonic forces and external surface processes, weather- ing and erosion, both acting to drive tectonic exhumation. The rate of erosion can be constrained by thermochronology on detritus in the sediment record in modern rivers, the foreland basin or of minerals obtained from the bedrock as exposed in crystalline cores of mountain belts. Each of these different approaches has been applied to vari- ous sectors of the Alps to constrain exhumation and present-day erosion (Garver et al., 1999; Carrapa, 2004, 2009; Von Eynatten and Wijbrans, 2003; Wölfl er et al., 2016; Bertrand et al., 2017). Applications of detrital thermochronology on both retro- and pro-wedge basin sedimentary rocks have been applied on the Western Alps (Bernet et al., 2004; Carrapa, 2009, 2016; Garver et al., 1999), Central Alps (Spiegel et al., 2000, 2004; Von Eynatten and Wijbrans, 2003) and Eastern Alps (Kuhlemann et al., 2004).

71 Testing the 40Ar/39Ar detrital approach: Eastern Alps Rivers in the Alpine domain transport sediment from the high mountains to the foreland and thus their sediment load contains key information on sediment prove- nance, on the age range of rocks contributing to the sediment load, and on exhumation in the source area. However, so far, only relatively few studies have focused on the link between foreland basin record and mountain surface processes from Alpine river sediments (Bernet et al., 2004, 2009; Carrapa et al., 2004; Glotzbach et al., 2011; Reiter et al., 2013). In the present study, we use 40Ar/39Ar dating of white mica and biotite single crys- tals using a laser fusion technique. White mica and biotite have limited resistance to physical abrasion and chemical weathering and relatively low closure temperatures (350 - 425°C and 300 - 350°C, respectively) (Harrison et al., 2009; McDougall and Harrison, 1999), and therefore are well suited to reveal information concerning recent tectonic events in their source areas. White mica and biotite age signals record cool- ing and exhumation from crustal depths of ~10-12 km in the orogen, assuming a nor- mal geothermal gradient of 25⁰C/km and commonly adopted values for the closure temperature for the two target minerals. In recent years, detrital mineral datasets have been collected from the modern river sediments world-wide in several studies, and new numerical methods have been proposed to extract quantitative information about the history and the lateral variation of the erosion/exhumation of active regions (e.g. Rhul and Hodges, (2005)). Brewer et al., (2006), for example, used a new approach to a set of detrital samples from the Marsyandi River (Nepal) and its tributaries to derive, by mixing, the theoretical and observed detrital cooling age, and the variation of erosion rates from confi ned neighboring catchments. Enkelmann et al., (2015) coupled detrital age distributions of low-temperature thermochronometers from river sand samples and from sedimen- tary basins to reconstruct the thermo-kinetic evolution of a glaciated landscape. Using a thermal model, Bracciali et al. (2016) compared their multi-thermochronometer dataset with the computed heat fl ow solutions to infer the thermal evolution of the “pop-up” of the Namche Barwa syntaxis (Eastern Himalaya) coupled with the Siang- Brahmaputra river capture. However, most of such methods rely on the assumption of a steady-state between tectonic uplift and erosion over known geological time in- tervals. This assumption precludes a test of the transitory response of the orogeny to changes in tectonic and climate forcing. In this paper, we applied a new linear inversion approach (Mixing model) to the “raw” detrital age distributions from river sand samples to extrapolate information about the relative spatial-variability of the present-day erosion from catchments along

72 Chapter 4 the river trunk and from in-between tributaries. We use detrital cooling age peaks (bins), as a set of “colors” or “passive markers” that, without relying on any thermal parameters or steady-state assumption, allow to infer the present-day mixing of dif- ferent eroding sources in the river drainage system. The model is based on a limited number of important and measurable parameters that describe the lateral variation of lithology among different exclusive catchments and on the binned raw age distribu- tions. The bins are arbitrarily chosen to represent known tectonic events in the area under consideration. The method is highly dependent on the cooling-age distributions, and to test whether 20-40-60 single grains are representative of an entire catchment, we compared our age distributions with surface in-situ dates from literature. The Alps form an appropriate test-case for such an approach as there is a sub- stantial database of ages obtained from minerals in the crystalline source rocks (i.e. 4 Hunziker et al., 1992; Luth and Willingshofer (2008), Scharf et al., 2013). In the fi rst part of the paper, we compare our new single grain river sand biotite and white mica age data from modern rivers with in-situ thermochronology to test for consistency with the patterns found in crystalline basement rocks. In the second part of the paper, we apply our new numerical approach to three Inn river-trunk catchments and to one of its lateral tributaries to predict the relative present-day erosion rates of the catch- ment areas. Finally, the detrital white mica and biotite cooling ages of fi ve samples collected from three rivers draining the Tauern Window are used to constrain the relatively young (Miocene) exhumation of the area. Using this simple approach, we derive fi rst-order exhumation rates of ~0.10-0.65 mm/yr for the ~45-7 Ma period. 4.2 Geological summary of the Eastern Alps

The Alpine orogen is the result of the convergence, subduction and collision be- tween the Adria and Eurasian plates since the Cretaceous (Frisch et al., 1998; Stamp- fl i et al., 1998; Berger et al., 2008). The current confi guration of the Alps is caused by a complex geodynamic evolution characterized by three major orogenic phases known as Pre-Alpine or Varisican, Eo-Alpine, and Alpine. The Late Paleozoic Vari- sican orogeny affected the Austroalpine basement nappes (Fig.4.1) and was followed, in the Late Paleozoic-Mesozoic, by the break-up of Pangea leading to the rifting of the ocean basins of the Neothetys (Handy et al., 2010). The northern branch of the Neothetys refers to the opening during the Middle-Jurassic of the Meliata-Maliac- Vardar (Meliata) ocean and is kinematically linked to the rifting of the Atlantic ocean (Handy et al., 2010). During the Early Cretaceous, the inversion of the Mesozoic ba- sin-related extensional phase resulted in a new pulse of contractional deformation that closed the Meliata ocean. During the Cretaceous orogenic phase, the Austroalpine

73 Testing the 40Ar/39Ar detrital approach: Eastern Alps acted as lower plate and was affected by the Eo-Alpine metamorphic overprinting (Frisch and Gawlick, 2003) (Fig. 4.1). The Early Cretaceous orogeny that caused the stacking of the Austroalpine units in a pro-and-retro wedge position, was followed by a prolonged period of extension during the Late Cretaceous. This extensional phase was accompanied by topography development and erosion, causing sediment depo- sition in intra-orogenic sedimentary basins (Dal Piaz et al., 2003; Froitzheim et al., 1994). A new stage of contractional deformation led during the Tertiary to the “Alpine” orogeny characterized by crustal-scale folding coupled with orogen-parallel exten- sion and lateral-extrusion processes. Lateral extrusion, in turn, led to the exhumation and over-thrusting of the high-grade rocks of the Penninic units during the Cenozoic whereas the upper plate Austroalpine units remained unaffected by tectonic overprint- ing because they were located in the upper crust (Schmid et al., 2004, 2013; Stampfl i et al., 1998). Alpine overprinting can be found in the metamorphic domes of the East- ern Alps located in the Tauern Window. Such domes formed due to the exhumation of deeply buried units during the Neogene (Neubauer et al., 1999).

74 Chapter 4 o o 46 o 47 48 ed blue lines the major

Vienna EA18 o EA19 16 o EA17 Maribor 16 EA15 EA16 EA14 4 o EA13 14 o Trieste 14 Fault systems systems Fault Rivers EA10 EA11 Po Plain Po EA12 EA9

periadriatic line Tauern window Tauern o o ed after (Schmid et al., 2004). The main litho-tectonic units are indicated in the ed after (Schmid et al., 2004). 12 fi EA8 12

Munich EA7 Molasse EA6 Austroalpine nappes Austroalpine nappes Penninic Varisican basement Varisican Nappes Helvetic EA5 Engadin window EA3 o o 10 10 EA4 EA2 EA1 Milano ed tectonic map of the Eastern Alps modi ed tectonic map of the Eastern fi

rich Tertiary-Pliocene Covers Tertiary-Pliocene intrusions Tertiary Southern Alps Northern Alps Calcareus (Upper Austroalpine)

u

Z Simpli Legend of the tectonic units Legend o o o 47 48 46 river paths. Figure 4.1. Figure legend together with the major tectonic discontinuities (red lines). The white stars indicate the samples location and dott legend together with the major tectonic discontinuities (red lines).

75 Testing the 40Ar/39Ar detrital approach: Eastern Alps Our work is focused on the north-verging sector of the Eastern Alps, north of the Periadriatic line. In this sector, the main tectonic terranes are from the internal to the external side of the Austroalpine nappe system (Adriatic passive continental margin), the Penninic metamorphic nappes system and the Helvetic zone that was thrusted on top of the Molasse foreland (Fig. 4.1). The Austroalpine nappes are made of sedi- mentary cover units and basement that underwent Early-mid Cretaceous (Eo-Alpine) metamorphism and that over-thrusted the Mesozoic Penninic units that bound the Tauern Window and the Engadin Window (Dal Piaz et al., 2003; Schmid et al., 2008). The Austroalpine nappe complex includes the Northern Calcareous Alps (NCA), situ- ated at the northern margin of the European accretionary wedge (Frisch and Gawlick, 2003), and crystalline units including ortho-gneisses, para-gneisses, and mafi c intru- sives, dykes and gabbros of Permian age (reference, to Mohn, Petri, Manatschal). The Austroalpine domains consist of pre-Alpine crystalline basement rocks, low-grade Paleozoic blocks (phyllitic and grauwacke rocks) and post-Variscan sedimentary se- quences (calcareous rocks) (Frisch et al., 1998). The stacking in the Austroalpine nappe complex is classically related to the subduction of the Piedmont-Liguria Ocean during Cretaceous and to the collision, during the Tertiary, with the Penninic base- ment (Liu et al., 2001; Pfi ffner, 2001). 4.2.1 The Tauern Window In the Eastern Alps, the most recent exhumation associated with crustal-scale ex- tension is localized along the axial zone of the belt, where the Penninic units of the Tauern Window show the youngest cooling ages as recorded by different thermochro- nometric systems in the region (Bertrand et al., 2017; Frisch et al., 1998; Fodor et al., 2008; Luth and Willingshofer, 2008; Rosenberg et al., 2009; Viola et al., 2001; Wölfer et al., 2011) (Fig. 4.2). The Penninic units were overprinted during the burial and exhumed due to the gravitational collapse of the Eastern Alps coupled with substan- tial lateral extrusion toward the east along a conjugate system of shear zones during the Miocene (Frisch et al., 1998; Ratschbacher and Frisch, 1991). The Tauern Win- dow and the Engadin Window have been described as a crustal-scale antiformal stack bounded by Austroalpine basement rocks (Liu et al., 2001; Schmid et al., 2013). The present-day surface rock units and cooling ages of the Tauern Window present a con- trasting cooling history when compared with the adjacent Austroalpine surrounding units. The exhumation of the Tauern Window, associated with localized deformation, occurred from early to late-Miocene (Rosenberg et al., 2009). Lateral extrusion start- ing from the mid-Miocene was associated with an N-S contractional phase defi ned by the displacement at high angles to the strike of the mountain range (Neubauer et al.,

76 Chapter 4 1999; Ratschbacher and Frisch, 1991; Robl et al., 2008; Wölfer et al., 2008). In-situ thermochronology iso-age contours are subparallel to the axial plane of the large-scale folding, suggesting that folding coupled with erosion was the mecha- nism controlling the exhumation of the Tauern Window during Miocene (Bertrand et al., 2017 and reference therein). This interpretation is an alternative to models that emphasize the role of the eastward lateral extrusion associated with lateral extension components and subordinate N-S normal faulting (Frisch et al., 1998; Neubauer et al., 1999; Robl et al., 2008) and argues for a more complex exhumation history where the steady-state assumption may have been fulfi lled only for limited periods of time starting from Late Oligocene (Neubauer et al., 1999; Kuhlemann et al., 2006). In the Penninic units of the Tauern Window, a diachronous character of the cooling has been described in cooling maps displaying the results of different thermochronometers (e.g. Luth and Willingshofer, 2008). Contrasting rates of Miocene exhumation from high to 4 low between western and eastern sectors of the Tauern are derived from numerical in- version of low-temperature cooling ages (Bertrand et al., 2017 and reference therein). In the western Tauern Window, younger cooling ages post-date by ca. 5 Ma the age patterns found in the eastern sector of the window. 4.3 Detrital 40Ar/39Ar analytical method

Nineteen modern river sand samples were collected from rivers in the Eastern Alps between eastern Switzerland and Liechtenstein in the west, Austria in the centre, and northeast Slovenia in the southeast of the study area (Fig. 4.1 and Table. 4.1). The sampling sites were chosen at least 1km away from tributary junctions to the main trunk of the river and any landslides to avoid bias toward one particular source in the main river stream. Approximately 2kg of medium grained sand was collected from the top 10 cm sediment at each sampling location from the edge of the active river channel. Biotite and muscovite were separated for radio-isotopic 40Ar/39Ar analysis. Min- eral separation was performed using standard procedures in the mineral separation laboratory at the Vrije Universiteit Amsterdam. Organic material was removed by density separation. The samples were sieved to obtain 400-200 μm grain size frac- tions. A Faul vibration table was used to separate the fl at micas from the non-fl at minerals. Biotite and muscovite were separated from each other by heavy liquid sepa- -3 -3 ration (ρmuscovite=2.77-2.9 g cm , ρbiotite=2.9-3.3 g cm ) and by making use of the higher magnetic susceptibility of biotite in a Franz isodynamic magnetic separator. Finally, all samples were hand-picked under a binocular microscope to remove any signifi cant weathering variation or inclusions and to obtain 200-250 grains of pure muscovite and

77 Testing the 40Ar/39Ar detrital approach: Eastern Alps biotite. After separation, the samples were wrapped in Al-foil and loaded in 9 mm ID quartz tubes together with the monitor standard Drachenfels sanidine dated at 25.43 ± 0.03 Ma. This value is compatible with the set of (Kuiper et al., 2008). Samples were irradiated at the Oregon State University TRIGA reactor in the CLICIT facility for 12 hours. Muscovite and biotite ages determinations were conducted at the argon geochronology laboratory of the Vrije Universiteit of Amsterdam. Single white mica or biotite grains were loaded into a copper disk with 185 holes of 2mm-diameter and 3mm-depth. The copper disk was heated overnight at 150°C in an ultra-high vacuum sample house fi tted with a multispectral ZnS externally pumped double vacuum seal

window. Single mica grains were fused under a 25W Synrad CO2 laser instrument. The gas was cleaned in a sample purifi cation system by exposure to SAES St707 (Fe-V-Zr alloy) getters and a SAES NP50 getter device (Hiden based system) or a SAES NP10 getter (Helix system) each fi tted with ST101 cartridges. The Ar isotope spectrum was analyzed on ThermoFisher Helix MC+ multi-collector noble gas mass spectrometer (Helix) and Hiden HAL 3F Series 1000 Pulse Ion Counting Triple Filter quadrupole mass spectrometer (Schneider et al., 2009). A system blank was measured before every fi fth sample measurement and three blanks and two gas pipette air aliquots were analyzed at the start and the end of the tray measurement respectively. Total system blank levels were approximately 2-6×10−17 moles for 40Ar and 0.5-6×10−18 moles for 39Ar,38Ar and 36Ar, and 1-2×10−17 moles for 37Ar. The data reduction software ArArCALC2.5 was used for data reduction and age calculation (Koppers, 2002). Corrections were applied for 37Ar and 39Ar decay follow- ing sample irradiation and for procedure blanks.

78 Chapter 4 Table 4.1. Samples locationsa River Lab-ID Longitude Latitude Minerals Rhine EA1 9°39′39″ 47°26′58″ Ms and Bt Rhine EA2 9°35’53″ 47°13’59” Ms and Bt Inn EA3 10° 4’45″ 46°44’52” Ms and Bt Inn EA4 10° 4’23″ 46°44’58” Ms and Bt Inn EA5 10° 38’14″ 47° 6’50” Ms and Bt Inn EA6 11° 4’27” 47°18’4” Ms and Bt Inn(Sill) EA7 11°27’27” 47° 6’29” Ms and Bt Inn(Ziller) EA8 11°51’51″ 47°20’51” Ms and Bt Salzach EA9 12°21’14″ 47°16’18” Ms and Bt Salzach EA10 13°11’47″ 47°20’21” Ms and Bt 4 Salzach EA11 12°56’21″ 47°56’32” Ms and Bt Tiroler EA12 12°30’19″ 47°49’10” Ms Enns EA13 14° 5’15″ 47°30’58” Ms and Bt Mur EA14 15°13’50″ 47°24’6” Ms and Bt Mur EA15 15°31’27” 46°53’1” Ms and Bt Drau EA16 15°29’51″ 46°32’37” Ms and Bt Raab EA17 15°45’5″ 47° 3’44” Ms and Bt Raab EA18 16° 9’57″ 46°59’52” Ms Leitha EA19 16°10’13″ 47°43’44” Ms aSummary of sample number, rivers and locations expressed as longitude and latitude. Min- erals indicate the type of target analysis (Ms = muscovite; Bt = biotite).

4.4 Age distributions and comparison to existing in-situ ages

The analysis of detrital modern river sand samples has produced a remarkably representative fi ngerprint of the bedrock ages drained into the basin by erosional and tectonic (i.e. exhumation along faults zones) processes as demonstrated by several studies over the past decades (Bernet et al., 2004; Brewer et al., 2006a; Garver et al., 1999). Three major exhumation pulses (Permo-Carboniferous, Late-Cretaceous, and Eocene-Miocene) are recorded for both biotite and white mica as shown in the composite total age Probability Density Plots (PDP) (Fig. 4.2). The relative peaks of muscovite ages are generally higher for the Varisican and Eo-Alpine events in com- parison with biotite ages that records mostly Alpine Cenozoic exhumation.

79 Testing the 40Ar/39Ar detrital approach: Eastern Alps PDF N=444 Biotite Muscovite Relative probability Relative N=326

0 50 100 150 200 250 300 350 400 450 500 Age (Ma)

Figure 4.2. Composite Probability Density Plots for the biotite distributions (continuous line) and for the muscovite (dotted line); On the x-axis, the ages are expressed in million years and on the as a relative probability on the y-axis. The number of single grain analysis is indicated for both the target minerals.

We compared the consistency of our detrital signal with existing in-situ bedrock ages (references are shown in Fig. 4.3). The contribution of the source to the drain- age basins is tracked in the detrital age distribution by an age-related color code (Fig. 4.3). The overall detrital mineral ages were plotted together as histograms and Ker- nel Density Estimator (KDEs) (Vermeesch, 2012). Comparison of the detrital dis- tributions with the in-situ thermochronological data allows deriving multiple pieces of information. The fi rst piece of information comes from the observed cooling-age distributions from the river samples, which give an insight into how much a cooling event has imprinted the source surface unit rocks, under the assumption that it is representative of the entire catchment area. The second piece of information comes from the present-day mixing of the signal in the river, which gives an insight into the present-day erosion in the regional-scale geology. In this section, we will describe the river sand sample cooling age distributions and how they compare to the available in-situ thermochronological data and which additional information they bring to our understanding of the regional scale tectonics. Samples from the Rhine EA1 and EA2 sites and from the Inn EA3, EA4, EA5 sites contain minerals that originate in the internal Penninic and Austroalpine base- ment nappes and Northern Calcareous Alps. The bedrocks ages available for those basins area are varied and range from 400 Ma to ~10 Ma (Handy et al., 1996.; Von Eynatten et al., 1996; Challandes et al., 2003; Wiederkehr et al., 2009). Detrital min- eral ages from the Rhine and Inn samples are consistent and homogeneous for the two target minerals, as 70% of the total distributions are in the range 298±2 to 290±2 Ma. Downstream from the Inn river, sample EA06 presents a more scattered distribution with muscovite peaks at ages 80±20 Ma (40%), 300±20 Ma (50%) and 250 Ma. The biotite ages of EA6 concentrate on a peak at 80±20 Ma that includes 70% of the total distribution with the other 30% of measurements showing a broad scattering of older ages (Fig. 4.3). The Sill (EA7), Ziller (EA08) and the Salzach (EA9 and EA10) samples contain minerals from catchments draining the Penninic nappes of the Tauern Window. These samples yield muscovite and biotite grains younger than 50 Ma and are consistent 4 with a major cluster of published in-situ ages around Paleogene ages (Kurz et al., 2008; Liu et al., 2001; Ratschbacher et al., 2004; Scharf et al., 2013; Warren et al., 2012; Zimmermann et al., 1994). Downstream, the Salzach sample (EA11) contains minerals coming from the external Austroalpine units towards the north. This sample displays a bimodal distribution in the biotite age population with two narrow peaks at 40±5 and 300±20 Ma. In the muscovite data set, the <50 Ma peak is dominant and represent 95 % of the distribution, with one statistically unrepresentative single grain of 300±20 Ma. The sample downstream of the Salzach basin (EA11) contains mostly muscovite and biotite grains that are derived from the Tauern Window. Sample EA12 from the Tiroler Achen river received minerals from the Northern Calcareous Alps units; the detrital white mica ages distribution is made of a narrow peak at 300 Ma. The Enns (EA13) sample was collected in the middle of the Northern Calcareous Alps unit that is characterized by a range of muscovite bedrocks ages of 300-200 and 100-50 Ma (Liu et al., 2001). This sample yields a heterogeneous set of ages from 450 to 60 Ma in the biotite distribution, whereas, the white mica ages from the same samples form a 90-60 Ma peak. The Mur River where samples EA14 and EA15 were collected originates in the metamorphic internal zone of the Penninic nappes (Tauern Window) which has char- acteristic in-situ ages of 0-50, 100-200 and 200-300 Ma and fl ows east towards the Austroalpine Nappe units (Liu et al., 2001; Neubauer et al., 1995; Scharf et al., 2013). Downstream from sample EA14 site located in the Northern Calcareous Alps units, the river bends southward and fl ows across the Autroalpine Nappes units and Ceno- zoic cover. The Mur detrital age distribution presents a narrow peak at 80±20 Ma. The young (Miocene) age signal of the Tauern Window is not found in the distributions

81 Testing the 40Ar/39Ar detrital approach: Eastern Alps observed in the Mur river. The Drau river (EA16) originates in the Penninic units of the Tauern Window south of the main ranges and drains along a west-to-east section of the internal units of the Austroalpine basement nappes. Published bedrocks ages are comprised of several intervals, i.e. 50-5 Ma, 100-50 Ma, 300-200 Ma and 400-300 Ma (Ratschbacher et al., 2004; Wiederkehr et al., 2009) for the muscovite and 100-50 Ma for the biotite (Wiederkehr et al., 2009). This wide range of ages is recorded in our detrital musco- vite data also, which display peaks clustering around 80±3 Ma (70 %), 300±2 Ma (10%), 180±2 Ma (5%), 140±2 Ma (5%) and 47 ±10Ma (10%). A minor 10-30 Ma peak, related to the internal metamorphic core of the Tauern Window, is present in the biotite age distribution. Samples EA17 and EA18 from the Raab river receive mineral grains from Aus- troalpine nappe and Tertiary cover units that yield mostly pre-Varisican and Varisican age signatures and no Alpine overprinting as observed from several in-situ age analy- ses (Dallmeyer et al., 1998). Our detrital muscovite age distribution shows peaks at 240-310 Ma, 310-300 Ma and 200-80 Ma which is consistent with observed in-situ ages. Sample EA19 (for which we only have a muscovite age distribution) is derived from a location draining the Northern Calcareous Alps and the Austroalpine basement nappes and present circa 70% of the muscovite ages ranging within 298±2 Ma and 290±2 refl ecting a univocal Varisican range of ages. As described above our detrital basin-related age distributions are generally com- parable with the bedrocks ages, although information on in-situ bed rock ages is by necessity more fragmentary. Interestingly, the most recent Alpine (< 50 Ma) pulse of exhumation and metamorphism is spatially confi ned to the Tauern Window and En- gadin Window and within the catchments draining those units (Fig. 4.3). As pointed out earlier, these areas have experienced exhumation since the Oligocene along major normal faults during episodic lateral (E-W) extension in a convergent (N-S) steady re- gime (Ratschbacher and Frisch, 1991). The < 50 Ma cooling signal is recorded in the tributaries draining the Tauern Window and is transported downstream in the Salzach river; but is not seen in the catchments of the Mur, Drau, and Enns rivers. Those rivers mostly record pre-Alpine and Varisican exhumation of the Austrolap- ine units. The Varisican signal is pervasive and constitutes the major peak in all basins draining towards the North-west and in the rivers draining the Northern Calcareous Alps units.

82 Chapter 4 0 0 0 0 5 1 2 5 8 24 10 10 16 500 500 500 500 500 amples EA13 EA12 EA15 EA14 400 400 400 EA11 400 400 itu cooling 300 300 300 300 300 are mirrored rams with the 200 200 200 200 200 n=15 n=15 n=15 n=15 n=15 n=49 n=46 n=15 100 100 100 100 100 0 0 0 mineral is indicated. 0 0 h a color code as ex- 8 4 0 5 0 8 4 0 8 0 4 0 2 10 24 16 0 5 10 21 0 4 8 500 12 N 500

EA18

EA19 EA10 400 EA16 400 (1) (3) 300

Raab (1) 300

EA17

Leitha

EA15 200 200 Mur EA16 n=15 n=37 n=15 100 100 0 (1)

EA14 0 0 ) 20 10 5 0 12 10 ( 0 25 10 18 0 2 4 500 (5) 500 EA9

400 4

EA17

400 Mur

EA13 300

Drau 300 200 n=5 n=15 n=30n=30 n=30 n=30 200 100

(6) Enns 100 0 0 0 21 10 8 4 0 (6) 0 (7) 21 ) 5 15 11 500 (

EA10 (6) 500

EA11 EA8 400

EA9 (8) 400 EA18 300 (2)

EA12 (11)Zimmermann et al., (1994) (11)Zimmermann et al., (1995) (12)Neubauer et al., (1995) et al., (13)Eynatten (2009) et al., (14)Wiederkehr 300

200 n=15 Tiroler alzach

number S n=24 0 (4) 200 6 500 100 100

300

0 ) Bt Ms EA8 Ziller 0 KDE 11 Age Ma Age 22 11 0 ( 100 0 6 3 0

0 12 25 4 0 number

500

(4) (6)Liu et al., (2001) (6)Liu et al., (7)Scharf (2013) et al., (2012) et al., (8)Warren (1996) (9)Handy et al., (2003) (10)Challandes et al., (13) Sill

EA7 Key 500 EA7 400 EA19 400

n=14 EA6 300 50-0Ma 400-300Ma 300-200Ma 200-100Ma 100-50Ma

Bt

300 Inn 200 Ms

EA5 200 n=30n=27 n=30 (14) 100 (13) (1)Dallmeyer et al., (1998) (1)Dallmeyer et al., (2008) et al., (2)Kunz (1999) (3)Muller et al., (2004) (4)Ratschebacher et al., (2006) (5)Wlesinger et al., 100 0 (13) 0 20 10 0 0 4 2 0 4 8 EA3

12

12 14 Inn 500 (9)

EA4 EA6

EA1 400

EA2 Milan

10 Rhein 300 8 200 n=50 n=50 (10) 47 46 45 44 100 0 0 0 0 0 2 4 6 0 2 4 6 8 2 0 4 6 4 6 12 4 2 500 500 500 500 500 EA1 EA3 EA2 EA4 EA5 400 400 400 400 400 300 300 300 300 300 Tectonic map and in-situ bed rocks ages of the Eastern Alps. The major river network is indicated by the pale blue paths. The s The major river network is indicated by the pale blue paths. Alps. map and in-situ bed rocks ages of the Eastern Tectonic 200 200 200 200 200 100 100 100 100 100 n=15 n=15 n=30 n=15 n=15 n=15 The numbers associated with the in-s ected on the upper side of box whereas white mica dates are bottom box. n=30 n=30 n=30 n=15 fl 0 0 0 0 0 4 2 0 8 4 0 5 0 6 0 6 4 2 0 10 12 same color code of the bed-rock ages. The age range from 0 to 500 million of years. The histograms and the KDEs for biotite The age range from 0 to 500 million of years. same color code of the bed-rock ages. plained in the legend. On side, boxes display new 40Ar/39Ar plotted as Kernel Density Estimator (KDE) and histog are indicated by the yellow stars. The square represents the muscovite data, circle biotite and they are associated wit are indicated by the yellow stars. and re ages describe the references of work as explained in legend map. In each box, number analysis per target Figure 4.3. Figure

83 Testing the 40Ar/39Ar detrital approach: Eastern Alps 4.5 Inversion of age distributions to estimate relative present-day erosion

4.5.1 A new approach to extract relative present-day erosion estimates from detrital age data In the study area, the detrital record yields a consistent picture of cooling ages in the downstream sediments of the Eastern Alps rivers. To assess the spatial variability of the present-day erosion for each exclusive source draining into the Inn River basin, we use the age distributions of three samples along the Inn river trunk and one tribu- tary draining into it. We have developed a method based on the linear inversion of the raw binned age data points without involving any thermal calculation. Previous methods where a thermal model was used (i.e. Brewer et al., (2006)) tried to compare their data to theoretical density age distributions that rely on thermal model predictions. The use of the “raw” binned detrital ages, allows us to avoid any complication or bias that may arise from assumptions about the past geothermal gradient or rock thermal con- ductivity and heat production. Such assumptions can lead to unnecessary uncertainty in interpreting the data. Due to the simplicity of its basic assumptions, our method is, however, highly limited by the number of samples in the age distribution (whether 10, 30 or 100) as each distribution/sample needs to contain a suffi cient number of analy- ses to provide a robust interpretation that can be applied to an entire catchment area. The method regards the ages as passive markers that defi ne a characteristic signa- ture of a given catchment. The age distributions observed at different sites along the river are the results of mixing of these signatures that depend, among other things, on the mean present-day erosion in each catchment, i.e. with contributing areas to the sections of the river between successive sites. The relative size of these individual catchments also infl uences the mixing process as well as the relative concentration in surface rocks of the mineral used for the linear inversion. Recently, evidence of bias related to so-called mineral fertility or concentration (defi ned as the target mineral abundance in the source rocks) has been explored (Malusà et al., 2016) within the Western Alps. In their work, Malusà et al. (2016) argue that any geological interpreta- tion obtained from a detrital record in modern and ancient settings can be signifi cantly improved when mineral fertility is properly considered. The importance of the estima- tion of mineral abundance in the source rocks (Concentration factor α), to correctly interpret signal dilution over the length of a river convinced us that we should also take into account this factor. Our approach was to give to each site along the river a position number that cor- responds to the relative location of the site in the river network and that is expressed

84 Chapter 4 as a progressively increasing (downstream) numerical value. The position is defi ned by a positive value for samples located in the main river trunk and by a negative value for the samples located in a tributary (second column in Table 4.2). The areas of the exclusive catchment corresponding to each site has been calculated from a DEM and is expressed in square km (third column in Table 4.2). We call α (fourth column in Table 4.2) a semi-quantitative estimate of the relative concentration of the dated min- eral in the surface rocks of the corresponding catchment. To compute it, we make the simplifying assumption that rocks have an average mineral composition that can be estimated from the lithology as indicated on a geological map of the region for each exclusive area (Bigi et al., 1990). To do this, we overlapped the catchment areas on top of the geological map. We then estimated the number of surface units in the area assuming that, i.e. carbonate rocks of 40% of the total area would have produced nil concentration of white mica and biotite, although the 60% of medium-to- high-grade 4 metamorphic rocks would have contributed 60% of the total area. Because the con- certation α is expressed as a relative number (0<αi<1) we derive a concentration of 0.6 of white mica and biotite.

Table 4.2. Input parameters used for the inversion of the detrital age distributionsa. Sample Position Area (km2) α value Mineral EA3 1 1345 0.75 Bt EA4 -2 72 0.7 Bt EA5 3 1571 0.35 Bt EA6 4 2975 0.65 Bt EA3 1 1345 0.7 Ms EA4 -2 72 0.75 Ms EA5 3 1571 0.4 Ms EA6 4 2975 0.7 Ms aFitted parameters used for the inversion of the Inn basins age distributions. See equation (3.1) in chapter 3.

In the Appendix, we give a detailed description of the method we have designed to retrieve this information from the data. The method yields estimates of present-day erosion, as well as estimates of the present-day distribution of ages in surface rocks for each catchment. The method works best when successive age distributions are markedly different from each other, implying that each successive exclusive catch- ment has a distinct signature that can be used to infer the mean erosion characterizing it. We also show how the method needs to be slightly modifi ed to obtain independent

85 Testing the 40Ar/39Ar detrital approach: Eastern Alps relative erosion estimates from detrital age distributions measured in tributaries. One drawback of our simple approach is that it cannot provide estimates of the confi dence we should attach to our estimates of relative present-day erosion and sur- face rock age distributions in each catchment. In order to derive such an estimate, we used a bootstrapping method that relies on computing the relative present-day erosion and surface age distribution estimates from a large number of sub-samples of the original dataset by arbitrarily and randomly removing a set number of data points. Simple statistics are then applied to a large number of erosion estimates and surface rock age distributions.

4.5.2 Application of the mixing model to the Inn river detrital age distributions Mean present-day erosion estimates, their standard deviation and modal values obtained by the inversion of the age distributions as described above are shown in Table 4.3. The observed age distributions and predicted concentrations of surface ages are displayed as normalized histograms in Fig. 4.4a and b (from the biotite and white mica data respectively). Predicted relative modal erosion together with the relative surface age concentration for the muscovite and biotite are summarized in map form in Fig. 4.5. The raw age data are available in the data repository. Looking at the relative present-day erosion distributions obtained by using the bootstrapping technique (Table. 4.3), we can see that the standard deviations are com- monly large, i.e. of the order of 35-55 % of the mean predicted erosion estimates for both the muscovite and the biotite age samples. We note also that, for both systems, the modal value is signifi cantly smaller than the mean value where present-day ero- sion values are highest. In each panel of Figure 4.4, the light gray bars represent, for each site, the observed age distributions in the detrital sample. The dark gray bars represent the distribution among the different age bins of the mean of the predicted relative surface concentra- tions obtained at each site from the bootstrapping procedure. The upper panels are for the biotite ages, while the lower panels are for the muscovite ages. We note that, in general, the observed distributions substantially vary from one station to the next. This implies that the method should yield meaningful estimates of the relative present-day erosions in all catchments. Indeed, the dataset does not suffer from the uncertainty that we describe above and results from the diffi culty in inter- preting two consecutive age distributions that are very similar.

86 Chapter 4 Table 4.3. Results of the Mixing method and bootstrapping approacha. Locations Mean erosion St. deviation Modal value Mineral EA-03 0.957 0 0.672 Bt EA-05 0.561 0.206 1.235 Bt EA-06 1.481 0.825 1.032 Bt EA-04 0.488 0.297 1.231 Bt Locations Mean erosion St. deviation Modal value Mineral EA-03 0.411 0 0.314 Ms EA-05 0.213 0.118 0.471 Ms EA-06 2.376 1.346 1.629 Ms EA-04 0.208 0.125 0.471 Ms We note that the distribution of predicted surface concentrations in the fi rst catch- 4 ment is not exactly equal to the observed distribution in the river sample, despite the method requiring that it be the case. This is because, at each step of the bootstrapping procedure, we generate a sub-sample that is different from the original dataset, there- fore, generating a mean set of surface concentrations that is similar but not identical to the observed, complete dataset. We also note that where the distribution of predicted surface concentrations (dark grey bars) are very similar to the observed distributions (light grey bars), such as for the white mica data at site EA6, it implies that the signal coming from the correspond- ing catchment dominates the signal in the river; in other words, this means that the local contribution strongly overprints the signal coming from upstream. This must mean that the erosion rate in the catchment must be high or that the catchment area is markedly larger than the cumulative upstream catchment area, which in the case of site EA6, is not true. This explains why the method predicts that, according to the white mica data, relative present-day erosion in catchment area EA6 is very large (Figure 4.5). Figure 4.5 contains the same information as Figure 4.4 but displayed in a geo- graphical manner to help interpret the results of the mixing model. For ease of inter- pretation, we have also normalized the estimated media values of erosion rate by their modal. Importantly these remains relative values that should only be interpreted as giving us information about the spatial variability or gradient in relative present-day erosion, not its absolute value. We see that both datasets suggest an increase in rela- tive present-day erosion downstream, but that the biotite data favors the highest rela- tive present-day erosion in the central section (catchment EA5) and the white mica in the lowest catchment (EA6). In view of the substantially larger number of ages in

87 Testing the 40Ar/39Ar detrital approach: Eastern Alps the muscovite dataset, compared to the biotite dataset, the conclusion arising from the white mica data should be preferred. Interestingly, both datasets predict a similar relative present-day erosion in the tributary (EA4) compared to its “parent” catch- ment (EA5), despite the fact that the observed distribution in the tributary is markedly different from that observed for the entire parent catchment. This demonstrates the strength of our method and the fact that its prediction in terms of erosion rate is inde- pendent of the absolute value of the ages in the observed distributions. Finally, we note that the predicted surface concentrations of the youngest ages (0-200 Ma) are, in general, larger in the downstream catchments, while higher rela- tive concentrations of older ages (200-400 Ma) are found in the upstream catchments.

a) EA3 EA5 EA6 EA4 Normalized height (H) Normalized

0.012345 0.2 0.4 0.6 0.8 1.0 12345 12345 12345 b) EA3 EA5 EA6 EA4 Normalized height (H) Normalized

0.012345 0.2 0.4 0.6 0.8 1.0 12345 12345 12345

Figure 4.4. Results of the computation of the age distribution. In the x-axis, each double-box represent one bin range expressed in Ma. From the origin of the x-axis the bins are: 1) 0-50 Ma, 2) 50-100 Ma, 3) 100-200 Ma, 4) 100-300 Ma and 5) 300-500 Ma. In the y-axis is expressed the normalized height of the bins (i.e. the number of grains in the bin range) a) Observed sur- face distributions of ages (light grey) for the samples collected at locations showed in Figure 4.5 (a) and predicted surface age distributions (dark grey) in corresponding catchment areas for the biotite ages and for the muscovite (b).

88 Chapter 4 10o00’ 11o00’ a) 8 10 12 14 47 B 46 EA6 45 Milan 47

00’ 44 o o EA5 00’ 47 EA2

EA4 EA3

A 46 o 00’ 00’ o 0 50 km 46

10o00’ 11o00’

A B 2200 m EA04 2000 m EA03 EA05 EA06 1750 m 1500 m 4 1250 m 1000 m 750 m 500 m 0 km 25 km 50 km 75 km 100 km 125 km 150 km 175 km

b) Bin: 0-50 Ma Bin: 50-100 Ma Bin: 100-200 Mac) Bin: 0-50 Ma Bin: 50-100 Ma Bin: 100-200 Ma

1 1 1 1 1 1 0.5 0.5 0.5 0.5 0.5 0.5 0 0 0 0 0 0

Biotite modal erosion rate Muscovite modal erosion rate

1.5 2 0.75 1 0 0

Bin: 200-300 Ma Bin: 300-500 Ma Bin: 200-300 Ma Bin: 300-500 Ma

1 1 1 1 0.5 0.5 0.5 0.5 0 0 0 0

Figure 4.5. Topographic map, catchments area (light gray) and sample locations (grey dots) used for the inversion (a). The Inn river and the fl owing direction are indicated in by the white arrow and the main trunk river profi le is displayed from A to B. Predicted modal present-day erosions and relative (normalized as that the sum of the 5 bins is 1) concentrations of surface age distributions for the biotite (b) and muscovite (c).

89 Testing the 40Ar/39Ar detrital approach: Eastern Alps 4.5.3 Estimating long-term exhumation rate

4.5.3.1 Classic lag time method The exhumation rate can be roughly estimated using the cooling age and closure depth of closure temperature of mineral from river system:

Elag=ZC/ tc = ((Tc-Ts)/D)/L (4.1)

Where Elag is the exhumation rate, ZC is the closure depth of mineral and to is the time for bringing mineral from closure depth to surface which is approximately equal

to L when the transport time is short in the maintains. Ts (20°C) is surface temperature

and Tc is the closure temperature of mineral. The geothermal gradient of the Alps spans from 20°C/km to 45°C/km based on various methods. We chose the medium value (D = 32.5°C/km) as geothermal gradient for this study. We assume that the clo- sure temperature of muscovite and biotite are 350-425°C and 280-345°C, respectively (Hames and Bowring, 1994; Harrison et al., 2009). Samples EA7 - EA11 were col- lected from rivers originating in the Tauern Window (Fig. 4.6). We chose the young- est age peak center of each sample as the cooling age (L). The detrital muscovite and biotite age distributions of these samples are used to constrain the average exhuma- tion rate of the Tauern Window. Because surface rocks of the Tauern Window form only part of the catchment area of sample EA11 (Fig. 4.6), we infer that the muscovite and biotite grains that form the young age peak of this sample were derived from the Tauern Window, whereas the older muscovite grains originated from other areas (e.g. Austroalpine units). Based on the equation (4.1) and detrital muscovite and biotite ages (Fig. 4.3), the approximate exhumation rate for each samples was calculated. The results are shown in Figure 4.6. EA7 yield highest exhumation rate (Ms: 0.68 - 0.84 mm/yr and Bt: 0.52 - 0.65 mm/yr) among fi ve samples compared with EA11 has lowest exhumation rate (Ms: 0.37 - 0.46 mm/yr and Bt: 0.34 - 0.42mm/yr). In general, the exhumation rates calculated from muscovite are higher than those of biotite in all fi ve samples.

4.5.3.2 Detrital age-elevation method Comparing detrital ages and bedrock ages suggests that almost all young micas (<50 Ma) were derived from the Tauern Window (section 4.3). These young mica ages are therefore likely to mostly provide information about the exhumation rate in the Tauern Window. The white mica and biotite 40Ar/39Ar ages record the time taken for each mineral to be exhumed from the depth of its closure temperature to the sur- face (assuming that the surface transport time, i.e. of the order of thousands of years,

90 Chapter 4 is short compared to the time necessary for exhumation, i.e. of the order of millions of years). Consequently, a cooling age can be used to calculate a fi rst-order approxima- tion of the exhumation rate, E, according to:

E=Rrange/trange=(Zmax-Zmin)/( tmax-tmin) (4.2)

where the Zmax and Zmin are the highest and lowest elevations, and tmax and tmin are the oldest and youngest ages in a given catchment/sample. We make the simplifying assumption that the difference between highest elevation (Zmax) and lowest eleva- tion (Zmin) is the main cause for the difference between the oldest (tmax) and youngest

(tmin) ages observed in a given distribution. The youngest, respectively oldest, detrital muscovite or biotite ages in a river sample are expected to originate from the lowest, respectively highest, elevation. The validity of this assumption can be evaluated by 4 comparing the hypsometric curve computed in the catchment and the observed cool- ing age cumulative curve (Fig. 4.6). If the two curves are similar, our assumption is valid and we can derive from it the fi rst-order estimation of exhumation rate, E. Based on the detrital age and elevation data, we calculated an approximate value for the exhumation rate for every sample catchment (EA7-EA11) over the time range (~5-47 Ma) (Fig. 4.6), by using Equation (4.2). The results are shown in Figure 4.6. They confi rm that the exhumation rate of samples from the western Tauern Window is higher than east in general, which is consistent with the younger muscovite and biotite ages in the western Tauern Window. The exhumation rates based on biotite data of samples EA7 (0.28 mm/yr) and EA8 (0.30 mm/yr) are similar but are lower than those of EA7 (0.52 mm/yr) and EA8 (0.64 mm/yr) calculated from white mica data of same sample for the period that ranges from ~8-25 Ma. However, muscovite and biotite ages of samples (EA9-EA11) from the eastern Tauern Window yield similar exhuma- tion rate. Note that for samples EA09-EA11, that contain surfi cial rocks from an area that drains also Austroalpine units, we only use the young (<50 Ma) ages to calculate the exhumation rates. The discrepancies in exhumation rate between samples as well as between target mineral’s estimates (i.e. EA7 and EA8) are also insignifi cant.

91 Testing the 40Ar/39Ar detrital approach: Eastern Alps 8 12 EA11

46 N Milan 44 Inn River EA9 EA10 EA8 Salzach River EA7 Tauern Window

0 30km

1.0 1.0 1.0 EA7 EA9 0.8 0.8 EA8 0.8

0.4 0.4 0.4 Probobility Probobility 0 0 0 0 0.4 0.8 1.0 0 0.4 0.8 1.0 0 0.4 0.8 1.0 t*, z* t*, z* t*, z* E =0.28mm/yr E =0.30mm/yr E =0.53 - 0.66 mm/yr E =0.17mm/yr E =0.35 - 0.43mm/yr (e)Bt E(l)Bt=0.52 - 0.65 mm/yr (e)Bt (l)Bt (e)Bt (l)Bt E =0.41mm/yr E =0.68 - 0.84 mm/yr (e)Ms (l)Ms E(e)Ms=0.64mm/yr E(l)Ms=0.57 - 0.70 mm/yr E(e)Ms=0.15mm/yr E(l)Ms=0.39 - 0.48mm/yr

1.0 1.0 EA11 Key 0.8 EA10 0.8 Biotite Muscovite 0.4 0.4 Cumulative hypsometric curves

Probobility Sample catchment Probobility Probobility 0 0 Sample location 0 0.4 0.8 1.0 0 0.4 0.8 1.0 t*= Normalized age t*, z* t*, z* z*= Normalized elevation

E =0.19mm/yr E(e): calculated from age - elevation method (e)Bt E(l)Bt=0.36 -0.45 mm/yr E(e)Bt=0.11mm/yr E(l)Bt=0.34 - 0.42mm/yr E =0.15mm/yr E =0.38 - 0.47mm/yr E : calculated from lag time method (e)Ms (l)Ms E(e)Ms=0.12mm/yr E(l)Ms=0.37 - 0.46mm/yr (l)

Figure 4.6. Results of the Exhumation rates calculated from the Tauern Window using the lag-

time (El) and age-elevation profi le (Ee) methods. The shaded area is the Tauern Window. The dash lines represent the sample catchment and white lines represent the current river. 4.6 Discussion

4.6.1 Relative present-day erosion The Inn river originates in the pre-Alpine basement of the Grisons-Bernina units (Austroalpine) and crosses the Cenozoic Penninic nappes of the Engadin Window in the upper reaches of the river (Fig. 4.1 - 4.2). The observed detrital age distributions of this area record mostly pre-Alpine metamorphism and exhumation (compare gray histogram bars in Fig. 4.4). From the inversion of the detrital ages the highest present- day erosion estimates are predicted to occur in the lower part of the main trunk (sam- ple EA06). Interestingly, there is a good correlation between catchments where high concentrations of young surface ages (0-50 Ma and 50-100 Ma) are seen and catch- ments where higher present-day erosions are predicted by the inversion. Similarly, in

92 Chapter 4 the inversion predictions, there is a good correspondence between the predicted lower present-day erosion estimates and the size of older age bins generated by the algo- rithm (compare central panel displaying the predicted relative present-day erosion with the upper/lower panels showing the relative age concentrations of Fig. 4.5 b-c). Note that the relative erosion estimates inferred with the new method are obtained by considering the ages as passive markers. The predicted surface concentrations are also independent of what the ages mean (i.e. whether they are old or young). The fact that younger surface ages are found in the fastest eroding catchment (and vice-versa) must be regarded as an independent confi rmation of the predicted relative present-day erosions and that this gradient in erosion rate has existed for some time. There could be situations where surface ages in the fastest eroding catchments are older than in the slower eroding catchments. This would indicate that the present-day erosion is relatively recent and have not been sustained for suffi ciently long periods of time to 4 reset the ages set by a previous tectonic event to younger values (Braun, 2016). We describe an increase of the predicted values of relative erosion downstream in the Inn trunk river (Fig.4.5). This is associated with the relative change in the age distributions between EA3 and EA6, that shows an increase of younger bin ages (5- 100 Ma) ages in the downstream catchments. This contribution is probably related to inset in the river system of sediments eroded from the younger Penninic units of the Engadin Window. For example, from the inversion of the white mica ages, the large difference in the predicted concentration of ages of the 50-100 Ma bin between EA6 and its upstream catchments (EA4, EA5), only can be explained by one step increase of the erosion of the younger surface rocks units contributing to the downstream. As observed earlier, the new method works well with our dataset because the age distributions are very different from one site along the river to the next and the infor- mation they contain can be used to derive relative present-day erosion, independently of the absolute age values. Because the absolute ages are very old, in part because the closure temperature of the systems being considered is relatively high, they contain relatively little information concerning the present-day erosion rate. In summary, the inversion of the binned detrital-age distributions has allowed us to quantitatively constrain the spatial variability of the relative present-day erosion along the Inn river.

4.6.2 Late-Oligocene-Miocene exhumation rates The rapid exhumation of the Tauern Window commenced from the late Oligocene- early Miocene (Luth and Willingshofer, 2008; Schneider et al. 2013). This is sup- ported by the younger muscovite and biotite ages for samples (EA7 - EA8) from the

93 Testing the 40Ar/39Ar detrital approach: Eastern Alps western Tauern Window when compared with samples (EA9 - EA11) from the eastern Tauern Window (Fig. 4.3). Both samples EA7 -EA8 come from points where the sedi- ment load in the river drains from a relatively small part of the total drainage area cov- ered by the Tauern Window. The classic lag time method and age-elevation method both show that the exhumation rates of EA7 - EA8 from the western Tauern Window are higher than those of EA9 - EA11 from the east, which is consistent with the previ- ous studies based on the thermochronological data (Bertrand et al., 2017; Wölfl er et al., 2012; 2016). This could be caused by a westward increase of shortening, leading to a west ward increasing of vertical thickening and exhumation rate (Bertrand et al., 2017). The exhumation rates calculated from the classic lag-time method are higher than those of the corresponding sample estimated from age-elevation method (Fig, 4.6). Moreover, the exhumation rates are about 0.5-0.9 mm/yr during 20-10 Ma in the western Tauern Window based on various thermochronometric data (Bertrand et al., 2017; Fox et al., 2016 and references therein) and are generally in agreement with those derived from the classic lag-time method. Therefore, the exhumation rates estimated from age – elevation method could be to some extent underestimated due to a limited number of analyzed grains. The exhumation rates calculated from muscovite and biotite using classic lag-time method are generally consistent when compared to those using age-elevation method (Fig. 4.6). The exhumation rates derived from biotite using age-elevation method is signifi cantly lower than those estimated from classic lag-time method and bedrock ages. The hypsometric curve and cooling age cumulative curve of sample EA7 and EA8 are more similar than in any other sample except biotite in EA7 (Fig. 4.6), implying that the exhumation rates obtained for these two samples are the more reliable. The lower exhumation rates estimate for EA10 - EA11 might be due to their larger river catchment which provided older mica to these two samples from Austroalpine units. 4.7 Conclusions

We have demonstrated here that using detrital age data obtained from modern river sands in the Eastern Alps, we were able to derive fi rst-order information on the spatial variability in exhumation rate averaged over large segments of the belt. We also predicted the distribution of in-situ bedrock ages exposed in each drainage basin that was assessed. The age distributions were compared with published in-situ ages and used to refi ne the existing information on the relative contribution of different units to the erosion and consequently to river sediment load. Three main exhumation pulses, classically described within the Alps as Varisican (~350-300 Ma), Pre-Alpine (~100-50 Ma) and Alpine (~50-5 Ma) are clearly recorded in our dataset.

94 Chapter 4 We applied a new numerical approach to the detrital age distributions, in order to get fi rst order magnitude about the relative present-day erosion and the mixing of the different sources in the main river trunk. This relatively simple method has the advan- tage of looking at the raw age distribution as reliable markers of the amount, extent and lateral variability of the erosional processes happening in adjacent catchments. We constrained, using a bootstrapping approach the uncertainties (σ) of our estimates that are generally high. Despite large errors, we can infer the downstream evolution of the present-day erosion in the Inn River. Our model shows that low erosion characterizes the catchments that drain the Austroalpine units (EA3, EA4, EA5) for both target minerals. A two-fold increase of the present-day erosion estimates is recorded in the downstream EA06 sample where a clear input of younger surface ages from the Penninic units of the western Tauern Window and Engadin Window is added to the signal derived from erosion of Aus- 4 troalpine units. Based on the white mica and biotite age distributions derived from the Tau- ern Window, we compared catchment age-elevation relationship to estimate exhumation rates over the ~5-47 Ma interval. We derived the higher estimates of the exhumation rates from smaller-size catchments draining the western sector of the Tauern Window (sample EA8 from the Ziller River). The ex- humation rates calculated from the classic lag-time method and bedrock ages are generally comparable, which is higher than those from the age – elevation method. The lowest exhumation rate estimates (0.11mm/yr and 0.34mm/yr) are found in the eastern parts of the Tauern Window (EA11, Salzach River). Our estimates are consistent with previous studies, and highlight the asymmet- ric cooling evolution of the western and eastern sector of the Tauern Window during Miocene. This work also demonstrates the strength of single grain Ar- dating of detrital mica minerals, to constrain the past and present exhumation history of a given tectonic area. From the analysis of modern detritus, in addi- tion, we derive a reliable picture of the provenance of the contributing sources in the Eastern Alps as previously done using ZFT dating, or 40Ar/39Ar dating in the southern Alps and Himalayas (Bernet et al., 2001; Brewer et al., 2006b; Carrapa et al., 2004). We also demonstrate that modern river catchment-relat- ed thermochronology is suffi ciently precise to characterize the tectonic history of middle to high-grade rocks within an actively deforming mountain orogen. However, further improvements are needed to precisely unravel the dynamic of transport and comminution of the target minerals from the source to the basin and the bias that could result from differences in so-called mineral abun-

95 Testing the 40Ar/39Ar detrital approach: Eastern Alps dance (concentration) within the source rocks. Acknowledgments

Roel van Elsas is acknowledged for his strategic help and support in the mineral separation laboratory at the VU Amsterdam. We want to thank Christel Bontje and Bertram Uunk for their support during total fusion analysis at Ar-Laboratory of the VU, Amsterdam. The results and analysis presented here were supported by the FP7/ People/2012/ITN, grant 316966. This work is dedicated to the memory of our extraor- dinary collegue, Gwladys Govin. Supplementary information

Data Repository for the Eastern Alps can be found at the following URL: https:// doi.org/10.6084/m9.fi gshare.5031410.v1

96 Chapter 4 Chapter 5 Present-day and long-term erosion of the eastern Hima- laya as detected by detrital thermochronology

L. Gemignani, P.A. van der Beek, J. Braun, Y. Najman, M. Bernet, E. Garzanti, J.R Wijbrans

Based on: Gemignani, L., van der Beek, P.A., Braun J., Najman, Y., Bernet, M., Garzanti, E. Wijbrans, J.R. (Submitted), Present- day and long-term erosion of the eastern Himalaya as detected by detrital thermochronology. EPSL. Ab stract Detrital thermochronology allows inferring spatial variability of exhumation rates in an actively evolving landscape. The Namche Barwa syntaxis in the eastern Hima- laya is exhuming extremely rapidly compared to the rest of the Himalaya and this relatively small area provides a signifi cant proportion of the material fl ux drained by the Yarlung-Siang-Brahmaputra River. We present new detrital zircon fi ssion-track (ZFT) and muscovite 40Ar/39Ar (MAr) data from modern sediments in rivers draining the eastern Himalaya. The cooling-age populations for both thermochronometers con- tain a characteristic <2 Ma signature related to the rapid exhumation of the Namche Barwa syntaxis, and a major component ranging at 5-15 Ma. These young ages can be traced in river sediments hundreds of kilometers downstream from their source in the Brahmaputra foreland, with only limited dilution from downstream tributary catch- ments. To estimate present-day erosion in the catchments, we apply a mixing model based on linear inversion of the binned age distributions. The inversion predicts high (but not extreme) erosion estimates in basins draining the Namche Barwa syntaxis and the eastern Himalaya. Signifi cantly slower erosion estimates are predicted in riv- 5 ers draining Tibet and the Lohit plutonic suite, where ages >20 Ma form the major peak in the age distributions. Our data defi ne consistent trends in regional MAr and ZFT cooling ages in the source rocks and can be used to assess sediment provenance and drainage-basin averaged bedrock exhumation in different sectors of the eastern Himalaya. 5.1 Introduction The course of the Indus and Brahmaputra rivers draining the Himalaya, character- ized by two symmetrical bends (Fig. 5.1a), is amongst the most distinctive on Earth and refl ects a complex interaction of crustal deformation and drainage pattern reor- ganization (Bracciali et al., 2015; Brookfi eld, 1998; Clark et al., 2004; Hallet and Molnar, 2001). The spatial link between the sharp bends and major knickpoints in these rivers with the two syntaxial regions of the Himalaya, which represent the most rapidly exhuming parts of the mountain belt, has sparked hypotheses on coupling be- tween tectonics and surface processes (Finnegan et al., 2008; King et al., 2016; Wang et al., 2014; Zeitler et al., 2001; 2014). The eastern Himalayan Namche Barwa syntaxis (NB) is characterized by rapid localized exhumation linked to an active crustal-scale antiform (Booth et al., 2009). The young metamorphic core of the NB has experienced exhumation at rates of ~3 to 10 km/Ma since 4-10 Ma (Bracciali et al., 2016; Zeitler et al., 2014, and references therein). Petrographic, geochemical and thermochronological analyses of detritus in

99 Exhumation patterns from the Eastern Himalaya Brahmaputra River sediments suggest that its sediment fl ux is dominated by substan- tial input (estimated at ~35% to as much as 70%) from the NB, which constitutes less than 5% of the catchment in surface area (Enkelmann et al., 2011; Garzanti et al., 2004; Singh and France-Lanord, 2002; Stewart et al., 2008). Detrital thermochronology can be applied to a range of tectonic and geomorphic problems, including assessment of sedimentary provenance, exhumation history of the source area, and reconstruction of paleo-drainage patterns (e.g. Bernet and Spiegel, 2004; Brewer et al., 2006). Because of the ubiquity and durability against weathering of their host minerals, 40Ar/39Ar thermochronology of white mica (MAr) and fi ssion- track analysis of zircon (ZFT) have been widely applied in detrital thermochronology (e.g., Bernet and Garver, 2005; Von Eynatten et al., 1996; Gemignani et al., 2017). Their nominal closure temperatures of 300-450 °C and 200-360 °C respectively, de- pending on the cooling rate and the integrity of the mineral lattice, allow tracking of exhumation from upper- to mid-crustal depths (e.g., Reiners and Brandon, 2006). Detrital thermochronology can also be used as a tracer of present-day erosion rates. Several methods have been designed to extract quantitative information on the spatial distribution of source-area erosion rates from detrital thermochronology datasets (e.g., Brewer et al., 2006; Ruhl and Hodges, 2005; Stock et al., 2006). However, interpretation of detrital mineral-age distributions depends on understanding how the signal evolves as a function of numerous infl uences, including the erosion-rate patterns, the lateral variation of lithology and abun- dance of the target mineral, the mechanical and chemical breakdown of the crystals during transport, etc. (Brewer et al., 2006; Malusà et al., 2016). In the eastern Himalaya, distal (>1000 km) sedimentary records of the Brahma- putra do not record rapid exhumation of the NB before 1-2 Ma (Bracciali et al., 2016), whereas more proximal deposits show this signature as early as 6 Ma (Govin et al., 2016; Lang et al., 2016). The reason for this discrepancy is not clear: is the earlier signal diluted downstream leading to a bias in the more distal records? The downstream modifi cation of detrital signals as it pertains to grain-size variations is a subject of active research (e.g., Armitage et al., 2011), but the implications of such modifi cation for provenance signals have been less studied (Garzanti et al., 2015; Malusà et al., 2016).

100 Chapter 5 80oE 90oE a Eurasian Plate Indus 30oN b Tsanpo Indian Siang Plate Ganga Brahmaputra

20oN 400 km 0 0 b 90 0 E Lhasa Block 96 E NB Tzangpo Gangdese thrust T1* Siang Jiali Fault zone T2* ITSZ

TSS Mishmi Lohit

Subansiri S6 Dibang 0 0 STD S5 Lohit thrust 28 N D7 GHS L8 LHS S3 Kameng Manas B4 K1 MCT BT M9 M 5 MFT B2 Brahmaputra AATB B5 Indo-Burma Adaman terrane Shillong plateau Legend Q. Alluvium Lhasa Block (Cretaceous- Cenozoic ) Tertiary Siwalik group Indus-Tsangpo Suture Zone TSS Tethys Sedimentary Seq. Major thrust GHS Greater Himalaya Seq. Major detachment

24 0 N LHS Lesser Himalaya Seq. Minor fault Rivers B6 Siang antiform Sb-3 Sand Sample Ganga 0 50 100 Lohit plutonic suite kilometers Gangdese plutons

Figure 5.1. Studied area in the eastern Himalaya. (a) Overview map, showing the major Himalayan rivers; western and eastern syntaxes are indicated by grey stars. (b) Regional geo- logical map (modifi ed after Booth et al., 2009; Chirouze et al., 2013) showing the sample locations, the main tectonic features and the main rivers. Abbreviations: STD - South Tibetan Detachment; MCT - Main Central Thrust; MBT - Main Boundary Thrust; MFT - Main Fron- tal Thrust; ITSZ - Indus-Tsangpo Suture Zone; TSS - Tethyan Sedimentary Sequence; GHS Greater Himalayan Sequence; AATB - Assam Akaram Thrust Belt; NB – Namche Barwa. Here, we address this question by systematically sampling modern river sand from the Brahmaputra catchment downstream of the NB and by analyzing the downstream evolution of the detrital age signal. We utilize the unique thermochronological signal of the syntaxis (Seward and Burg, 2009; Stewart et al., 2008; Enkelmann et al., 2013; Bracciali et al., 2016) to analyze if and how the signal is transformed downstream by

101 Exhumation patterns from the Eastern Himalaya dilution from tributaries. We sampled the main trunk streambed at regular intervals, as well as its main tributaries, and applied both detrital MAr and ZFT analyses to investigate potentially different behavior of these two systems. After initial qualita- tive evaluation of the data, we use a stochastic mixing model in order to: (1) use the relative concentration of observed detrital cooling-age “peaks” to predict the spatial variability of relative present-day erosion estimates; (2) explore the robustness of the detrital signal to quantitavely constrain present-day erosion of the major catchments of the eastern Himalaya. 5.2 Geological Setting The 2900-km-long Yarlung-Tsangpo-Siang-Brahmaputra River fl ows from Ti- bet (where it is named Yarlung-Tsangpo), via India into the Bengal Fan (Fig. 5.1). The Yarlung-Tsangpo is sourced near Mount Kailash in southwestern Tibet and fl ows >1000 km eastwards, along strike of the Himalaya, along the India-Asia suture zone. The river bends sharply at the NB massif and continues to the south through the Tsangpo gorge, forming a >2-km high knickpoint as it crosses the eastern Himalayan syntaxis (Finnegan et al., 2008; Wang et al., 2014; Zeitler et al., 2001). 5.2.1 The Lhasa block and the Himalayan units The Yarlung-Tsangpo suture zone (YTSZ) defi nes the collisional contact between the former Indian and Asian continents (e.g., Yin and Harrison, 2000). North of the YTSZ, the Lhasa block is composed of Paleozoic and Mesozoic meta-sedimentary sequences intruded by Cretaceous-Paleogene calc-alkaline plutonic rocks of the Gangdese (Transhimalayan) batholith (Yin and Harrison, 2000). Deformation of the Gangdese thrust belt (Fig. 5.1b) initiated during Cretaceous subduction of Tethyan lithosphere beneath Tibet (Kapp et al., 2007) and persisted as intracontinental thrust- ing into the Oligocene-Miocene (Yin et al., 1999). The Himalaya developed after Asia-India collision at ~55 Ma (Najman et al., 2010) and are subdivided from north to south into four main tectonic units (Hodges, 2000; Fig. 5.1b): Tethyan Sedimentary Sequence (TSS), Greater Himalayan Sequence (GHS), Lesser Himalayan Sequence (LHS), and sub-Himalaya (Siwalik group). The TSS comprises Paleozoic-Eocene rocks deposited on the Indian plate margin (Gaetani and Garzanti, 1991), and is separated from the underlying GHS by the extensional South Tibetan Detachment system (STDS) (Hodges, 2000). The STDS was active until 13-11 Ma in the central and eastern Himalaya (Leloup et al., 2010; Schultz et al., 2017). The GHS is composed of metamorphic rocks intruded by Miocene leucogran- ite bodies. An inverse regional shear zone, the Main Central Thrust (MCT), separates

102 Chapter 5 the GHS and LHS. Exhumation and metamorphism of the GHS is constrained at ~25- 17 Ma (Hodges, 2000). The LHS consists of meta-sedimentary rocks of Precambrian to Mesozoic age, that are deformed in a complex fold-and-thrust belt (DeCelles et al., 2016). Finally, the Siwalik group consists of Tertiary foreland-basin deposits sepa- rated from the LHS by the Main Boundary Thrust (MBT). 5.2.2 Eastern Syntaxis: Namche Barwa (NB) In the easternmost Himalaya, the suture zone bends a dramatic >90o around the eastern (NB) syntaxis. The NB massif is a NE verging crustal-scale antiform charac- terized by young metamorphism (<10 Ma) and rapid exhumation (~3-5 km/Ma up to 10 km/Ma; Bracciali et al., 2016; Zeitler et al., 2014), the onset of which is contested: around 10 Ma (Zeitler et al., 2014) or 3-4 Ma (Seward and Burg, 2008). The NB syntaxis is enclosed by the Lhasa block and the Transhimalayan plutonic belt. The core of the massif consists of orthogneisses with mafi c, pelitic and rare carbonatic in- tercalations, of GHS affi nity (Burg et al., 1998). Proposed mechanisms to explain the rapid exhumation and erosion rates in the NB syntaxis include: (1) structural buckling 5 at the indenter corner (Burg et al., 1998), further developed as “norward migration of the syntaxis” (Seward and Burg 2008; King et al., 2016) that is consistent with a number of mechanical models (Benidickt and Ehlers, 2014; Rong et al., 2017); (2) vigorous erosional downcutting following capture of the Yarlung-Tsangpo by the Siang-Brahmaputra River, weakening the crust and resulting in a “tectonic aneurysm” (Zeitler et al., 2001; 2014). The Siang window to the south of the NB (Fig 5.1b) exposes Meso-Cenozoic sedi- mentary and basic volcanic rocks (Abor volcanics). East of the syntaxis, the Mishmi crystalline units and Lohit plutonic complex are separated by NW-SE aligned thrusts orthogonal to the Himalayan strike (Misra, 2009). Serpentine lenses in the Mishmi hills have been interpreted as constituting the southeastern prolongation of the suture zone (Singh, 1993). 5.2.3 Thermochronology data from the eastern Himalaya The summary of available in-situ thermochronology data from the eastern Hima- laya is presented in Figure 5.2. Within the Lhasa terrane and Gangdese batholith, mica and feldspar 40Ar/39Ar ages suggest post-emplacement cooling of plutonic rocks be- tween ~20 and 30 Ma (Yin et al., 1999). Apatite fi ssion-track (AFT) and (U-Th)/He (AHe) ages suggest rapid cooling to surface temperatures during the mid-Miocene (~15-20 Ma; Li et al., 2016). Likewise, low-temperature AFT, detrital ZFT and zircon (U-Th)/He (ZHe) thermochronology data from the northern TSS indicate rapid mid-

103 Exhumation patterns from the Eastern Himalaya dle Miocene cooling (Shen et al., 2016), whereas MAr, ZHe and AFT ages from the footwall of the STDS are ~11-16 Ma, implying rapid tectonic exhumation at this time (Carrapa et al., 2016; Schultz et al., 2017). In the eastern syntaxis in-situ thermochronology data show a bulls-eye pattern around the NB massif (Bracciali et al., 2016 and references therein; Fig. 5.2). Both medium- (biotite 40Ar/39Ar, BAr) and low-temperature (ZHe, AFT) systems show ages < 2 Ma, which extend toward the NE in the Parlung River basin (Fig. 5.2; Zeitler et al., 2014). Detrital thermochronology of Siang River sediments is consistent with this rapid exhumation: at Pasighat (location S5 in Figs 5.1-5.2), detrital ZFT ages <2 Ma make up between 45% (Stewart et al., 2008) and 70% (Enkelmann et al., 2011) of the age distribution. Detrital MAr ages at Pasighat contain a minor (~ 15%) <2 Ma age group (Lang et al., 2016). In the Siang window (Fig. 5.1), ZHe ages of 4-14 Ma and AFT ages <2.5 Ma were reported (Salvi et al., 2016). Detrital MAr ages >15 Ma are recorded in sediments of the tributaries draining into the Siang River from the GHS (Lang et al., 2016). Further west in Bhutan, AHe, AFT and ZHe data (Fig. 5.2) show consistent pat- terns, with a northward increase of cooling ages across the LHS followed by a de- crease north of the MCT and a renewed increase toward the STD (Coutand et al., 2014). ZFT ages are ~8-17.5 Ma, whereas AFT and AHe ages are ~2.5-7 Ma. MAr ages across the MCT in eastern Bhutan are ~9-14 Ma and ~1 Ga in the LHS (Stüwe and Foster, 2001). In Arunachal Pradesh, both AFT and ZFT ages from the GHS are younger than in Bhutan, ~1-3 Ma and ~5-9 Ma, respectively whereas they are ~6-13 and ~11-14 Ma in the LHS (Adlakha et al., 2013). These variations are explained by lateral changes in fault kinematics (Adlakha et al., 2013; Coutand et al., 2014; Robert et al., 2011). MAr cooling ages reported from the GHS of Arunachal Pradesh are ~8- 12 Ma (Yin et al., 2010). The youngest AHe, AFT and ZHe ages from the Shillong plateau are ~8-14 Ma, with older ages of up to ~100 Ma for AHe/AFT and ~400 Ma for ZHe (Biswas et al., 2007; Clark and Bilham, 2008). No data are available for the Lohit and Dibang drain- age basins.

104 Chapter 5

Lohit < 2 2-5 5-10 >30 10-15 15-20 20-30 (Ma)

JFZ

E

o Ages

96 Ar L8 39 Mishmi Ar/ Ar Legend 39 40 Dibang D7 Ar/ 40 ITSZ S5 Detrital Samples (This study) Zircon (U-Th)/He Apatite (U-Th)/He Biotite White mica *Detrital Samples Bracciali et al., (2016) Namche Barwa Apatite FT Zircon FT Siang S6 Dating methods B4 S3

100km N B2 50 5 0

Tzangpo

Subansiri

arlung-

Y (Gangdese) K1

Trans Himalaya Trans Brahmaputra

Kameng

GTS B5

T2*

Lesser Himalaya Lesser STD

M9 ITSZ

Manas Greater Himalaya Shillong Plateau T1* B6 Tethyan Himalaya Tethyan

E

o

MCT

90

MBT

Ganga N N o o 24 28 a.

105 Exhumation patterns from the Eastern Himalaya 1-2 t symbols 0-0.5 0.5-1 2-5 5-10 >10 urces of data. Age (Ma) Age ively. Ar ages (b) and zircon (U-Th/He) 39 50 km Ar/ 40

Jiali Fault zone

Thrust Nam-La a

M Rong Chu Thrust Chu Rong

< 2 <2 ? ? ? ? ? < 5 Ma U-Th/He Zircon and ZFT U-Th/He c. . Note that the color code (to right of c) is common to these two panels but Zeitler et al. (2014)

50 km L

Jiali Fault zone ed from fi

Thrust

Nam-La N

< 2 Ma

t

s

ust

r

h

T

u

Chu Thrust Chu Rong Rong < 5 Ma Ar-Biotite Ar-Biotite . (a) (Previous page) Digital elevation model (DEM) of the eastern Himalaya showing major tectonic features, main rivers, and 39 Ar/ ssion-track ages (c), modi fi 40 b. published bedrock thermochronological ages; ages are represented by a color code and techniques differen different from that in (a). The 2-Ma and 5-Ma age contours are shown with a dotted yellow line continuous red line, respect from that in (a). different and Dashed line shows location of insets (b) and (c): zoom on the eastern syntaxis showing in-situ biotite Figure 5.2 Figure (see legend in (a)). In areas with high sample density symbols represent averages of several data points. See Appendix D for so (see legend in (a)). In areas with high sample density symbols represent averages of several data points. See 5.3 Methods

5.3.1 Detrital thermochronology We sampled modern river sediments in the Brahmaputra River downstream of the syntaxis (S6), from its entry point into the Assam plain at Pasighat (S5) to its confl u- ence with the Ganga (B6), nearly 1000 km downstream of the Namche Barwa. We also sampled the main tributaries to the Brahmaputra River (Subansiri S3, Kameng K1, Manas M9, Dibang D7, and Lohit L8) including the Siang (Table 5.1; Figures 5.1 and 5.2). Both detrital MAr and ZFT analysis was applied on all samples where pos- sible. The detailed analytical techniques are reported in Appendix A. 5.3.2 Inversion of age distributions In order to quantitatively assess the spatial variability of erosion between the dif- ferent sampled catchments, we applied a linear inversion of the binned age distribu- tions for each sample, under the basic assumption that the age distribution observed in a downstream sample can be explained by mixing of the age distribution observed 5 in an upstream sample and the fl ux of material from the intervening catchment area. Thus, the relative changes in bin height between two successive sites along the main stream inform us about the relative estimates of the present-day erosion in the in- tervening catchment. Age distributions from tributaries are included to improve the solution locally. The full analytical procedure on which the model is based is fully reported in chapter 3, whereas an application of the model to detrital minerals from eastern Alps river sediments is presented in Gemignani et al. (2017) (Chapter 4 of this thesis). 5.3.3 Concentration factors

The mixing model depends on the relative concentration of the dated minerals (ai) in the different catchments. Mineral concentration elsewhere addressed as “mineral fertility” (Malusá et al., 2016) can be source of bias while interpreting mineral age distributions. Mineral abundance variability can be observed in tectonic units with similar lithology (Malusá et al., 2016) and relationship between bedrock geology and mineral concentration are complex and river sediments can be used as a proxy to estimate relative estimates of α. We have estimated the abundance of zircon and mus- covite in source rocks based on their concentrations in river sands (Supplementary Table A3). Grain sizes of 100-500 μm, similar to those of the dated minerals, were used. Mineral concentrations were obtained by combining quantitative assessment of petrographic modes by point-counting 400 grains per thin section under the micro-

107 Exhumation patterns from the Eastern Himalaya scope, combined with weighing the dense sediment fraction separated with heavy liq- uids (>2.90 g/cm3). We determined 200 transparent grains in heavy-mineral separates using a size window as wide as possible. Hydraulic sorting and weathering effects were assessed to minimally infl uence the inferred concentration factors, as discussed in Appendix B. Table 5.1. Summary of the new and published data used for mapping the spatial variation of erosion rates in the Eastern Himalaya. Sample Latitude Longitude River Method References short label (N) (E) Main river trunk Yarlung- T1 29°21.281’ 90°43.589’ MAr+ZFT Bracciali et al. (2016) Tsangpo Yarlung- T2 29°15.529’ 91°39.909’ MAr+ZFT Bracciali et al. (2016) Tsangpo Yarlung- 302 29°25.989’ 94°31.591’ ZFT Stewart et al. (2008) Tsangpo Yarlung- P 29°36.442’ 94°56.189’ ZFT Enkelmann et al. (2011) Tsangpo Siang A --MAr+ZFT Lang et al. (2016) Siang S6 28°38.208’ 95°01.503’ MAr This study Siang B 28°34.599’ 95°04.212’ MAr Lang et al. (2016) Siang C 28°05.920’ 95°17.630’ MAr Lang et al. (2016) Siang S 28°05.982’ 95°17.631’ ZFT Enkelmann et al. (2011) Siang S5 28°06.140’ 95°17.984’ MAr+ZFT This study Siang Q 29°02.963’ 94°54.610’ ZFT Enkelmann et al. (2011) Siang R 28°34.628’ 95°04.222’ ZFT Enkelmann et al. (2011) Brahmaputra B4 27°26.664’ 94°45.531’ MAr+ZFT This study Brahmaputra B2 26°47.308’ 93°30.356’ MAr+ZFT This study Brahmaputra B5 26°11.965’ 91°46.349’ MAr+ZFT This study Brahmaputra B6 23°53.351’ 89° 41.106’ MAr+ZFT This study Tributaries to the main trunk upstream of the Namche Barwa syntaxis - H 29°36.385’ 94°56.212’ ZFT Enkelmann et al. (2011) Himalayan tributaries Yang Sang Y 28°58.123’ 94°54.583’ MAr Lang et al. (2016) Siyom X 28°13.160’ 94°51.953’ MAr Lang et al. (2016) Yanme Z 28°11.067’ 95°13.411’ MAr Lang et al. (2016) Subansiri S3 26°47.308’ 93°30.356’ MAr+ZFT This study Kameng K1 26°11.931’ 91°46.297’ MAr+ZFT This study Manas M9 26°46.955’ 90°57.441’ MAr This study - I 28°57.700’ 94°51.826’ ZFT Enkelmann et al. (2011) - J 28°54.599’ 94°46.441’ ZFT Enkelmann et al. (2011) Yang Sang K 28°58.700’ 94°54.283’ ZFT Enkelmann et al. (2011) - L 28°20.184’ 94°57.460’ ZFT Enkelmann et al. (2011)

108 Chapter 5 - M 28°13.187’ 94°51.321’ ZFT Enkelmann et al. (2011) Yanme N 28°11.099’ 95°13.211’ ZFT Enkelmann et al. (2011) Lohit / Mishmi hills tributaries Dibang D7 28°09.496’ 95°40.743’ MAr+ZFT This study Lohit L8 27°26.664’ 94°45.531’ MAr This study Abbreviations: MAr: muscovite 40Ar/39Ar; ZFT: zircon fi ssion-track.

5.4 Results Detrital thermochronology data are synthesized in Table 5.2 and Figure 5.3; de- tailed analytical results are provided in Data Repository. We include two samples from the Yarlung-Tsangpo River upstream of the syntaxis in our analysis, MAr and ZFT data for which was originally reported by Bracciali et al. (2016), and a sample from the Kameng River in western Arunachal Pradesh for which ZFT data that was re- ported by Chirouze et al. (2013). Age distributions were plotted as Probability density Plots (PDPs) using the Density Plotter software (Vermeesch et al., 2012). 5.4.1 40Ar/39Ar muscovite (MAr) ages 5 Single-grain MAr ages of samples T1 and T2 (X and Y of Bracciali et al., 2016) present no ages <5 Ma; most ages are between 10 and 15 Ma. Cooling ages <2 Ma are found in the Siang sample (S6); however the major age peaks are at ~11 Ma and ~20 Ma. The next sample downstream (S5) has a similar percentage of ages <2 Ma, with no grains ~2-10 Ma; the large majority of grains is in the range ~20-30 Ma. Eastern tributaries (Dibang, D7; Lohit, L8) show a large majority of ages > 20 Ma; D7 con- tains no ages <20 Ma, while L8 has a small age peak <10 Ma. The Brahmaputra samples (B2, B4, B5, B6) all yield a broad spectrum of ages, with peaks from ~2 Ma to ~40 Ma. 10-12% of the ages are <5 Ma and 5% are <2 Ma, except in the most downstream sample B6. Samples from rivers draining the southern Himalayan fl ank (Subansiri, S3; Kameng, K1; Manas, M9) show major age peaks between ~10-15 Ma, with smaller peaks around 3 Ma (S3, K1) and 6 Ma (K1, M9). 5.4.2 Detrital zircon fi ssion-track (ZFT) ages Both samples T1 and T2 show a major age peak at ~35 Ma, with smaller peaks at 16-18 Ma and 60-70 Ma. No ages <5 Ma occur in these samples. The Siang sample (S5) yields a positively skewed distribution with a majority of <5 Ma ages (55%; 13% <2 Ma). The Dibang sample (D7) shows a broad age peak at 15-25 Ma (~80%); 17% of the ages are >30 Ma. Brahmaputra samples all have a majority of ages <10 Ma, with signifi cant age peaks around 2 Ma (except for B2, where this peak is minor) and 9-12 Ma. 8-23% of the ages are

109 Exhumation patterns from the Eastern Himalaya >20 Ma. The Himalayan tributaries (S3, K1) have large age peaks encompass- ing ~50% of the data at 5-10 Ma and an older peak at ~17 Ma. eak fitting (Stewart and and (Stewart fitting eak r analysis the deconvolved age populations populations age deconvolved the r analysis 38.0 ± 0.0 [7] 41.6 ± 0.1 [19] 47.2 ± 0.1 [22] 16.3 ± 0.2 [30]11.6 ± 0.1 [37] 26.5 ± 1.0 [3] 15.6 ± 0.1 [25] 46.7 ± 1.0 [3] 20.2 ± 0.2 [3] 23.4 ± 0.1 [30] 28.7 ± 0.1 [26] 48.0 ± 0.1 [12] 19.9 ± 0.2 [40]23.4 ± 0.1 [60] 28.3 ± 0.6 [9] 32.0 ± 0.2 [6] 45.7 ± 1.8 [2] 43.7 ± 0.4 [3] 48] 18.6 ± 0.1 [20] 22.4 ± 0.1 [19] 38.0 ± 0.4 [3] 30] 17.0 ± 0.0 [26]36] 24.0 ± 0.0 [26] 27.0 ± 0.0 [17] 42.0 ± 0.2 [7] 29.3 ± 0.0 [4] 46.2 ± 0.1 [3] 18] 11.9 ± 0.1 [27] 19.0 ± 0.1 [38] 27.0 ± 0.0 [10] 28] 11.9 ± 0.0 [44] 15.6 ± 0.1 [20] 21.5 ± 0.2 [3] 13] 10.6 ± 0.0 [43] 16.2 ± 0.2 [25] 24.0 ± 0.1 [12] 21] 9.2 ± 0.2 [27] 13.9 ± 0.1 [36] 27.0 ± 0.2 [5] 2 [45] 67.8 ± 6.3 [17] 77.1 ± 8.9 [14] 1.2 [17] 33.3 ± 1.9 [49] 63.6 ± 5.5 [17] 1.5 [52] 16.8 ± 3.2 [20] 29.4 ± 9.0 [6] 2.3 [47] 25.6 ± 4.8 [35] 56.6 ± 36.3 [4] 1.8 [14] 8.9 ± 1.4 [53] 16.9 ± 2.9 [27] ± 0.8 [28] 12.5 ± 2.5 [15] 23.9 ± 3.8 [31] ± 0.6 [35] 7.9 ± 1.3 [28] 18.6 ± 3.3 [25] 30]39] 8.8 ± 1.4 [44] 9.3 ± 1.4 [47] 22.1 ± 3.8 [26] 34.9 ± 6.1 [14] of the computed age population. For MA population. age computed the of σ otter program (Vermeesch, 2012) and for the ZFT ages Binomfit p Binomfit ages ZFT the for 2012) and (Vermeesch, otter program ses: P is the peak age in Ma ± 1 age in ses: P peak is the is the size of the age population in %. in population age the of size the is Summary of the new detrital MAr and ZFT data, and peak-fitting results. data, and peak-fitting detrital MAr and ZFT of the new Summary S6S5 MAr MArS3 69 97 0.5-45 MAr 0.5-60 64 5 4 2.6-28S5 16 4 0S3 ZFT 6.3 ± 0.2 [34] 11.6 ± 0.2 [15] 0.6 ± 0.1 [4] 103 ZFT 4 0.5-66.4 16.3 ± 0.5 [28] 64 3.8 ± 0.1 [7] 12 2.4-43.5 9.2 ± 0.1 [ 46 0 1.0 ± 0.2 [11] 8 3.1 5.0 ± 0.9 [22] 10.4 ± L8 MAr 43 4.1-71 0 1 7.1 ± 0.3 [6] 21.0 ± 0.0 [26] T1T2 MAr MAr 34 39 6-198 6-183 0 0T1T2 0 0 ZFT ZFT 6.3 ± 0.6 [7] 6.5 ± 0.0 [9] 51 11.8 ± 0.0 [58] 47 9.2 ± 0.1 [25] 12-522 6.5-110 0 0 0 0 18.4 ± 1.4 [24] 8.6 ± 0.7 [17] 36.0 ± 2. 15.8 ± B4B2 MArB5 MAr 80B6 0.5-430 MAr 68 1.3-423 MAr 82 2B4 152 0.3-526 4 1.6-259B2 9 6 ZFTB5 1 6B6 ZFT 100 1.1 ± 0.1 [13] 0.5-46.4 9 10.6 ± 0.1 [ ZFT 74 1.6 ± 0.0 [7] ZFT 4 100 16 0.8-92 7.4 ± 0.1 [ 1.7 ± 0.0 [11] 78 0.2-93.6 11.5 ± 0.0 [ 11.0 ± 0.0 [40] 16.3 ± 0.0 [ 0.6-76.1 5 42 20 9 1.5 ± 0.4 [27] 15 33 4.7 39 1.2 ± 0.4 [7] 1.2 ± 0.2 [ 4.4 ± 2.7 ± 0.4 [ D7 MArK1 41 16-54 MAr 66 0D7 0.6-118 0 2 ZFTK1 24.2 ± 0.1 [11] 96 31.0 ± 0.1 [22] 8 3.5-115.4 ZFT 94 0 2.4 ± 0.3 [10] 2.6-272.6 6.4 ± 0.0 [ 0 0 8.8 ± 2.2 [15] 14 14.1 ± 6.4 ± 0.6 [51] 17.4 ± 1.9 [38] 138.9 ± 25.2 [11] M9 MAr 43 6.1-42 0 0 6.0 ± 0.7 [5] 9.5 ± 0.0 [ Sample Analysis N range Age <2 Ma N <5 Ma N P1 P2 P3 P4 P5 Table 2. Note: N is the number of single-grain analy single-grain of number Note: N is the were calculated with the Mixture Model function of the DensityPl the of Model function Mixture the with calculated were Brandon, 2004) was used; in brackets brackets in used; 2004) was Brandon,

110 Chapter 5 T1 T2 N=32 N=33 NB Tsangpo Siang S6 N=67            

Manas Subansiri Kameng       S5 M9 K1 S3 D7 N=95 N=43 N=52 N=63 N=41 Dibang

               

B5 B2 B4 L8 N=82 N=65 N=80 N=43 Brahmaputra trunk

                       

L10 B6 N=152 Lohit

      LEGEND PDP X0 Sample Number хϯϬDĂ фϮDĂ N=72 ϮϬͲϯϬDĂ ϮͲϱDĂ Number of grains ϭϱͲϮϬDĂ ϱͲϭϬDĂ Probability N 5 ϭϬͲϭϱDĂ Age (Ma)

b) Zircon fission-tracks

T1 T2 N=34 N=43 NB Tsangpo Siang

       

Kameng Subansiri S5 N=103

K1 S3 Db-7 N=94 N=64 N=97 Manas Dibang



              

B5 B2 Br-4 N= 100 N=74 N=100 Brahmaputra trunk

                  Lohit

B6 L10 NN= = 78

      Figure 5.3. Probability density plots (PDP) for the measured MAr (a) and ZFT (b) age dis- tributions. All age axes are scaled between 0 and 50 Ma; probabilities are relative and scaled to 1. Samples are positioned along the schematic river courses. Sample code is indicated in top-right of each plot; N = number of single grains analyzed. The pie charts represent the rela- tive abundances of different age groups (see legend). The PDP are smoother for the ZFT ages compared to the MAr ages due to the lower analytical precision of the single-grain ages.

111 Exhumation patterns from the Eastern Himalaya 5.4.3 Inversion of detrital age distributions For our modeling, we combined our new dataset with additional data from the lit- erature to obtain the most complete view possible of present-day erosion rates within the eastern Himalaya. Additional literature data are summarized in Table 5.1 and the results of the linear inversion are displayed as map view in Figures 5.4 and 5.5. The linear inversion was run separately for the MAr and the ZFT data, using bin ages of <5 Ma, 5-10 Ma, 10-20 Ma, and >20 Ma. Signifi cant bins of ages were calcu- lated using the mixture model function of Density Plotter by Vermeesch et al. (2012) and were compared with bedrocks thermochronology map of fi gure 5.2 for consist- ency. Relative age concentrations in the eroded source rocks were scaled such that the sum of the four age-bin concentrations equals 1 in each catchment. The results of the computed relative erosion estimates (scaled to an average erosion estimate of 1) are presented with the observed and predicted binned age distributions in Figures 5.4 (MAr data) and 5.5 (ZFT data), respectively. The bootstrapping results are shown in Figures 5.6 and 5.7. Supplementary Tables B1 and B2 report the input parameters, mean and modal erosion rates and their standard deviation, for each catchment. Note that (1) because of very high erosion estimates in some catchments, most of the rela- tive erosion rates are <1; (2) all estimates of relative erosion rates have signifi cant uncertainties, on the order of 30-60%, and for some of them 100 % (supplementary table B1). The inversion of the MAr dataset predicts low erosion rates (modal relative rates of ~0.02) upstream of the NB syntaxis (T1 and T2). Downstream of NB, relative ero- sion rates increase 10- to 20-fold in samples A through S5+C, as signifi cant propor- tions of young (<10 Ma) ages are mixed into the populations. Note that the tributaries (X, Y, Z) do not contribute any young ages, which therefore should come from the exclusive drainage area of the trunk stream, possibly explaining the extreme predicted relative erosion rate (modal value 3.1) for sample B. Very low relative erosion rates (<0.0005) are predicted for the Lohit and Dibang tributaries, as their characteristic age distributions, with large proportions of old ages, do not signifi cantly infl uence the trunk-stream distribution. The model predicts intermediate erosion with modal values around 0.2 for the Brahmaputra catchments and their south-fl ank Himalayan tributaries. The Subansiri forms an exception to this, however the computed estimate present an high error on the order of 100 %. Brahmaputra B4 and B2 have similar age distributions, which leads to low predicted relative estimates of erosion with a very large error for this section of the catchment. The model predicts high erosion

112 Chapter 5 estimates (~1.1) for the most downstream catchment (drained by B6) and its tributary the Manas (M9). Inversion of the ZFT data similarly leads to low predicted relative erosion es- timates along the Yarlung-Tsangpo upstream of the syntaxis: samples T1, T2, 302 and P all have modal relative estimates of <0.04 (and for two of them, <0.01). The estimates of erosion increase by an order of magnitude (modal rates >0.1) as the river crosses the NB massif, both for the trunk-stream samples P, Q, and R, and the tribu- tary samples H-K. All these samples are characterized by a high bin of ages <5 Ma. In contrast to what was inferred from the MAr data, tributaries draining the region south of NB (L, M, N), as well as the trunk-stream sample at Pasighat (S5+301+S), do not require high erosion (modal values <0.06). As was the case for the MAr data, none of these tributaries contribute young ages to the Pasighat sample. The Dibang sample (D7) shows a different behavior for the ZFT when compared with the results of the MAr, as relative erosion in the Dibang is predicted to be high (modal value 0.17), because it contributes to the large proportion of >10 Ma ages observed in the downstream trunk sample B4. This large proportion of older ages disappears again in 5 the downstream sample B2, leading the model to predict the highest erosion estimates (3.69) for this part of the Brahmaputra catchment. Interestingly, as for the MAr data, erosion estimates based on the ZFT data in the Subansiri are predicted to be very low (and unconstrained) due to the strong similarity between the ages of samples S3 and B2. Further downstream, erosion estimates for the Brahmaputra catchment and its Kameng tributary are predicted to be high (relative modal estimates of 0.4-0.7) but not as high as predicted for the MAr data.

113 Exhumation patterns from the Eastern Himalaya T1 T2 A Y S6 B X Z S5* 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

T2 T1 A S6 Y X D7 X B Z S5* L8 S3 B4 K1 M9 B2 Modal value >1 B5 erosion 0.5 0.2 0.1 0.05 0.02 0.01 0.005 0.002 0 100 km 0.001 0.0005 B6 0.0002 0.0001 <0.0001

B6 M9 B5 K1 B2 S3 B4 L8 D7 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Figure 5.4. Present-day relative erosion-rate pattern in the eastern Himalaya as predicted by linear inversion of the MAr ages. The map at center shows exclusive catchments included in the model, shaded according to the modal values of relative erosion rate (logarithmic scale). Samples (catchment outlets) are indicated by red dots for the tributaries and orange dots for the main trunk. Above and below the map, histograms of the observed (light gray bars) and predicted (dark gray bars) age distributions are displayed for the 4 age bins used for the inver- sion: <5 Ma, 5-10 Ma, 10-20 Ma and >20 Ma. The bin heights are normalized such that their sum is equal to 1. Plots are ordered following the position of the sample in the river network from left to right in the top row and from right to left in the lower row.

114 Chapter 5 T1 T2 302 P H Q K I J R 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.60.0 0.8 0.2 1.0 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

H T2 P T1 302 I Q J K R L N D7 M S5

S3 B2B4 K1 B2 Modal value >1 B5 erosion 0.5 0.2 0.1 0.05 0.02 0.01 0.005 0 100 km 0.002 0.001 0.0005 B6 0.0002 0.0001 <0.0001 5 B6 B5 K1 B2 S3 B4 D7 S5 N M L 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.60.0 0.8 0.2 1.0 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.60.0 0.8 0.2 1.0 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.00.20.40.60.81.0 0.00.20.40.60.81.0

Figure 5.5. Present-day relative erosion-rate pattern in the eastern Himalaya as predicted by linear inversion of the ZFT ages. See Fig. 5.4 legend for explanation. Note that some tributar- ies (Lohit, Manas) were not sampled for ZFT and are included in the exclusive area of the downstream trunk sample.

115 Exhumation patterns from the Eastern Himalaya Figure 5.6. Probability Density Plots of predicted present-day relative erosion rates obtained by the bootstrapping approach for the MAr age distributions. Note different scales for both the x and y-axes in these plots. Numbers in top-right of each plot give mean, standard deviation and modal values of the relative erosion rates. The plots are ordered following the position of the sample in the drainage network, from upstream (top-left) to downstream (bottom-right). A box around their identifi cation code identifi es the tributary samples.

116 Chapter 5 T1 T2 302 P H 0.03 0.01 0.02 0.03 0.14 0.0 0.01 0.01 0.01 0.07 0.03 0.009 0.01 0.02 0.11 0246810 0 5 10 15 20 0 50 100 150 0 100 200 300 400 500 0 10203040

0.00 0.02 0.04 0.00 0.04 0.08 0.002 0.008 0.00 0.10 0.20 0.1 0.4 0.7

Q K IJR 0.14 0.19 0.23 0.2 0.23 0.07 0.18 0.16 0.18 0.16 0.11 0.13 0.13 0.14 0.13 02468 01234 0246 01234 0246810

0.1 0.4 0.7 0.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0

LMN S5+301+S D7 0.06 0.06 0.06 0.09 0.27 0.05 0.05 0.05 0.05 3.0 0.19 5 0.0009 0.05 0.001 0.06 0.17 0 1020304050 02468101214 010203040 0 1020304050 0.0 1.0 2.0

0.0 0.2 0.4 0.0 0.2 0.4 0.0 0.2 0.4 0.0 0.2 0.4 0.0 1.0 2.0

B4 S3 B2 K1 B5 0.57 0.06 7.18 0.84 0.84 0.33 0.47 5.52 0.65 0.65 0.39 0.001 3.69 0.43 0.43 0246810 0.0 0.4 0.8 1.2 0.00.51.01.5 0.0 0.4 0.8 1.2 0.00 0.05 0.10 0.15 0.0 1.5 3.0 0246 02040 0246 0246 B6 sample id - Mean erosion 1.82 - Standard deviation 1.86 - Modal value 0.73 Density 0.00.40.81.2

0.0 0.1 0.2 0.3 0.4 02468 Erosion rate 02040

Figure 5.7. Probability Density Plots of predicted present-day relative erosion rates obtained by the bootstrapping approach for the ZFT age distributions. See Fig. 5.6 legend for explana- tion.

117 Exhumation patterns from the Eastern Himalaya 5.5 Discussion

5.5.1 Downstream evolution of the detrital age distributions To answer our initial question of how much the characteristic young age signal from the NB syntaxis is preserved or lost downstream, we look at the evolution of the youngest age population (<2 Ma) in the Brahmaputra River samples (Fig. 5.8b, c). All Siang and Brahmaputra samples present both MAr and ZFT ages <2 Ma, which make up 1-6% of the total age population for MAr and 7-20% for ZFT. There is a down- stream decrease in this age population, in particular the most downstream sample B6, which only contains 1% MAr ages <2 Ma, but has 12% ZFT ages this young. The signal of <5 Ma MAr ages is similar with ~15% of the age distribution in samples S6 and B4, and ~12% in the composite sample S5+C (Fig. 5.8), evolving downstream to 6-9% at B2 and B5. In the Brahmaputra, the proportion of <5 Ma ZFT ages is con- stant at 40-50% (except in B2, where it is only 22%). However, Himalayan tributar- ies (except Manas) contribute ~6-9% MAr ages <5 Ma, and ~13% ZFT ages in this range. A similar distribution of detrital MAr ages has been reported for the Narayani River catchment in central Nepal (Brewer et al., 2006; Copeland et al., 2015; Ruhl and Hodges, 2005), while detrital ZFT data from both Nepal and Arunachal Pradesh consistently show a young age peak with a lag time of ~4 Ma (Bernet et al., 2006; Chirouze et al., 2013). In contrast, these catchments do not contribute grains with ages <2 Ma, neither for MAr nor for ZFT (only 2 grains have MAr ages <2 Ma in the Kameng sample). The <2 Ma ages thus appear like the most robust signal of input from the NB syntaxis; it is present throughout the Brahmaputra River as far as >1000 km downstream, with only limited dilution through input from Himalayan tributaries. As the syntaxial age signal endures downstream, with exception of the white mica grains, it should be visible in sedimentary deposits in the foreland. Such ages are encountered in sediments deposited around 5-6 Ma in proximal sites (Siwaliks in the Pasighat area; Govin et al., 2016; Lang et al., 2016) but only in sediments <2 Ma in the more distal Surma Basin in Bangladesh (Bracciali et al., 2016), while Chirouze et al. (2013) did not encounter such ages at all in the Kameng Siwalik section of west- ern Arunachal Pradesh. We suggest that these differences may result from potential drainage reorganization in the foreland downstream of the NB syntaxis, rendering the record from the distal locations less reliable, and by comminution of the target min- erals downstream. The abundant record for the Namche Barwa signal would thus be obtained in either proximal (Pasighat) or very distal (Bengal Fan) records, the latter encompassing a very large source area independent of drainage paths in the foreland.

118 Chapter 5 a) B2 B6 B K1 B4 M9 Y S6 1 A B5 B6 T1 H Q S5*D7 B2 XZ B5 0.1 T2 B4 J K R I S5* S3 K1 Trunk Trib. 0.01 T1 M 302 N Ar-Ar 0.001 T2 ZFT L Modal erosion rates Modal erosion rates D7 L8 0.0001 S3 -500 0 500 1000 1500 NB Pasighat Ganges Distance along the main river trunk (km) b) Ar-Ar 20 Trunk Trib. < 2Ma S6 B < 5Ma B S5* K1 10 B5 S5* B2 A S6 B4 B5 B6 5

proportion of ages (%) A S3 B2 L8 K1 B6 T1 T2 Y X Z D7 B4 S3 M9 0 -500 0 500 1000 1500 NB Pasighat Ganges Distance along the main river trunk (km) c) 100 H ZFT I Trunk Trib. 80 R < 2Ma Q < 5Ma 60 Q R S5* J B4 40 B5 B6 I S5* P B5 proportion of ages (%) 20 B2 K1 B6 B4 S3 M T1T2 302 J K L N D7 K1 0 S3 B2 -500 0 500 1000 1500 NB Pasighat Ganges Figure 5.8. (a) Modal relative erosion rates along the Yarlung-Tsangpo-Siang Brahmaputra river versus distance downstream (measured from where the river enters the Tsangpo Gorge) as predicted for both target minerals (circles: MAr; squares: ZFT). Trunk-stream samples are connected by continuous (MAr) or dashed (ZFT) lines. Tributary samples are plotted individu- ally. Error bars indicate standard deviation around the mean. (b) Proportion of MAr ages <2 Ma (black circles) and <5 Ma (grey squares) downstream. (c) Proportion of ZFT ages <2 Ma (black circles) and <5 Ma (grey squares) downstream. In both (b) and (c), trunk-stream sample are connected by solid lines; tributary samples are plotted individually. Namche Barwa massif (NB), Pasighat (where the river enters the foreland and becomes the Brahmaputra), and junc- tion with Ganges River is indicated on the x-axes of each plot.

119 Exhumation patterns from the Eastern Himalaya 5.5.2 Relative erosion estimates compared to exhumation rates We have mapped the relative estimates of erosion in contiguous catchments through a mixing model and linear inversion of the detrital age distributions in samples from the Brahmaputra River and its tributaries. A qualitative comparison of the predicted relative erosion estimates with the long-term exhumation rates in- ferred from the detrital age distributions shows fi rst-order agreement between the two. Catchments contributing a high proportion of young ages also experience rapid recent erosion (i.e., those along the main Siang River trunk), whereas catchments or sub-catchments with a high proportion of older ages are inferred to be eroding much slower (e.g. the Yarlung-Tsangpo catchments upstream of the syntaxis and the east- ern tributaries). A few exceptions to this pattern may indicate spurious results that require more detailed investigation: high relative erosion estimates are predicted for the most downstream sample (B6) and its tributary the Manas (M9) for both the MAr and ZFT systems, while the proportion of young ages in these samples is smaller than in samples located upstream. This is due to the relative increase in the age bin 10-20 Ma between samples B5 and B6, which could only have been contributed by M9; the large upstream catchment area implies a high sediment fl ux, requiring high erosion in the contributing upstream catchments to infl uence the age distribution downstream. Another example is provided by samples B4, B2 (Brahmaputra) and D7 (Dibang) in the ZFT system: because B4 contains a large proportion of old ages, which the model assumes to have been contributed by the Dibang, high estimates of erosion are pre- dicted in the latter. As this peak disappears again in the downstream sample B2, very high intermediate estimates of erosion are predicted. More quantitatively, the mixing model, despite relatively large errors, shows (1) a 10-20-fold increase in erosion estimates between the Yarlung-Tsangpo and Siang samples, upstream and downstream of the NB, respectively. This value is in agree- ment with previous estimates of erosion derived from the detrital thermochronology of the area (Stewart et al., 2007; Enkelmann et al., 2011); (2) erosion rates in the Siang that are approximately twice those on the southern fl ank of the Himalaya; (3) low erosion relative estimates in the eastern tributaries of the Brahmaputra (except for the Dibang ZFT sample noted above). We can convert the major age populations en- countered in the different samples into a fi rst-order estimate of long-term exhumation rates in the contributing catchments using a simple steady-state 1D thermal model (Reiners and Brandon, 2006; Willett and Brandon, 2013) with standard thermal and diffusional parameters. Most MAr and ZFT ages from the Yarlung-Tsangpo are 10-20 Ma, corresponding to exhumation rates of 0.4-0.8 km/Ma for MAr or 0.3-0.6 km/Ma

120 Chapter 5 for ZFT. However, we know from analyses on in-situ data that exhumation rates in the southern Tibetan plateau were high in the mid-Miocene and dropped signifi cantly after 10 Ma (Carrapa et al., 2014; Shen et al., 2016). In contrast, MAr and ZFT ages <2 Ma in the NB massif require exhumation rates >2 km/Ma. Seward and Burg (2008) estimated exhumation rates at the core of the NB at 3-5 km/Ma, while Zeitler et al. (2014) inferred higher rates of 7-10 km/Ma. On the south fl ank of the Himalaya, MAr and ZFT ages between 5 and 15 Ma imply average exhumation rates of 0.6-1.3 (MAr) or 0.4-1.0 (ZFT) km/Ma, i.e. less than half of the rates inferred for the NB. Note that these are only fi rst-order estimates, as the simple 1D thermal model used here does not take into account any topographic effects on the age distributions, which will be important in rapidly exhuming regions such as the NB (Braun et al., 2006). Overall, our data support high erosion rates in the NB syntaxis, i.e. on the order of 2-4 mm/a rather than ~10 mm/a. This fi nding leads us to assess potential sources of error in our erosion-rate inversion. First, there is a linear relationship between the inferred erosion rates ( and estimated concentrations of the target mineral in the eroded rocks ( cf. Eq. 3.18. chapter 3), which are incompletely known. We have 5 used petrographic and heavy-mineral analyses on river sands to constrain  in our catchments, and the results show standard deviations on the order of 30-60 % and 100% of the mean, both within single sample and between samples (Supplementary Table A3). While this parameter induces signifi cant uncertainty in our model results, it cannot explain order-of-magnitude variations of the type we see here. Likewise, the resolution with which the age data can constrain the erosion rates was assessed by our bootstrapping analysis. Again, the results show errors in our modal value estimate on the order of 30-60% but these remain much smaller than the order-of-magnitude variations in erosion rates that we infer from the data. A remaining issue concerns the representativeness of our data. We use relative peak heights of (arbitrarily chosen) age groups to constrain the model, without assess- ing their representativeness or reproducibility. A rigorous assessment of reproducibil- ity would require resampling at different times and redoing the experiment. We can do something similar by comparing the data from our Pasighat sample (S5) with those published by Stewart et al. (2008), Enkelmann et al. (2011) and Lang et al. (2016) for samples collected at the same location. The ZFT data of Stewart et al. (2008) and Enkelmann et al. (2011) contain a larger proportion of young grains than our sample S5, despite similar analytical procedures. The reason for this difference is not clear; it may refl ect a limit to the reproducibility of these samples, or result from the criteria used to count specifi c zircon grains for fi ssion-track analyses. However, we combined these samples in the mixing model used to extract erosion rates, thereby smoothing

121 Exhumation patterns from the Eastern Himalaya potential biases that may have existed between the datasets. 5.5.3 Sediment fl uxes The proportion of mica and zircon in our samples that we can attribute to a NB source (cf. Section 5.1) seems relatively low, compared to earlier estimates of the contribution of the NB to the sediment fl ux in the Brahmaputra that were in the order of ~35% to 70% (Enkelmann et al., 2011; Garzanti et al., 2004; Singh and France- Lanord, 2002; Stewart et al., 2008). Although this discrepancy could be explained by low abundance of the target minerals in the NB source rocks, we note that (1) our petrographic analysis of river sands does not point to anomalously low concentrations of these minerals in rivers draining the NB (Supplementary Table A3) and (2) some of the earlier estimates were also made on the basis of detrital ZFT data, and thus should have led to comparable estimates. To assess what concentrations we would expect in the Brahmaputra River sands, we performed a simple sediment-fl ux calculation (Chirouze et al., 2015), assuming no signifi cant differences in abundance of mica and zircon in the different contributing units. Based on bedrock thermochronology data from the literature (Fig. 5.2 b, c) that include different thermochronometers, we estimate the areas exposing <2 Ma and <5 Ma ages for the BAr, ZFT and zircon (U-Th)/He (ZHe) systems (Supplementary Table C). By multiplying each of the computed areas (km2) with the assumed erosion rates (mm/a), we predict the expected yearly sediment fl ux (m3/a or, after multiplying by rock density, Mt/a) from these areas (Supplementary Table C). We then compare this sediment fl ux with the total predicted sediment fl ux at the locations of samples S5 and B6, calculated in a similar manner. We take an average erosion rate of 2-4 mm/a for the NB massif, 0.2-0.5 mm/a for the Yarlung-Tsangpo drainage upstream of the NB and 0.5-0.8 mm/a for the total Tsangpo-Siang-Brahmaputra drainage area, includ- ing the southern Himalayan tributaries. We estimate the total sediment fl ux at Pasighat to be 148-370 Mt/a, which compares favorably with the measured value of 210±92 Mt/a (Stewart et al., 2008). From our calculations, we expect 30±20% of the ZFT ages and 18±12% of the MAr ages in the Pasighat sample to be <2 Ma. These numbers are somewhat higher than the proportions we measured, suggesting that we may be miss- ing some of the youngest ages in this sample, but consistent with the combined sam- ples we used in the mixture model. For sample B6, these proportions are expected to decrease to 7±4% for ZFT and 4±2% for MAr, in line with the observed proportions. We note, however, that given the lack of bedrock MAr ages and the relative scarcity of ZFT ages in the NB massif, the inferred source areas are based on the BAr and ZFT+ZHe systems and are therefore somewhat overestimated, because the closure

122 Chapter 5 temperatures of the BAr and ZHe systems are somewhat lower than those of the MAr and ZFT systems, respectively (Reiners and Brandon, 2006). The above estimates of the contribution to the sediment fl ux are therefore upper limits. Overall, we fi nd that these fi rst-order independent estimates of sediment fl uxes are consistent with our mixing-model results, suggesting erosion rates in the NB syntaxis on the order of several, but probably not exceeding 5 mm/a. 5.6 Conclusions Understanding the evolution of the eastern Himalayan syntaxis is key to discrimi- nating different models of coupling between tectonics and erosion. Detrital thermo- chronology plays an important role in expanding this understanding. We have quanti- fi ed the evolution of the detrital thermochronological MAr and ZFT age signatures downstream of the NB, along the Brahmaputra River and its tributaries. Our data show that the characteristic signal of young (<2 Ma) ages originating from the NB massif is preserved up to at least 1000 km downstream. However, the <2 Ma signal seems to be partially diluted either by sediment fl uxes from tributary catchments or by hydraulic sorting of the target minerals in the river’s fl ow. This signal can be used in 5 the sedimentary record as indicating rapid exhumation of the NB. Previous confl icting conclusions as to the onset of this rapid exhumation may result both from the incom- pletely understood drainage evolution of the Himalayan foreland, and from the effect of dilution of the mineral grains downstream. We have used the data to map out relative erosion rates in the different sampled catchments using a mixing model and linear inversion of the age distributions. The target-mineral concentration in the rocks eroded from each catchment was estimated from sand petrography and heavy-mineral analyses. Although our approach is associ- ated with signifi cant errors of the order of ~40-80%, robust predictions of the inver- sion are: (1) a 10-20-fold increase in erosion rates the Yarlung-Tsangpo and Siang samples, upstream and downstream of the NB, respectively; (2) erosion rates in the Siang that are approximately twice those on the southern fl ank of the Himalaya; (3) very low erosion rates in the eastern tributaries of the Brahmaputra. Relative erosion rates from the mixing model are compared with a quantitative estimate of steady-state exhumation rates required to produce the major age groups observed in the differ- ent samples. This comparison suggests erosion rates of ~0.2 mm/a in the catchments drained by the Yarlung-Tsangpo upstream of the syntaxis, 2-4 mm/a within the NB massif, and 0.5-1 mm/a on the southern Himalayan fl ank. Overall, our data suggest very high but not extreme erosion/exhumation rates at Namche Barwa.

123 Exhumation patterns from the Eastern Himalaya Acknowledgements This work was supported by FP7/People/2012/ITN, grant agreement no 316966. Natalie Vögeli, Gwladys Govin and Katu Bage are acknowledged for support during fi eldwork. Klaudia Kuiper helped in the treatment of MAr data. Roel van Elsas is ac- knowledged for his support during mineral separation and Onno Postma for his help in the 40Ar/39Ar laboratory. Mélanie Balvay helped with ZFT analyses. Data Repository Data repository tables can be downloaded from the following link address https:// fi gshare.com/s/a1468579d841f07f3a6f .

Supplementary information

Appendix A. Analytical procedures for muscovite 40Ar/39Ar and zircon fi ssion- track analyses.

A1 40Ar/39Ar-dating on detrital muscovite Individual muscovite grains ranging in size from 125 μm to 1000 μm were separated using standard mineral separation procedures at the Vrije Univer- siteit, Amsterdam. Subsequently, an average of 250 grains/sample were hand- picked under a binocular microscope (Olympus SZ series), wrapped in Al-foil and loaded in 9 mm ID quartz tubes together with the fl ux-monitor standards FCT (Fish Canyon Tuff; 28.20±0.05 Ma) and DRA (Drachenfels; 25.43±0.03 Ma) sanidine (Schneider et al, 2009). Samples were irradiated at the Oregon State University TRIGA reactor in the (cadmium-lined) CICLIT facility for 12 h (batch VU-101). After irradiation, single grains were loaded in a copper tray and pre-heated in a separate pre-bake vacuum line at a temperature of 250 °C to remove undesirable atmospheric argon. Analytical experiments were car- ried out using both a Helix MS (ThermoFisher) and a single-collector AGES quadrupole mass spectrometer at the argon geochronology laboratory of the Vrije Universiteit Amsterdam. Both mass spectrometers have been equipped with a Synrad 48-series laser. A blank analysis was performed after every third grain analyzed. Data reduction for individual isotopes, blank corrections, air corrections and mass discrimination corrections was performed with ArArCalc (v2.20c) software (Koppers, 2002). Correction factors for the interferences 39 37 36 37 of Ca and K isotopes are: ( Ar/ Ar)Ca = 0.00698, ( Ar/ Ar)Ca= 0.000268, 40 39 ( Ar/ Ar)Ca = 0.00465 (Wijbrans et al., 1995). Analytical results for detrital muscovite 40Ar/39Ar dating are provided in Table A1.

124 Chapter 5 A2 Detrital zircon fi ssion-track dating Detrital zircons from the same samples were separated at the Vrije Universiteit Amsterdam using standard heavy-liquid and magnetic separation techniques. We ob- tained a suffi cient number of grains (in the size fraction 125-250 μm) for seven sam- ples, which were further processed and analyzed at Université Grenoble-Alpes. Zir- con aliquots (200-1000 grains) were mounted in PFA Tefl on, polished and etched in a NaOH-KOH melt at 225-230 °C between 50-70 hours. Two mounts per sample were prepared and etching was done in a stepwise manner to establish a reasonable num- ber of fi ssion tracks for the majority of the grains, focusing on younger grains in one mount and older ones in the other (Bernet and Garver, 2005). The samples were cov- ered with muscovite sheets as external detectors and sent for irradiation to the FRM II Research Reactor at the Technische Universität München, Germany, together with FCT (Fish Canyon Tuff) and BLK (Buluk) standards. The dosimeter glass employed was IRMM541. Standards and samples were counted at 1250´ magnifi cation on an Olympus BH2 microscope, using the FTStage 4.04 program. We aimed at obtaining at least 100 single-grain ages per sample where possible. Analytical results for detrital 5 zircon fi ssion-track dating are provided in Table A2. Appendix B. Assessment of hydraulic sorting and weathering effects on inferred concentration factors. Bedrock abundance is assessed only if sand composition refl ects bedrock compo- sition: i.e. if sediments can be assumed to have been produced by mechanical grinding, in the absence of both hydraulic sorting and selective physical and chemical removal of less durable minerals during erosion, transport and deposition. Mechanical break- down of sand-sized grains has been demonstrated to be low during transport in river water, even over distances of thousands of km in very high-energy settings (Garzanti et al., 2015 and references therein). Rapid physical erosion characterizes most of the Siang-Brahmaputra catchment (Garzanti et al., 2004). As far as hydraulic processes are concerned, the effects of selective entrainment can be detected by calculating the grain density of a sand sample from the density of each detrital component weighted by its abundance (SRD index of Garzanti and Andò, 2007), compared with the as- sumed weighted average density of source rocks, which in an orogenic belt lies at ~2710 kg/m3. To minimize the error caused by hydraulic effects and to obtain robust estimates of mineral abundances, we analyzed several replicate samples for each river tract. Samples falling outside the range 2710±3 kg/m3 were discarded.

125 Exhumation patterns from the Eastern Himalaya Appendix D. Bedrock thermochronology data from the Eastern Himalaya. The data plotted in Figure 5.2 of the main text was compiled from the following references: Adams et al. (2015), Adlakha et al. (2013), Biswas et al. (2007), Clark et al. (2004), Copeland et al. (1995), Coutand et al. (2014), Finnegan et al. (2008), Gru- jic et al. (2006), Li et al. (2015, 2016), Salvi et al. (2016), Stüwe and Foster (2001), Yin et al. (1999, 2010), Zeitler et al. (2014). Supplementary Tables Table A3. Mineral concentration in modern sand samples of the eastern Himalayan rivers used

to estimate the concentration index (αi) for the different catchments.

River Site #samples #classes1 Ms SD2 Zr SD Tsangpo Quxu 1 1 0,00 0,02 Tsangpo Shannan 6 1 0,29 0,30 0,08 0,75 Siang Yingkiong 1 8 1,35 0,01 Siang Pashigat 1 1 1,23 0,56 Brahmaputra Dibrugarh 1 1 1,55 0,09 Brahmaputra Tezpur 2 1 1,89 1,24 0,03 0,03 Brahmaputra Guwahati 2 1 2,38 2,24 0,12 0,09 Yamuna Brahmaputra 8311,38 0,21 0,03 0,04 Bridge Manas at MFT 2 1 2,85 1,65 0,01 0,01 Subansiri at MBT 3 1 0,53 0,31 0,00 0,03 Dibang at MBT 1 1 1,08 0,16 Lohit at MBT 1 1 0,74 0,00 Kameng at MBT 2 1 2,08 1,22 0,03 0,01

1Classes refers to the number of grain-size fraction as referred in Garzanti et al. (2010). 2Stand- ard deviation. Table B1. Sample positions in the river network, contributing exclusive catchment areas (Ai)

and mineral concentrations (αi) of the MAr age distributions used for the numerical inversions, together with predicted mean relative erosion rates, standard deviations and modal value of the distribution of relative erosion rates from the bootstrapping analysis. Values are normalized such that the mean is 1. Sample locations are numbered from upstream to downstream along the Tsangpo-Siang-Brahmaputra River system; negative numbers indicate tributaries.

Mean Sample Position Area (km2) α value SD Modal value erosion T1 1 55395 0.0100 0.0215 0.0000 0.0215 T2 2 13265 0.0300 0.0213 0.0153 0.0149 A 3 41374 0.0014 0.2613 0.1026 0.2035 Y -4 1200 0.0130 0.5551 0.2688 0.4252

126 Chapter 5 S6 5 4526 0.0135 0.5551 0.2688 0.4252 B 6 292 0.0135 5.2742 3.5981 3.1428 X -7 2000 0.0140 0.5341 0.3203 0.3442 Z -8 2500 0.0130 0.5353 0.3193 0.3447 S5+C 9 8775 0.0123 0.5355 0.3190 0.3447 D7 -10 11210 0.0108 0.0080 0.0572 0.0004 L8 -11 20315 0.0073 0.0076 0.0534 0.0003 B4 12 16435 0.0155 0.4336 0.2529 0.2917 S3 -13 26135 0.0053 0.3175 0.3786 0.0182 B2 14 28600 0.0189 0.3976 0.3396 0.2169 K1 -15 9000 0.0208 0.3205 0.2555 0.1737 B5 16 46262 0.0238 0.3212 0.2550 0.1712 M9 -17 28560 0.0285 2.1760 1.7564 1.1223 B6 18 97660 0.0138 2.1784 1.7546 1.1033

Table B2. Sample positions in the river network, contributing exclusive catchment areas (Ai) and mineral concentrations (α ) of the ZFT age distributions used for the numerical inversions, i 5 together with predicted mean relative erosion rates, standard deviations and modal value of the distribution of relative erosion rates from the bootstrapping analysis. See Table B1 legend for explanation.

Area Mean Sample Site (km2) α value erosion SD Modal value T1 1 55395 0.0002 0.0398 0.0000 0.0397 T2 2 13265 0.0008 0.0173 0.0117 0.0094 302 3 25423 0.0042 0.0025 0.0012 0.0018 P 4 8973 0.0040 0.0316 0.0176 0.0220 H -5 390 0.0005 0.1458 0.0727 0.1122 Q 6 8184 0.0056 0.1458 0.0727 0.1122 K -7 3700 0.0015 0.1944 0.1824 0.1380 I -8 1900 0.0040 0.2350 0.1696 0.1334 J -9 450 0.0020 0.2074 0.1837 0.1430 R 10 4818 0.0056 0.2350 0.1696 0.1334 L -11 450 0.0002 0.0623 0.0595 0.0009 M -12 11450 0.0003 0.0695 0.0596 0.0585 N -13 2490 0.0014 0.0613 0.0587 0.0013 S5+301+C* 14 8775 0.0056 0.0943 0.0599 0.0616 D7 -15 11210 0.0016 0.2763 0.1912 0.1703 B4 16 16435 0.0009 0.5734 0.3332 0.3979 S3 -17 26135 0.0001 0.0670 0.4722 0.0017 B2 18 28600 0.0003 7.1855 5.5221 3.6911 K1 -19 9000 0.0003 0.8499 0.6576 0.4303 B5 20 46262 0.0012 0.8499 0.6576 0.4303 B6 21 97660 0.0003 1.8250 1.8638 0.7334

Table C. (Top) Sediment fl ux (Sf) out of the syntaxial region characterized by young (<2 Ma and <5 Ma) MAr and ZFT ages, calculated by multiplying the area enclosed by the 2-Ma and 5-Ma age contours (Fig. 5.2b, c) with average erosion rate (Ea; taken as 4±1 mm/a). Multiply- ing by an average rock density of 2850 kg m-3 gives fl ux in Mt/a. (Bottom) Total estimated

average sediment fl ux (Sft) for the Tsangpo-Siang-Brahmaputra catchment at Pasighat (loca- tion of sample S5) and at the location of the most downstream Brahmaputra sample (B6) for an average basin-wide erosion rate of 0.35±0.15 mm/a and 0.65±0.15 mm/a, respectively.

2 2 -1 3 -1 -1 Contours thermochronometer Area (km ) Area (m ) Ea (m yr )Sf (myr ) Sf (Mt yr ) < 5 Ma U-Th/He & FT Zircon 10084 10,084,000,000 0.003 30,252,000 86 < 2 Ma U-Th/He & FT Zircon 6649 6,649,000,000 0.004 26,596,000 76 < 5 Ma Ar-Ar Biotite 6889 6,889,000,000 0.003 20,667,000 59 < 2 Ma Ar-Ar Biotite 3831 3,831,000,000 0.004 15,324,000 44 2 2 -1 3 -1 -1 Basin Area (km ) Area (m ) Ea (m yr )Sft (myr ) Sft (Mt yr ) Tsangpo-Siang basin at Pasighat 260000 260,000,000,000 0.0005 130,000,000 371 Tsangpo-Siang basin at Pasighat 260000 260,000,000,000 0.0002 52,000,000 148 Brahmaputra basin at B6 544177 544,177,000,000 0.0005 272,088,500 775 Brahmaputra basin at B6 544177 544,177,000,000 0.0008 435,341,600 1,241

128 Chapter 5 Chapter 6

Improving the precision of mica Ar- dating on smaller and younger mus- covite grains: implication for prove- nance studies.

L. Gemignani J. R. Wijbrans K. Kuiper

Based on: Gemignani, L., Kuiper, K., Santato A., Wijbrans, (in preparation for Chemical Geology). Improving the precision of mica Ar-dating on smaller and younger muscovite grains: im- plication for provenance studies. Abstract

Current generation multi-collector mass spectrometers allow for increasingly precise measurement of small ion beams. The improvement of instrument sensitiv- ity and resolution compared with older generation mass spectrometers has impor- tant implications for 40Ar/39Ar dating and allows to expand its range of applicability. Thermochronological analysis of detrital modern river sands is a powerful proxy for unraveling provenance and exhumation histories of eroding hinterlands. Improving the precision of dates for young and small grains contributes to advance an interpreta- tion of the detrital signal in tectonically active mountain ranges. Previous studies used the 40Ar/39Ar method to speculate on how the detrital signals can evolve downstream in the river trunk. So far, however, there has not been a robust assessment of how grain-size variability can introduce biases in the analysis of age distributions. The Plio-Pleistocene evolution of the Eastern Himalaya has been recently studied using detrital thermochronology. The white mica signal from the Namche Barwa syntaxis seems to be diluted downstream its source. Major questions concern how the dilution of this target mineral is related to the analysis of medium-sized grains of muscovite: can the analysis of smaller grain sizes shed light on the dilution effect? Due to its re- cent uplift and therefore young thermochronological signal this area is ideal to answer our research question. 6 Here we use the latest generation mass spectrometers to (1) test if the precision in the analysis of young and small muscovite samples is improved by use of this new instrumentation and (2) test the variability of the age distribution as a function of the grain size from fi ve modern rivers samples draining the Eastern Himalaya. The Helix MC plus at VUA is equipped with 1013 Ohm amplifi ers on the H1-H2 Faraday cups. We compare the functioning of the 1013 Ohm amplifi ers with the 1012 amplifi ers on the in-house Drachenfeld (DRA) standard. The use of the 1013 Ohm amplifi er improved our analysis on standards by a factor of two. We show that for larger catchment areas the use of multi grain-size analysis leads to a more precise assessment of the full spec- trum of ages from the signal drained from different sources. The analysis of smaller grain sizes (< 250 microns) shows how the previous speculations about the dilution of the Namche Barwa syntaxis signal for the muscovite minerals were biased by the grain-size of the analyzed samples. This outcome has important implications for fu- ture provenance studies.

131 Improving the precision of Ar-dating: implication for provenance 6.1 Rationale

The 40Ar/39Ar method is one of the most precise and widely applied dating meth- ods for geochronology. However, due to the low contents of radiogenic argon in young and small potassium minerals, improving the precision of dates of these grains has been the challenge of the last decades. In detrital 40Ar/39Ar thermochronology stud- ies, muscovite is one of the most studied minerals due to its high potassium content, 40 40 low tendency to incorporate excess Ar ( Arexcess), and its widespread occurrence in granites and in low- to medium-grade, regionally metamorphosed terranes (McDou- gall and Harrison, 1999). In 40Ar/39Ar studies, Neogene muscovite grains with sizes of 500-1000 micron could yield, with proper attention to experiment design, ages with analytical precisions of 0.2 % or better (Hodges et al., 2005). However, due to uncertainties of the age of standards, the absolute accuracy of the technique has been increased by using the inter-calibration technique from ~2.5% (Hodges et al., 2005) to 0.25 % (Kuiper et al., 2008). Young (Pliocene-Pleistocene) samples with grain sizes below ca 400 micrometers yield ages with much poorer precision, generally > 30% (Hodges et al., 2005). The recent improvements in the sensitivity of the latest generation mass-spectrometers provide the opportunity to extend the applicability of the technique to precisely measure even smaller beam intensities to get more precise dating of young (Plio-Pleistocene) grains smaller than 400-300 micrometer. Achiev- ing higher precision on smaller and young detrital grains has important implications for the understanding, for example, the dilution downstream in rivers of the detrital signal in provenance studies progressively further away from the source area. Study of progressive dilution can help to improve our understanding of provenance signals obtained for sediment sources located > 500 km away from the source rocks. Previous studies focused on studying the evolution of the detrital signal using 40Ar/39Ar single-grain dating. by looking at tributaries catchments competing in the main river trunk (Rhull and Hodges, 2005; Brewer et al. 2003; 2006; Gemignani et al., 2017). Gemignani et al. (2017) used the detrital age distributions as a tool to infer the exhumation history of the eroded sources using standard grain size minerals (250-500 and 500-1000 microns). Because erosional processes and river transport play a critical role in the comminution of detrital minerals, the analysis of smaller grain sizes is particularly useful in regional-scale provenance studies. Sediments (i.e. river’s sediments, ash layers, foreland sedimentary sequences etc.) during transport from source to sink are transformed by both mechanical and chemical processes. Most of these processes reduce the size of the crystals as a function of the distance from the source rock and the processes in the river’s fl ow. Nevertheless, the downstream fi n-

132 Chapter 6 ing of sediments in the riverbed is regulated by complex dynamic processes, such as abrasion (mechanical breakdown) and hydraulic sorting. Hydraulic sorting seems to act as the dominant process instead of mechanical breakdown for the process of com- minution of muscovite grains (Surian, 2002; Garzanti et al., 2015). High-precision dating of smaller and young grains, that until now was not possible, can be used to test for differences in grain-age distributions between grain size fractions and thus shed new light on the age distributions coming from eroding hinterlands. Removing the fi ner fractions (<400) from the analysis of samples collected from large and complex rivers could cause a bias in age distributions when not accounted for. Depending on the original grain size distributions in the source rock, the larger crystals found in the river near the source of sediment would possibly record the most reliable age-peak distributions, whereas moving downstream along the river we would need to decrease grain size in order to record the full spectrum of all the different sources draining into the river from adjacent and more distal sources. 6.1.2 Improvements in instrumentation High-precision analysis of detrital modern river samples is routinely performed in the Argon Geochronology Laboratory at the Vrije Universiteit Amsterdam We use a Hiden quadrupole for large, older mica grains and a Helix MC Plus Noble Gas MS produced by ThermoFisher Scientifi c for younger and/or smaller grains. The 6 latter system is equipped with a high-resolution 120o defl ection magnetic sector ana- lyzer and with an adjustable multi-collector array, where fi ve dual-detectors (Faraday – SEM) allow measurement of fi ve ion beams simultaneously (see Monster, 2016 for a review). The detector array of the Helix MC plus includes a fi xed axial (AX) detector, two adjustable high mass channels (H1 and H2), and two adjustable low mass (L1 and L2) channels. Each channel has both a Faraday and CDD (Compact Discrete Dynode) ion counting multiplier to allow a maximum range of beam intensities. To improve the precision of analysis of small beam sizes, the amplifi ers connected to the H1 and H2 Faraday detectors have been equipped with 1013 Ohm resistors instead of the original 1011 and 1012 Ohm resistors. A series of experiments on mica crystals and on sanidine mineral standards were performed to test the sensitivity of the mass spectrometer (Monster et al., 2016). Pre- vious work (mainly thermal ionization mass spectrometry) compared the performance of 1011, 1012 and 1013 Ohm amplifi ers, and they established that the new 1013 Ohm resistor increased the instrument sensitivity of a factor of 10 and 100 while the noise increased by 3 and 10 respectively when compared with the analysis made with 1011 and 1012 Ohm. The signal to noise ratio is thus improved which results in improved

133 Improving the precision of Ar-dating: implication for provenance analytical precision for the given instrumental sensitivity (Santato et al., 2015; Zhang et al., 2016, Koornneef et al., 2014; 2015). The 1013 Ohm resistor has been success- fully used in thermal ionization mass spectrometry (TIMS). As example, in Koornneef et al., 2015 and Timmerman et al., 2017 the isotopic ratio of small amounts of Sr (2 ng) and Nd (30 pg) were determined in primitive melts inclusions trapped in olivine phenocrysts from samples of lavas from the Italian Magmatic Provence . However, the new amplifi er hasnot been tested for argon noble gas analyses. Recent studies tested the performance of the Helix MC plus showing its potential high mass resolution (>1800) and high mass resolving power (>8000) and to compare different amplifi ers (1011, 1012 and 1013 Ohm) (Zhang et al. 2016). Performance stud- ies were carried out also on the new Argus VI mass spectrometer. The Argus VI mass spectrometer is a 13 cm radius 90° extended geometry magnetic sector that operates in a static mode that presents higher sensitivity but lower resolution compared to the Helix MC plus (Mark et al., 2009; Phillips and Matchan, 2013). Tests for FCT sanidine standards using the Argus VI mass spectrometer, show reliable single grains ages of ~28.4 ± 0.5 Ma using the 1012 Ohm amplifi er setting within the recommended values (28.294 ± 0.036 Ma)(Kim and Su-In Jeon, 2015). So far, no experiment was devised to test the precision of the single-grain dating of small and young muscovite minerals using the latest generation 1013 Ohm feedback resistors in combination with a Helix MC plus mass spectrometer. In this study, we show new results obtained from the analysis of detrital muscovite modern river samples using the Helix MC plus instrument using the new 1013 Ohm resistors on the H1 and H2 Faraday cups and CDDs for the remainder of the collector positions. Our results on detrital white mica from modern river sediments shed light on how important it is to include smaller grain sizes in order to get a full spectrum of detrital age signals. This study shows the potential of using high resistor amplifi ers for provenance analysis. In the fi rst part of this manuscript, we have tested improve- ment in precision of the two amplifi ers (1012 and 1013 Ohm) using a set of in-house Drachenfels sanidine standards. This standard, when analyzed in single grain single fusion mode, yields a normal age distribution. Overall, we observe an improvement of a factor of ~2 for the standard deviation (SD) from analysis obtained from a grain of size of 355-1000 microns with the 1012 Ohm amplifi er if compared with the analysis using the 1013 Ohm amplifi er. In the second part of the paper, we present ~400 new single grain analysis of detrital mica from fi ve modern river sand samples from the Eastern Himalayan syn- taxis. The river samples are located at increasing distances into the foreland basin (Brahmaputra river Figure 1 and Table 1). The core of the Eastern Himalayan syn-

134 Chapter 6 taxis (Namche Barwa) is one of the most tectonically active places on Earth, and it is characterized by both in-situ and detrital cooling ages ranging from ~4 Ma to ~0.1 Ma (Seward and Burg, 2009; Zeitler et al., 2001; 2014) and extremely high erosion rates (4-10 mm/yr, see Bracciali et al., 2016 for a review). Present-day sediment fl uxes and detrital thermochronology from the Namche Barwa syntaxis and Siang River, indicate that the ~45 % of the total Brahmaputra sediment transport is sourced by input from erosion in the Namche Barwa syntaxis area (Singh and France-Lanord, 2002; Enkelmann et al., 2011). However, the young thermochronometric signal, as recorded using 40Ar/39Ar on muscovite, seems to be diluted progressively downstream in the Brahmaputra foreland, i.e. only 1 % of the transported sediment seems to origi- nate from Namche Barwa syntaxis in the most downstream sample (Gemignani et al., paper in review). The origin of this dilution it is not clear and potentially can be caused by not sampling the smaller grains. Therefore, testing different fractions of mica grains as a function of their age distributions can give additional insight into the evolution of the detrital records downstream the Namche Barwa syntaxis source in the Brahmaputra river. The extremely young and locally well-constrained thermo- chronological signal makes the Namche Barwa a suitable study area to test wheather different fractions will have different age distribution. With the present study, we aim 1) to quantify the precision of analysis that we can 6 obtain using the new 1013 Ohm resistors when compared with the 1012 Ohm resistor from the analysis of the in-house sanidine standards. 2) To investigate to which extent grain-size variability can bias the interpretation of sample age distributions towards older age peak, and how the progress of the 40Ar/39Ar dating technique of smaller and younger grain can minimize this bias. We reach this aim by looking at the downstream evolution of the syntaxis (Namche Barwa) signal in the Brahmaputra river foreland.

135 Improving the precision of Ar-dating: implication for provenance Figure 6.1. Studied area and sample locations. The dotted grey line shows the Tsangpo-Siang- Brahmaputra river’s fl ow direction.

6.2 Methology

River samples were collected from the bar of the active channel of the Siang- Brahmaputra river. The location of the samples is summarized in table 6.1. The sam- pling sites were located at least 1km away from tributary junctions to the main step of the river and any landslide to avoid bias toward one particular source in the main river stream. Approximately 2kg medium grained sand was collected from the top 10 cm sediment at each sampling location from the edge of the active channel. Muscovite was separated for radio-isotopic 40Ar/39Ar analysis. Mineral separation was performed using the standard procedure in the mineral separation laboratory at the Vrije Universiteit of Amsterdam.

136 Chapter 6 6.2.1 Mass spectrometry using the Helix MC plus The Thermo Fisher Scientifi c Helix MC plus is a 350 mm radius high-resolution multi-collector mass spectrometer that allows precise and simultaneous measure- ments of up to 5 noble gas isotopes (Honda et al., 2015). Depending on experiment type, either the Faraday cups or the CDDs can be selected for each collector set-up. In practice 38Ar, 37Ar and 36Ar are measured on the CDDs as these ion beams are very small, whereas both 40Ar and 39Ar are measured on Faraday cup or CDD depending on signal size. The Faraday cups can be installed with either 1011, 1012 or 1013 Ohm resis- tor amplifi ers. A simplifi ed overview of the system at the VUA is shown in fi gure 2. The VUA instrument was fi tted with either 1012 Ohm or 1013 Ohm Faraday amplifi ers on the H2 and H1 channels for the experiments described in this chapter. After ionization of the gas following the electron bombardment principle, the ion- ized beam is aligned using an optical lens system to enter the fl ight tube. A coiled W-fi lamentemits electrons with a 200 micro-Amp electron-trap current, the ionization voltage, i.e the energy of the electrons, is minus 70 eV, and source magnets cause the electron path ways to be spiraled in order to optimize ionization effi ciency. The argon ions are pushed by a ca 3 V ion-repeller voltage from the ionization chamber towards the accelerator part of the source which operates at an acceleration voltage of 10 kV. The fl ight tube has specifi c 35 cm radius and a curvature of 120° (Fig. 6.2.a-b). The 6 instrument resolution is ca 800 for the standard collectors H2, H1, AX, L1, and >1500 for L2, fully resolving all hydrocarbons peaks and partially resolving 1H35Cl from 36Ar on the L2 peak (Santato et al., 2015). Faraday amplifi ers for the L2, L1, and AX col- lectors are set up with 1012 Ohm resistors, whereas the Faraday amplifi ers of the H1 and H2 channels have been equipped with 1013 Ohm resistors and are compared with the previously installed 1012 Ohm resistor. The 1013 Ohm resistor amplifi er presents 4-5 times higher precision in measurements of small ion beams with respect to the previous 1011 and 1012 Ohm amplifi ers (Santato et al., 2015). The L1, L2, H1, and H2 collectors can be physically moved whereas the AX collector is fi xed. The beam size of the CDDs should be limited to beam intensities less than 50 fA (equivalent to about 300,000 cps). A schematic map of the Helix MC plus operating at the VUA is presented in Figure 6.2c. Most of the additional components of the vacuum extraction system and the laser beam delivery system have been designed and produced in-house at the VUA. The system at the VUA is fully automated and is based on modular con- trol devices controlled using the new Qtegra mass spectrometer operating platform of ThermoFisher. The Qtegra software platform by Thermo Fisher Scientifi c controls both the mass spectrometer and auxiliary devices via scripts launched using C# and

137 Improving the precision of Ar-dating: implication for provenance .NET coding. Gain biases between different collectors are determined by measuring

the beam intensity at mass 44 (CO2) in dynamic mode and these data are processed using an in-house designed Excel macro. For data reduction, we convert the multi- collector data format from .csv format as produced by Qtegra to formatted text fi les as used by the ArArCALC software of Koppers et al. (2002). Factory specifi cations for 1012 Ohm amplifi er baselines are tested during 30 min with 4s integration time and meet specifi cations when 1 standard deviation (SD) is <0.1 fA (Santato, et al., 2015). These specifi cations were also tested in-house for both 1012 and 1013 Ohm amplifi ers in addition to tests with 33s integration times. Baselines were stable for both amplifi ers; however, noise levels were about 2 up to 2.5 times lower for the 1013 Ohm amplifi ers (Monster, 2016). The 1013 Ohm amplifi er used for this study shows an increase of a factor of 10 of the instrument sensitivity when com- pared to the 1012 Ohm amplifi er (Santato et al., 2015). Thus, the new amplifi er shows 4-5 times better signal/noise ratio and this results in an improvement in the measure- ment precision and reproducibility of small sample analysis. 6.2.2 Experimental design As far as we are aware, as yet no 40Ar/39Ar experiments were reported on the preci- sion of analysis of standards and unknown samples for different grain sizes using 1013 Ohm amplifi ers. In this study, we compare results on both standards and unknowns. We performed a series of experiments on our in-house fl ux monitor standard Drachen- fels sanidine (DRA) (Schneider et al., 2009) using two different amplifi ers, a 1012 and 1013 Ohm, on the Faraday H1 and H2 cups. We used a modifi ed age calibration follow- ing Kuiper et al. (2008). We used samples from two irradiations (VU101, VU108) that have been irradiated for respectively 12 and 7 hours in the CLICIT facility of the OSU TRIGA reactor. The 39Ar values were normalized to the 7 hours irradiation as shown in section 6.3.1 to allow direct comparison. Analysis of muscovite detrital samples was performed on fi ve selected grain size fractions: 125-180, 180-250, 250-355, 355-500 and 500-1000 microns. All samples were separated using standard mineral separation procedures. In total ~400 grains have been analyzed. For each size fraction, we analyzed a variable number of single laser fusion analyses as shown in Table 6.1. The data for total fusion experiments of DRA sanidine standard single crystals on both 1012 and 1013 Ohm amplifi ers is shown in Fig. 6.3. Raw data (blanks, stand- ards, air pipettes, unknowns) reduction was performed using the ArArCALC2.5 free-

ware data reduction package (Koppers, 2002). CO2 measurements in dynamic mode performed prior or after the samples analysis were used for inter-channel gain bias

138 Chapter 6 correction and processed using a separate in-house developed Excel macro. The in- house procedure developed for gain bias correction using CO2 is described in Monster, (2016). This procedure allows for the inter-channel gain bias corrections factors to be determined. The uncorrected intercept values for blanks, unknowns, standards, and airs are obtained from regression within ArArCalc. Intercept values are corrected for gain bias by dividing the measured intensities by each respective gain bias correc- tion factor using a VBA macro for Microsoft Excel. The corrected values of the mass discrimination factor (MDF) are fi nally added to the measurement of standards, and later the inter-channel bias corrected and MDF-corrected J-factors are added to the sample measurement. Since the interchannel gain bias factors are stable over periods of weeks, ignoring/omitting gain correction hardly infl uences the fi nal ages as long as standards, unknowns and air-shots are measured in the same period (within a week) and with the same cup confi guration (see. e.g. Phillips and Matchan, 2013). Instead of weighing individual grains, we use the 39Ar intensity as a proxy for sin- gle crystal grain size (although this is not completely correct if the K concentrations are not constant). Since the 39Ar intensity is a function of irradiation time the VU101 39Ar intensities are systematically higher when compared to VU108 intensities and we 39 need to correct for this. We defi ne Arnorm intensity as the normalized value to 7 hrs irradiation following equation (6.1): 6 39 Arnorm = A × t1 / t2 (6.1)

Where, A is the value of intensity in femto Ampere (fA), corrected for blank and mass discrimination, t1 is the time of irradiation for VU108 = 7 hours, and t2 is the time of irradiation for VU101 = 12 hours. The standard deviation (1σ) associated with the 39Ar intensity was calculated using Eq. 6.1. The normalized values were plotted together in a scatter plot (Figure 6.3).

139 Improving the precision of Ar-dating: implication for provenance Figure 6.2. Design of the Helix MC plus installed at the Geochronology Laboratory of the Vrije University Amsterdam (VUA). a) Major components of the Helix MC plus operating at the VUA b) detector set-up (modifi ed after Monster, 2016). c) Schematic map of the system at the VUA (Designed by Onno Postma).

6.3 Result

6.3.1 Improved precision of 1013 Ohm amplifi er tested on in-house Drachen- fels standard. The J value is a dimensionless irradiation parameter that can be measured by ir-

140 Chapter 6 radiation of a mineral of known age (fl ux monitor) (McDougall and Harrison, 1999). Figure 3 shows the reproducibility of J values and their standard deviations (2SD) for analysis on standards using the 1012 Ohm amplifi ers or the 1013 Ohm amplifi er on H1 and H2. The analytical uncertainty of individual measurements achieved with the 1012 Ohm amplifi er ranges from 0.5 - 1 % (2SD, n=45), whereas using the new 1013, we obtained an uncertainty that ranges from 0.1 - 0.3 % (2SD, n=30). Overall, analy- 13 sis on standards (sanidine - DRA) using the 10 Ohm resistors (JX1 = 0.0019101 ±

0.0000045, JX2 = 0.0019133 ± 0.0000043) improve the standard deviation by a factor 12 of ~2 when compared to the 10 Ohm resistor (JF13 = 0.00187111 ± 0.0000094, JF14

= 0.0018510 ± 0.0000097, JF15 = 0.0018409 ± 0.000009). In fi gure 3 we see that for lower intensity of the signal the 1012 Ohm amplifi er measurements result in higher er- rors when compared to the analyses performed using the 1013 Ohm amplifi er. Further, the measurement of standards with the 1012 Ohm amplifi er show a negative relation between J-value and beam intensity not shown in the 1013 data. The linear decrease of the J-values with increasing intensity is particularly evident in samples F-13, F-14 and to a minor extent in F-15 (Fig. 6.3). 6

Figure 6.3. J-values of the in-house Drachenfels standards analyzed with 1012 and 1013 Ohm amplifi er versus 39Ar intensity (fA). The values are normalized to 7 hours irradiation as ex- plained with Eq. 1.1. The median value is indicated for each standard with the same color code. The error bars display the 2σ uncertainty.

141 Improving the precision of Ar-dating: implication for provenance 6.3.2 Measurement of different grain size of detrital muscovite from river samples Different fractions of white mica from fi ve modern river sand samples (Table 1), from two different irradiation batches, were analyzed using the 1013 Ohm amplifi er on the H2 and H1 Faraday cups of the Helix MC plus at the VUA.

Table 1. Analyzed fractions of mica grains km No. of analysis River Sample ID Irradiation downstream* abcde Siang ~214 S1 VU 101-108 15 15 15 15 Siang ~318 S2 VU 101-108 15 45 15 15 Brahmaputra ~610 S3 VU108 7 15 Brahmaputra ~795 S4 VU 101-108 35 30 15 Brahmaputra ~1222 S5 VU 101-108 6 20 68 15 45 Note. * km downstream from the Namche Barwa syntaxis; a) 125-180; b) 180- 250; c) 250-355; d) 355-500; e) 500-1000

The composite age distributions of the Siang-Brahmaputra river samples are pre- sented in fi gure 6.4 as Kernel Density Estimator (KDE) and Probability Density Plots (PDPs) using DensityPlotter software (Vermeesch et al., 2012). Overall, we observe four major age peaks of ~0.1 - 3.5 Ma, ~5 – 10 Ma, ~10 - 20 Ma and ~20 - 50 Ma (fi g. 4). The plot shows ages as young as few thousand years but most ages range between ~10-30 Ma. Ages younger than ~ 3 Ma refl ect the onset of source material drained from the Namche Barwa syntaxis. The young signal has been recorded in both in-situ and detrital thermochronology in the area (Lang et al., 2016; Bracciali et al., 2016). The PDP presented here does not account for variation in the grain size of the dated muscovite or its exact location downstream, but it is representative of the age distribu- tions observed up to ~1500 km downstream from the Namche Barwa syntaxis in the Brahmaputra river. The 39Ar isotope intensity measured using the H1 Faraday channel fi tted with the 1013 Ohm amplifi er is plotted for each sample’s grains sizes (Figure 5a). As the 39Ar signal is a proxy for the potassium content of the mineral it serves as a measure for the grain size, provided the K concentration (and irradiation time, therefore we normalized our 39Ar signals) is constant. In fi gure 5 each fraction is represented by a color code and a symbol that are characteristic for the irradiation and the fraction’s size. Overall, the ages are scattered when plotted relative to their 39Ar intensities with a major group clustering at intensities of 0.5 - 10 (fA) 39Ar. The grain sizes >250 mi- crometers (we analyzed 250-355, 355-500, 500-1000 mm size fractions) are charac-

142 Chapter 6 terized by an increase in the 39Ar intensity to ~100 fA and show more or less the same 39 scatter for Arnorm < ~25 fA, while above this intensity they seem to cluster between 25 and 35 Ma. The 355-500 μm and 500-1000 μm fractions show a similar behavior (Figure 5a). The 500-1000 μm fractions are characterized by ages older than 10 Ma and, as expected, the highest values of 39Ar intensity.

Figure 6.4. Composite age distribution of the fi ve Himalayan samples plotted using Radial 6 Plotter by Vermeesch (2012) using all size fractions. The age on the x-axis is displayed on a logarithmic scale. The gray area shows the Kernel Density Estimator (KDE), the black line the Probability Density Plot (PDP) and the black dotted lines the histograms. The range of the young signal draining from the Namche Barwa syntaxis is indicated in the top.

The data points for grain sizes smaller than 250 mm (Figure 5b) show a range of ages scattered between 0.1 Ma and ~45 Ma (Figure 6.4) and are characterized by 39Ar intensities lower than 10 (fA). However, if we zoom in we see that the 180-250 μm fraction records an increase of the 39Ar intensity to values between 4f and ~8fA and ages at ~17-25 Ma (fi gure 6.5b). We notice that analytical uncertainties of 39Ar inten- sities <3 fA are generally high, on the order of 30-60% of the total measured intensity and that is refl ected in the higher analytical uncertainty of corresponding ages. Fur- thermore, it is important to note that a few data points present very low 39Ar intensities in the range of 0.05 - 0.08 fA, that is close to the electronic baseline for this amplifi er (Santato et al., 2015), and to the measured blank values, and possibly indicating that the samples were either not fused well or they were not muscovite grains. The 39Ar intensities versus age distributions are presented for each sample in fi g- ure 6.6. The data varies within the range of ages characteristic for this area, whereas the intensity varies as a function of the target grain’s size.

143 Improving the precision of Ar-dating: implication for provenance a. 100 90

80 Key: 70 VU108 125-180 VU108 180-250 60 b VU108 250-355 50 VU108 355-500 VU101 125-180 40 VU101 180-250 VU101 250-355 Age (Ma) Age 30 VU101 355-500 VU101 500-1000 20

10

0 020 40 60 80 100 120

39 b. Arnorm. [fA] 50

45

40

35 Key: 30 VU108 125-180 VU108 180-250 25 VU101 125-180 VU101 180-250

Age (Ma) Age 20

15

10

5

0 021 3 4 56 7 39 Arnorm. [fA]

Figure 6.5. Scatter plots of the age distribution versus 39Ar intensity of the muscovite data. Each color represents a fraction of grains. The circles represent the samples of the VU108 7-hours irradiation and the triangles the VU101 12-hours irradiation. The errors bars represent the one standard deviation of 39Ar intensity (horizontal or x error bars) and the two standard deviations of the absolute age (vertical or y error bars). a) All the data and fractions. b) frac- tions of 125-180 and 180-250 μm. Note the difference in the intensity on the x-axis between the two plots. Note also that some data points present very small age errors and the uncertainty bars are not visible. The upstream Siang sample S1 presents four fractions ranging from 125 μm up to 500 μm (Figure 6.6a). Sample S1 (Figure 6.6a) is characterized by ages ranging from ~0.1 and 31 Ma. The smaller fraction 125-180 μm is character-

144 Chapter 6 ized by a heterogeneous range of ages and generally low intensities that range between 0.05 - 2 fA. This fraction interestingly shows an age population at ~10-20 Ma that is lost for the other fractions. The errors for the intensities are generally high 30-50% for the 125-180 μm fraction and in the order of 20-25% for the 250-355 μm fraction. The other two fractions present lower errors in the order of 10-15 % (Fig. 6.6a). Furthermore, we notice that despite two single grain ages at ~1 and 15 Ma, the 355-500 grain size present the 95% of ages at 18-25 Ma. Sample S2 (Fig. 6b-c), presents similar age distributions with the smaller fraction that are characterized by 39Ar intensities lower than 3 fA (Fig. 6.6c). In this sample, the 400-500 μm is characterized by a cluster of grain ages between 25-35 Ma and intensities between 10-80 fA. Sample S3 (Fig. 6.6d) yield only data points for the smaller fractions (125-180 μm and 180-250 μm) with values for intensity of 39Ar around 0.05 - 2.5 fA and ages ranging from ~0.5 to 45 Ma. The analyzed grain sizes present the complete range of ages characteristic of the studied area. The downstream Brahmaputra sample S4 present three-grain size fractions (Fig. 6.6 e) and show 39Ar intensities ranging from ~0.1 to 12 fA. The smaller size fraction yields the youngest ages of ~0.2 - 3 Ma. The grain size fraction 355-500 microns show a wider range (~3 - 42 Ma) of ages when compared to previous samples. 6 The farther downstream sample S5 (Fig. 6.6 f-g), show fi ve different size fractions that are characterized by an age distribution ranging from ~1.2 – 50 Ma. However, the youngest < 2 Ma ages are only a few grains for this sample and are detected from the 125-180 and 250-355 grain fractions. The biggest fractions of this sample are charac- terized by two well-defi ned components of ages at ~30 Ma and at ~10-17 Ma (Figure 6f). The grain sizes <250 μm are characterized by most of the ages generally < 30 Ma. The 250-355 grain fractions present most of the ages at ~0.5 and 28 Ma with the 90 % of the total age distribution at ~10-30 Ma.

145 Improving the precision of Ar-dating: implication for provenance a. b. S1-Siang S2-Siang 50 50 40 40

30 30 20 20 Age (Ma) Age (Ma) 10 10 0 0 012345678910 020406080 39 39 Arnorm [fA] Arnorm [fA] c. d. S2-Siang* S3-Brahmaputra 50 50

40 40

30 30

20 20 Age (Ma) Age (Ma) Age 10 10

0 0 0123456 0123456 39 39 Arnorm [fA] Arnorm [fA] e. f. S4-Brahmaputra S5-Brahmaputra 50 50

40 40

30 30

20 20 Age (Ma) Age (Ma)

10 10

0 0 0 5 10 15 20 0 5 10 15 20 25 30 35 40 45 50 39 39 g. Arnorm [fA] Arnorm [fA] S5-Brahmaputra* 50 KEY 125-180 180-250

40 250-355 30 355-500 20 500-1000 Age (Ma) 10 <6 fA, sample S2 0 <6 fA, sample S5 0123456 39 Arnorm [fA]

Figure 6.6. Detailed scatter plots of the age distribution versus 39Ar intensity of the mus- covite data per sample. The error bars represent the 1σ uncertainty of the 39Ar intensity. The 39Ar values have been normalized using Eq. (6.1). Each grain size is identifi ed by a color code as displayed in the key of the fi gure. The x-axis scales between the different plots. A) Most upstream sample in the Siang river S1; B) Pasighat sample before the outlet of the Siang river in the Brahmaputra foreland S2; C) Inset of the smaller fractions for sample S2; D) Upstream sample of the Brahmaputra river S3; E) Median sample of the Brahmaputra river S4; F) Most downstream sample of the Brahmaputra drainage S5 (see fi g. 6.1 for locations). G) Inset show- ing the smaller fractions for sample S5.

146 Chapter 6 6.4 Discussion

6.4.1 In-house standard analyses using 1012 Ohm and 1013 Ohm amplifi ers Analysis of small and young sanidine minerals has previously been tested with Argus VI (Philips and Matchan, 2013; Kim and Jeon, 2015) showing that multi-col- lector system allows improvements of an order of magnitude in analytical precision compared with single collector analysis. Our new data using 1013 Ohm amplifi ers installed in the Helix MC show that using a grain size fraction of sanidine of 400-700 microns yield a factor 2 increase in precision for the fl ux monitor values. This result is in agreement with previous observations carried out on the Helix MC plus from air measurements and baselines that show important improvements of both signal/noise ratio and external precision (relative standard deviation of ~0.18 % of the 40Ar/39Ar ratio) (Santato et al., 2015). The total system sensitivity is ~3.5 × 10-13 mol/V for the Helix MC plus at VUA. This value is comparable to MAP215-50 at VUA ~7.08 × 10-19 mol/cps, whereas it is slightly lower than the Argus VI with ~1.5 × 10-14 mol/V. Furthermore, our results show that standards measured with the 1012 Ohm am- plifi er present an important increase of the standard error for grains with lower 39Ar intensities < 40 fA and higher J values. Whereas, standard measured with the 1013 Ohm amplifi er present a more uniform distribution of the single grain analysis and an 6 improved precision of the errors for smaller beam size (please compare error bars of fi gure 3). This comparison allows us to demonstrate that for smaller than ~50 fA beam intensity the 1013 Ohm amplifi er present a ~2 times better precision if compared with analysis with the lower resistor (1012). Remarkably, for standards measured with the 1012 Ohm resistor, we show a lin- ear decrease in the J-value with increasing intensity of the 39Ar. This linear drift is particularly clear for Drachenfels samples F-13 and F-14 J values but less evident in F-15. Potential causes for this drift in J-value are 1) procedural issues related to blank measurements. Blank ratios can infl uence the precision with which we can measure a 40Ar*/39Ar ratio (Kelly, 2002), low system blanks are therefore crucial in the 40Ar/39Ar analysis. A detailed assessment of the blanks of the VUA system has been performed in fall 2014. Samples F-13, F-14, and F-15 were measured in the winter of 2014 (VU101) when the Helix MC plus extraction line and mass spectrometers were rela- tively new and presented a total system blank of about 7-8 fA of 40Ar. Blanks obtained in the run analysis for the VU108 irradiation are generally ~ 1-2 fA of 40Ar. The high resolution of the Helix MC plus (>750 up to >1500 at 10% peak-valley, Santato et al., 2015) has high interferences resolving power that allow a complete resolution for the discrimination of contaminant effect (HCl and hydrocarbons). There is a low

147 Improving the precision of Ar-dating: implication for provenance probability that hydrocarbons contamination can be a source of bias for standards analysis and we would have expected a similar behavior for standard X1 and X2. 2) The difference in irradiation time between two batches; the amount of 39Ar produced by 39K is proportional to the time of irradiation and of the intensity of the neutron fl ux (McDougall and Harrison, 1999). J-values are determined in function of the position of the mineral standards in the cylinder and of the duration of the irradiation. The time of the irradiation (12 hrs) has been the same for F-13, F-14, and F-15. Neutron fl uences in a nuclear reactor are not homogeneous. The number of neutron fl uences that standards and unknown material receive are variable (Renne et al., 1998; Rutte et al., 2015). Neutron fl uence gradients are signifi cant both in vertical and horizontal direction and can, therefore, be a source of biases for high precision dating analysis in the order of 0.5 to 1.0 % (Renne et al., 1998). Shielding effects during irradiation procedures on few amounts of material are generally small and cannot lead to neu- tron fl ux variation, however, sample rotation during irradiation procedures might not always be suffi cient to reduce shielding effects suffi ciently to allow high precision dating (Rutte et al., 2015). We argue that the most likely explanation for the difference of the J-values for standards analyzed in 2014 (F-13, F-14, F-15) is associated with some procedural is- sues related to the blanks analysis. The analysis of the F-13, F-14, F-15 are older than the analysis of standards X1 and X2. In 2014 the Helix MC plus at the VUA was still under overhaul of some components of the extraction line (e.g. extraction line parts). Outgassing of stainless steel yielded blank values that were generally variable over the duration of one single experiment as previously reported by Monster et al. (2016), this could have propagated bias in the analysis of the standards analysed in that period. 6.4.2 Analysis of young and small muscovite grains from the Eastern Hima- laya We analyzed a set of detrital muscovite crystals from modern river sedi- ment samples located at increasing distance in the Siang-Brahmaputra river downstream from the Namche Barwa using fi ve different fractions per sample where possible. The purpose is to identify if different grain size intervals con- tain different age populations. If such differences exist, implications are far- reaching and would potentially disqualify previous studies that focused only on one-grain size fraction. If such a relation is non-existing, the one-grain size fraction approach is supported. As expected, our data show a signifi cant decrease of the 39Ar intensity (as well as for 40Ar intensity) when performing analysis on grains <180 μm that serves as a proxy for grain-size. We have also

148 Chapter 6 noticed that values of 40Ar % of ages < 3 Ma are generally at ~30-40 %, this has been noticed before from sample collected in similar locations in the Siang river (Lang et al., 2016). The high-precision analysis is therefore required to precisely measure low 39Ar intensities (<10 fA) associated with the measure- ment of grain sizes < 180 microns. The single-grain analysis with the 1013 Ohm amplifi er shows a good improvement of the cooling ages with a precision that varies from ~0.1-2.5 % for the analyzed fractions. The smaller fractions (< 180), however, present ages < 2 Ma that yields a higher error in the order of 10-30%. In order to discuss the effect of grain-size variability on the age distributions at increasing distances from the Namche Barwa syntaxis in the Brahmaputra river, the PDPs of three sequential samples (S1 ~170 km; S4 ~700 km; S5 ~1200 km) are dis- played in fi gure 6.7. The age distributions are generally not bimodal but show differ- ent components in the ~0-50 Ma range, with a few exceptions e.g. Sample S5 where the 355-500 fraction presents most of the ages at ~30 Ma. The young ~0.01 - 3 Ma population is derived from rocks exposed to young exhumation in the Namche Barwa syntaxis (Bracciali et al., 2016), however, the dilution of this signal has been previ- ously reported for muscovite analysis (Chapter 5). Our detrital data from the Siang locations S1 and S2 can be compared with pre- 6 vious studies (Bracciali et al., 2016; Lang et al., 2016), that document similar age distributions. Lang et al. (2016) reported 40Ar/39Ar muscovite analysis from a similar location at Pasigath (S2) using grain-size fractions of 250-500 and 500-1000 microns. In their work, Lang et al. (2016) assume ages older than 9 Ma to be diagnostic of the contribution of the Namche Barwa syntaxis. They argue that due to its higher closure temperature, the muscovite ages would have given older ages with respect to young ~< 3 Ma biotites dated in the bedrock of the core of the Namche Barwa syntaxis (Booth et al., 2009; Zeitler et al., 2014). The detrital muscovite age distribution refl ects both the thermochronologi- cal complexities of the eroded material, and the bias acquired during erosion, transport, and deposition, such as hydraulic sorting and mineral abundance of datable minerals at the soucre rock (Garzanti et al., 2015; Malusá et al., 2016; Braun et al., 2017). Many lines of evidence show how the components of age distributions can vary as a function of the analyzed fraction and of the distance from the source. If we look carefully at the age distribution derived for every grain size fractions (fi g. 6.7), we notice that we can obtain a different answer in function of the size of the analyzed target mineral and of the position in the

149 Improving the precision of Ar-dating: implication for provenance river. In particular, the multi-grain fractions analysis shows that the young (<3 Ma) signal, drained from the Namche Barwa syntaxis, is preserved in all the grain’s fractions (fi ne and medium) analyzed from the stations proximal to the source area (S1 of fi g. 6.7). Whereas, in the distal samples (S4 and S5 of fi g. 6.7) the young signal disappears from the grain fractions above 355 microm- eters. The only exception is the grain-size fraction of 180-250 from sample S5, where we do not record any ages < 3 Ma. It implies that for the proximal samples the age distribution would not have been affected by the analysis of grains of 250-1000 microns. Whereas, while proceeding at increasing distanc- es downstream in the main river stream the number of sources drained in the basin, as well as the biases introduced by the hydraulic sorting is increased. For example, in the farthest downstream sample, the grain size fraction 355-500 microns shows a monotonic ~33-37 Ma peak that is not recorded from any of the analyzed grain size fractions of sample S1. Importantly, we see that from the farthest upstream sample S1 there is a major event at ~10-20 Ma that is re- corded in all the grain-size fractions as a major component of the age distribu- tions. This cooling event is well recorded as well in the Brahmaputra samples in all the grain size fractions. This event is well described in the Himalaya and it is probably related to the rapid extrusion of lower crust material along major tectonic discontinuities (Beaumont et al., 2001; Kellet at al., 2013; Grujic et al., 2006). Rapid physical erosion characterizes most of the Siang-Brahmaputra catchments and hydraulic sorting processes are more signifi cant than mechanical breakdown, that is generally low even over distances of thousands of kilometers (Garzanti et al., 2004; 2015). As far as hydraulic sorting and mineral abundance can be considered a major source of bias for the accuracy of age distributions from river’s sediments (Malusá et al., 2016; Braun et al., 2017), the multi-grain size fractions analysis should be taken into proper account while increasing the size of a catchment area (~10000 km2). For most of the detrital mineral isotope provenance studies, the number of single grain analysis is an important parameter to consider for characterizing the consistency of the age distribution. Often we are about to ask ourselves, how many grains do I need to date to be sure that I do not miss any of the components? However, the bias introduced by the analysis of single grain size fraction analysis has been a much more poorly researched topic. Gemignani et al. (2017), analyzed 19 modern river samples from the Eastern Alps, and showed that 30-60 grains per samples for catchments of small-medium size (~<

150 Chapter 6 10.000 km2) yield precise enough coverage to detect all the eroding sources compet- ing in a river network when the in-situ ages of bedrocks are well constrained and limited in number. The Eastern Himalaya and the Brahmaputra river present rates of erosion/exhumation and sediment fl ux that are of one order of magnitude higher when compared to the Alps (Garzanti et al., 2004; Singh and France-Lanord, 2002; Enkelmann et al., 2011; Zeitler et al., 2014). Despite a great effort, bedrock thermochronology has been only poorly assessed due to the large areal extent and inaccessibility of much of the Himalaya. For those reasons, the analysis of the detrital age distributions requires a robust statistical num- ber, coupled with a multi grain-size fraction analysis. Vermeesch, (2004), computed using a statistical relationship the probability that no component is missed in prov- enance studies. In the best-case scenario, for example, if we want that no part of the component comprising the 0.05 of the total population is missed at the ~95% confi - dence level, at least ~95-117 grains should be dated. The <3 Ma signal is at S1 the ~7 % of the total age distribution and we do observe this component in all the analyzed grain sizes. We can, therefore, validate that the number of analysis of this sample (n = 108) is robust enough for our purposes, or follow Vermeesch (2004) that we have the 95 % confi dence that no fraction > ~0.9 % was missed. The total number of analysis for the downstream sample S4 is n = 92 and we do miss the <3 Ma component only for 6 the 355-500 microns fractions where we performed n = 26 single grain analysis. The number of analysis of the downstream sample S5 (n = 146), where we likely expect to record most of the different sources draining from the eastern Himalaya, is well in line with the recommendation of Vermeesch, (2004). However, for the medium-size frac- tion, the young signal is diluted, and previous analysis that only considered a grain size of 250-1000 microns could have been biased, despite a suffi ciently large number of analysis (~ n=100), by the fact that grain-size variability was not taken into account.

151 Improving the precision of Ar-dating: implication for provenance Siang trunk Brahmaputra trunk ~160 km from NB ~700 km from NB ~1300 km from NB

S1, 125-180 S4, 125-180 S5, 125-180 7 11 2 (n=33) (n=27) (n=5) eaiedniyRelative density 6 Relative density Relative density

5 8 Relative density 4 3 5 1 Number Number 2 Number 1 2 0 0 0 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 S5, 180-250 S1, 180-250 8 S4, 180-250 10 10 (n=15) 7 (n=33) (n=19) Relative density Relative density Relative density 6 7 7 5 4 5 5 Number 3 Number

Number 2 2 2 1 0 0 0 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 S5, 250-355 S1, 250-355 19 20 (n=63)

(n=65) Relative density 15 14

10 9 Number

Number 5 4

0 0 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 S1, 355-500 S4, 355-500 S5, 355-500 5 6 12 (n=14) (n=26) (n=18) 5 Relative density 4 Relative density Relative density 4 9 3 3 6 Number Number

2 Number 2 3 1 1

0 0 0 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 Age (Ma) Age (Ma) 23 S5, 500-1000 Tot analysis: n = 108 Tot analysis: n = 92 (n=41) 17 Relative density

11 Number

5

0 0 5 10 15 20 25 30 35 40 45 50 Age (Ma) Tot analysis: n = 146

Figure 6.7. Kernel Density Estimator (KDE; orange shaded area), Probability Density Plots (PDP; black line) and histograms (red dotted lines) of the 39Ar intensity (fA) of the upstream Siang sample S1, the middle Brahmaputra S4, and the downstream Brahmaputra sample S5 for the different grain’s fractions. The distance in km from the Namche Barwa and the total number (n) of analysis per sample are indicated. The age distributions are shown in the upper right corner with the number of analysis in the upper right corner. The age distributions present the KDE (Pale grey area), PDP and histograms (grey dotted line).

152 Chapter 6 6.5 Summary

We investigated the performance of the newly installed 1013 Ohm resistor on H2- and H1- Faraday amplifi er channels of the Helix MC plus multi-collector noble gas mass spectrometer. Comparing the 1012 and 1013 Ohm resistors on the in-house Drachenfels sanidine standards, we fi nd that the precision of the J-Values standard error increased by a factor of 2 using the 1013 Ohm amplifi er for smaller (< 10 fA) beam intensity. Our work yields a similar conclusion as a study carried out on TIMS standard analyses (Koornneef et al., 2014). We performed a robust analysis (total n=615) on mica samples using the 1013 Ohm set-up using fi ve fractions per sample where possible. Our new data are the fi rst attempt to test the dating of young and/or small detrital muscovite decreasing the fractions down to 125 microns. We fi nd that grains with ages > 2 Ma and grain size of 250-400 microns present acceptable relative standard deviation in the order of 0.1- 2.5 %. Whereas analyzing samples below 250 microns and ages < 2 Ma, the errors are generally slightly higher 0.5-10 % of the mean age. Overall, the uncertainties on DRA-standards are lower for the 1013 Ohm amplifi er when compared with the 1012 Ohm amplifi er and this is refl ected in a more precise determination of the mineral’s ages. We tested the effect of the grain size variability in age distributions from the same 6 samples. Grain-size variability seems to bias the age distribution towards unimodal components while narrowing the interval of the grain size. In particular, we notice that the range of grain sizes of 355-500 micrometers of sample S2 present a unimodal distribution of ages at 30-35 Ma. The same unimodal distribution of age is observed downstream in S5 for the same fraction. Our data support the hypothesis that grain- sizes variability can bias age distributions. We conclude that for the study of dynami- cally changing and complexes mountain belts the multi-grain size analysis approach is required in order to improve the reliability of the detrital age distributions. Acknowledgement: The work leading to the study presented here was supported by the People Pro- gramme (Marie Curie Actions) of the European Union’s Seventh Framework Pro- gramme FP7/People/2012/ITN, Grant agreement number 316966. Janne Koornneef is acknowledged for the early review of this paper. Onno Postma is aknowledge for having designed the Helix scheme presented in Figure 6.2.

153 Improving the precision of Ar-dating: implication for provenance Chapter 7 Synthesis and Conclusions 7.1 Synthesis and conclusions The aim of this thesis project is to improve knowledge on the possibilities and lim- itations of detrital thermochronology to constrain the temporal and spatial variability in exhumation on geological timescales. The objective of this thesis has been to study the link between climate-driven erosion, river transport, and exhumation, using detri- tal thermochronology. I have presented detailed multi-proxy data analyses and model studies of actively eroding mountain ranges. To achieve the objective of this thesis, the fi rst part of this manuscript has been devoted to introducing the fundamentals of radio-isotopic dating techniques. I have presented a new linear mixing model, which was developed to quantify the relative intensity in erosion of adjoining catchments drained by a trunk stream and its tributaries, using detrital age distributions. The sec- ond part of this thesis has applied the detrital thermochronology approach to differ- ent settings (the eastern Alps and the eastern Himalaya), to explore the possibility of deriving fi rst-order information about the exhumation history and present-day erosion of actively evolving hinterlands. In particular, I have explored the potential effect that dilution and grain-size variability exert on age distributions of modern river sands in river systems as a function of distance from the source. Finally, I present a study exploring the applicability of analysis of younger and smaller grains of muscovite, using a latest generation multi-collector noble gas mass spectrometer for argon dating. Chapter 2 focused on the 40Ar/39Ar and fi ssion-track techniques, presenting their historical background followed by their principles, applications, and limitations. Pre- vious studies have applied these techniques as a provenance tool and to derive con- 7 straints on regional scale exhumation, assuming a steady-state relationship between tectonics and erosion (e.g. Bernet and Spiegel, 2004). This assumption, however, pre- cludes the likely transient response of an orogen to changes in tectonic and climatic forces, and in many cases is not justifi ed (Rahl et al., 2007). Chapter 3 focuses on how detrital age distributions can be used as a tool to infer spatial variability of the exhumation rates. We have developed a numerical mixing model that does not require verifying the steady-state assumption or resolving the complex thermal structure of the eroding catchment. The model resolves the down- stream evolution of binned age distributions by linearly inverting the observed raw age bins in function of assessable parameters to determine the relative erosion for each catchment or sample station. The reliability of the computed erosion estimates depends on the observed detrital age distributions: do they represent the fi ngerprint of the exhumation/erosion of the studied source? A conceptual sketch of the linear mixing model is presented in fi gure 7.1. Detrital

157 Synthesis and conclusions age distributions are arbitrarily grouped in bins of ages (A1 and A2). Each contribut- ing basin is characterized by an exhumation rate (Ex and Ey). The linear inversion aim to quantify relative fi rst-order estimates of the erosion necessary to produce the age distribution that characterizes the catchment area (fi g. 7.1b). The analysis of the tributaries to the main trunk has been used to add constraints on the evolution of the signal with respect to local source infi ll. Because the detrital record refl ects both the thermochronology of the source (the eroded bedrock), and the bias derived from erosion, transport, and deposition, I have quantifi ed the amount of hydraulic sort- ing and mineral fertility bias by estimating the concentration factor α (target mineral abundance at the source). Using a bootstrapping approach I have also computed the errors associated with the estimates of the present-day erosion. The method provides estimates on the spatial distribution of relative erosion rates in the analyzed areas; independent analysis is needed to transform these into absolute numbers.

A) Age bins are colors Ex Ey Linear mixing model A2 A2 A1 B) option 1 A1 if Ex>Ey then (A1/A2)x > (A1/A2)y option 2 option 1 if ExEy (A1/A2)x < (A1/A2)y

? option 2 Ey>Ex

Figure 7.1. Conceptual sketch of the mixing model. A) Evolution of age bins in a river system. The river is indicated in blue; the circles represent the sampling stations. B) verifi cation of the mixing model. Colors are age and the two options are explained in term of the relationship between age bins A1 and A2 of area x and area y. In Chapter 4, I have tested 40Ar/39Ar analysis of detrital biotite and muscovite grains as a sediment provenance tool in the eastern Alps. Three major pulses of exhu- mation are recorded by the age distributions derived from 19 modern river samples collected north of the Periadriatic line. The clusters of ages correspond to the clas- sical Varisican (350 - 280 Ma), Pre-Alpine (150 – 50 Ma) and Alpine (50 - 0 Ma)

158 Chapter 7 tectonic episodes. To assess the reliability of the detrital age distributions, I compared them with bedrock ages from their known source areas. The application of the mixing model (developed in Chapter 3) to four catchments drained by the Inn River produced consistent estimates of present-day relative erosion rates. The mixing model predicted higher erosion rates in catchments contributing younger cooling ages, which are de- rived from the Penninic units of the Tauern and Engadin windows, whereas lower estimates for erosion rates are derived for catchments containing Austrolapine units. I have also quantifi ed the cooling rates of the crustal-scale dome of the Tauern Window during the Miocene using a detrital-age elevation method (Rhul and Hodges, 2005). Our results suggest asymmetrical exhumation at rates of ~0.8 mm/yr for the western and ~0.3 mm/yr for the eastern Tauern window. In the Eastern Alps, the cooling histo- ry of the Tauern window is known with a relatively high degree of precision (Bertrand et al., 2017; Fox et al., 2016; Zimmermann et al., 1994). The asymmetric exhumation pattern has been associated with a westward increase of shortening and crustal thick- ening. This study demonstrates how muscovite and biotite 40Ar/39Ar ages can be used as a proxy to infer exhumation and erosion patterns from the modern river sediments. However, the downstream evolution of the detrital signature can affect the inter- pretation of the relative erosion rates in the source areas. In particular, the potential effect of downstream dilution of such a signal can strongly affect interpretations but is not often considered. The variation of the detrital thermochronological signal within samples collected along a major trunk river is addressed using a multi-proxy approach in Chapters 5 and 6. In Chapter 5, I have focused on the downstream evolution of the detrital age sig- 7 nal in the Brahmaputra River in the eastern Himalayan foreland. The Brahmaputra drains the entire eastern Himalaya and, in particular, the Namche Barwa syntaxis. The rapidly exhuming Namche Barwa syntaxis produces a distinctive young (< 3 Ma) detrital age signal, making the Brahmaputra an ideal case to study dilution effects on isotope provenance interpretations in a large drainage system. Our results show that the young age peak distinctive of sediment derived from the syntaxis is present in all samples, in both the mica and zircon datasets. However, we show that whilst the diag- nostic young zircon peak endures up to a thousand kilometers downstream, the young mica population becomes heavily diluted, presumably either due to hydrodynamic and transport effects or due to input from tributaries draining the Higher Himalaya. In Chapter 5 we have mapped the patterns of present-day erosion in the eastern Himalaya. The inversion modeling shows high erosion rates in basins located south of the Namche Barwa syntaxis, whereas signifi cantly slower erosion rates are predicted in rivers draining the Tibetan Plateau and Lohit plutonic suite, which are character-

159 Synthesis and conclusions ized by age peaks of ~ >20 Ma in both the zircon fi ssion-track and muscovite 40Ar/39Ar systems. This approach defi nes consistent trends in regional 40Ar/39Ar muscovite and zircon fi ssion-track ages that can be used to better understand the exhumation history of the eastern Himalaya and to infer the spatial variability of the present-day erosion. Furthermore, in Chapter 5 erosion estimates from the mixing model are compared with a quantitative estimate of steady-state exhumation rates required to produce the observed major peaks of age distributions in the different samples. The comparison suggests erosion rates of ~0.2 mm/yr in the catchments draining into the Yarlung- Tsangpo before it enters the mountain range, 2-4 mm/yr within the Namche Barwa massif, and 0.5-1 mm/yr on the southern Himalayan fl ank. While previous studies have highlighted that detrital signal of the eastern Himalaya is dominated by young input from the Namche Barwa syntaxis, in Chapter 5 we have shown that the down- stream evolution of that signal presents a more complex behavior and that the dilution can bias interpretations on the cooling history of the area using the syn-depositional sediments of the foreland. The dynamics of comminution of the youngest muscovite grains 1000 km from the source it was however not clear. I have addressed this issue by looking at the effect of grain-size variability on the age distribution in Chapter 6. There, I have used the last generation multi-collector noble gas mass spectrometer technology (Helix MC plus) to analyze muscovite grains of different sizes, from 1000 mm down to 120 mm. Using the 1013 Ohm amplifi er on H2-H1/Faraday cups, we were able to improve the precision of the ages of the sanidine standards DRA by a factor of ~2 (in comparison with the 1012 Ohm setting). Measurements using the 1013 Ohm amplifi er showed an increase in the precision of the muscovite ages of a factor of ~5 for grain sizes rang- ing from 250-1000 mm and of ~2 for the grain size < 250 mm compared with the previous generation of mass spectrometer used in earlier studies. Thus, using the 1013 Ohm amplifi er we are able to detect small signal intensities of < 10 fA for the 39Ar ion beams for fractions < 250 mm. I have shown that grain-size variability can potentially bias the distribution of age peaks using fi ve modern river-sediment samples collected at increasing distances from the Namche Barwa syntaxis, along the Brahmaputra River in the Himalayan foreland. I have analyzed fi ve grain-size intervals for each sample and I have shown that the ages obtained from analysis from a narrow grain-size interval of 400-500 mm can bias the age distributions towards monotonic age peaks. The new results support- ed the hypothesis that grain-size variability can affect age distributions. I concluded that for the analysis of regional-scale areas a multiple grain-size fractions analysis should be the preferred methodology. This improvement in the methodology and our

160 Chapter 7 initial testing of its effects has important implications for future analyses of distal foreland river sand provenance ages. In summary, I tested the consistency of the detrital 40Ar/39Ar dating approach as a tool to characterize the tectonic history of source rocks within the river network of an evolving mountain range. This approach was initially tested in the Eastern Alps. In the following of the work, I analyzed the modern river sediments of the Tsangpo- Siang-Brahmaputra river and of its major tributaries using two different thermochro- nometers (muscovite 40Ar/39Ar and zircon fi ssion-tracks). Understanding the evolu- tion of the eastern Himalayan syntaxis is key to differentiating models of coupling between tectonics and erosion. The multi-proxy approach allowed to produce a syn- optic cooling-age map of the eastern Himalaya that highlighted the spatial variation in exhumation rates of the contributing sources to the fl uvial system. Using the mixing model approach, I inferred the present-day erosion patterns of the major river’s catch- ments of the area. The relative present-day erosion estimates were then compared with quantitative estimate of steady-state exhumation rates required to produce major age components observed in the detrital samples. This comparison suggested erosion rates of ~4 mm/yr for the catchments located south of the Namche Barwa syntaxis from the Siang river. Our estimates are in line with previous rates derived in the prox- imity of the Namche Barwa syntaxis (Burg et al., 1998; Finnegan et al., 2008), but slightly lower if compared with erosion rates of ~10-17 mm/yr derived from detrital zircon fi ssion-track data analysis from similar position in the Siang river (Enkelmann et al., 2011; Stewart et al., 2008). We noticed that whilst the young age peak is distinctive for the studied minerals 7 and endures many kilometers downstream, the young mica population is much more suppressed, both in proximal and distal samples. The potential effect of dilution of the analyzed target minerals has been addressed by looking at different grain-size frac- tions. This work has shown that grain-size variability can bias age distributions when studying large catchment areas, such as the Brahmaputra foreland. We show that for extended catchment areas, multiple grain-size analysis allowed to have a better reso- lution of the sources drained in the catchment area. In particular, we showed that for > 1000 km downstream the Namche Barwa syntaxis the analysis of grains < 250 micrometers allowed to detect the young < 3 Ma component that has been partially missed by looking at grain size > 250 micrometers. This thesis has explored the exhumation patterns of two dynamically evolving mountain ranges characterized by two distinct spatio-temporal evolutions. I have shown how the linear mixing model approach can be used for deriving fi rst-order in- formation on the erosion and cooling patterns within a large catchment area. Despite

161 Synthesis and conclusions large errors, the mixing model approach allowed mapping the spatial variability and temporal resolution of exhumation rates. This thesis also demonstrates that a combi- nation of multi-proxy thermochronology, numerical modeling, and analytical tech- nique improvement provides new opportunities to study the evolution of the transient response of mountain belts to changes in boundary conditions on geological (Ma) timescales.

162 Chapter 7 Bibliography

163 Bibliography Adams, B.A., Hodges, K.V., Whipple, K.X., Ehlers, T.A., van Soest, M.C., Wartho, J., (2015). Constraints on the tectonic and landscape evolution of the Bhutan Himalaya from thermochronometry. Tectonics 34, 1329–1347, doi:10.1002/2015TC003853. Adlakha, V., Lang, K.A., Patel, R.C., Lal, N., Huntington, K.W., (2013). Rapid long- term erosion in the rain shadow of the Shillong Plateau, Eastern Himalaya. Tec- tonophysics 582, 76–83, doi:10.1016/j.tecto.2012.09.022. Aldrich L.T., Nier A.O., (1948). Argon 40 in potassium minerals. Phys Rev 74:876- 877. Amstrong, R.L., E. Jager, , P., Eberhardt, (1966). A comparison of K-Ar and Rb-Sr ages on alpine biotite. Earth and Planet. Sci. Lett., Vol. 1, 13-19. Armitage, J.J., Duller, R.A., Whittaker, A.C., Allen, P.A., (2011). Transformation of tectonic and climatic signals from source to sedimentary archive. Nature Geos- ci. 4, 231–235, doi:10.1038/ngeo1087. Batt, G. E., and M. T. Brandon, (2002). Lateral thinking: 2-D interpretation of ther- mochronology in convergent orogenic settings, Tectonophysics 349 , 349, 185– 201. Beaumont C., Jamieson R. A., Nguyen M. H., Lee B., (2001). Himalayan tectonics explained by extrusion of a low-viscosity crustal channel coupled to focused surface denudation. Nature, 414, 738-742. doi :10.1038/414738a. Bendick, R., Ehlers, T.A., (2014). Extreme localized exhumation at syntaxes initiated by subduction geometry. Geophys. Res. Lett. 41, 5861–5867, doi:10.1002/2014GL061026. Berger, A., and R. Bousquet, (2008). Subductionrelated metamorphism in the Alps: Review of isotopic ages based on petrology and their geodynamic consequenc- es, in Tectonic Aspects of the Alpine-Dinaride-Carpathian System, edited by S. Siegesmund,B. Fugenschuh, and N. Froitzheim, Geol. Soc. Spec. Publ., 298, 117 – 144, doi:10.1144/SP298.7.

Berger, G.W., York, D., (1981). Geothermometry from 40Ar/39Ar dating experiments. Geochim Cosmochim Acta 45:795-811. Bernet, M., and C. Spiegel, (2004). Introduction: detrital thermochronology, Detrital thermochronology—Provenance Anal. Exhum. Landsc. Evol. Mt. belts, 378, 1–6, doi:10.1130/0-8137-2378-7.1. Bernet, M., Brandon, M. T., Garver, J. I. and Molitor, B. (2004b). Downstream changes of Alpine zircon fi ssion-track ages in the Rhône and Rhine rivers, J. Sediment. Res., 74(1). 82–94, doi:10.1306/041003740082. Bernet M., Garver J.I., (2005). Fission-track analysis of detrital zircon. Rev Mineral Geochem 58:205-238.

164 Bernet, M., van der Beek, P., Pik, R., Huyghe, P., Mugnier, J.-L., Labrin, E., Szulc, A., (2006). Miocene to Recent exhumation of the central Himalaya deter- mined from combined detrital zircon fi ssion-track and U/Pb analysis of Si- walik sediments, western Nepal. Basin Res. 18, 393–412, doi:10.1111/j.1365- 2117.2006.00303.x. Bernet, M., Brandon, M., Garver, J., Balestieri, M.L., Ventura, B., Zattin, M., (2009). Exhuming the Alps through time: Clues from detrital zircon fi ssion-track thermo- chronology. Basin Res. 21(6). 781–798, doi:10.1111/j.1365-2117.2009.00400.x Bertrand, A., C. Rosenberg, A. Rabaute, F. Herman, and B. Fügenschuh (2017). Ex- humation mechanisms of the Tauern Window (Eastern Alps) inferred from apatite and zircon fi ssion track thermochronology, Tectonics, 36, 207–228, doi:10.1002/2016TC004133. Bigi, G., Castellarin, A., Coli, M., Dal Piaz, GV, Sartori, R., Scandone, P. and Vai, GB,(1990). Structural Model of Italy, sheets 1-2: CNR, Progetto fi nalizzato Geodinamica. Biswas, S., Coutand, I., Grujic, D., Hager, C., Stockli, D., Grasemann, B., (2007). Exhumation and uplift of the Shillong plateau and its infl uence on the eastern Himalayas: New constraints from apatite and zircon (U-Th-[Sm])/He and apa- tite fi ssion track analyses. Tectonics 26, TC6013, doi:10.1029/2007TC002125. Booth, A.L., Chamberlain, C.P., Kidd, W.S.F., Zeitler, P.K., (2009). Constraints on the metamorphic evolution of the eastern Himalayan syntaxis from geochronologic and petrologic studies of NB. Geol. Soc. Am. Bull. 121, 385–407, doi:10.1130/ B26041.1. Bracciali, L., Najman, Y., Parrish, R.R., Akhter, S.H., Millar, I., (2015). The Brahma- putra tale of tectonics and erosion: Early Miocene river capture in the Eastern Himalaya. Earth Planet. Sci. Lett. 415, 25–37, doi:10.1016/j.epsl.2015.01.022. Bracciali, L., Parrish, R. R., Najman, Y., Smye, A., Carter, A., and J. R. Wijbrans, (2016). Plio-Pleistocene exhumation of the eastern Himalayan syntaxis and its domal “pop-up,” Earth-Science Rev., 160, 350–385, doi:10.1016/j.earsci- rev.2016.07.010. Brandon M.T., Vance J.A., (1992). Fission track ages of detrital zircons: implications for the tectonic evolution of the Cenozoic Olympic sub- duction complex. Am. J. Sci. 292:565–636. Brandon, M.T., (2002). Decomposition of mixed grain age distributions using Binom- fi t, On Track, Vol. 24, p. 13-18. Braun, J., (2002). Estimating exhumation rate and relief evolution by spectral analy- sis of age-elevation datasets, Terra Nov., 14(3). 210–214, doi:10.1046/j.1365- 3121.2002.00409.x. Braun, J., van der Beek, P., Batt, G., (2006). Quantitative thermochronology: numeri- cal methods for the interpretation of thermochronological data. Cambridge University Press, Cambridge, 271 pp. Braun, J., (2016). Strong imprint of past orogenic events on the thermochrono- logical record, Tectonophysics 683, 325–332, https://doi.org/10.1016/j.tec- to.2016.05.046. Braun, J., Gemignani, L., and van der Beek, P.: Extracting information on the spa- tial variability in erosion rate stored in detrital cooling age distributions in river sands, (2017). Earth Surf. Dynam. Discuss., https://doi.org/10.5194/es- urf-2017-42. Brewer, I. D., D. W. Burbank, and K. V. Hodges, (2003). Modelling detrital cooling- age populations: Insights from two Himalayan catchments, Basin Res., 15(3). 305–320, doi:10.1046/j.1365-2117.2003.00211.x. Brewer, I.D., Burbank, D.W., Hodges, K. V, (2006a). Downstream development of a detrital cooling-age signal: Insights from 40Ar/39Ar muscovite thermochronol- ogy in the Nepalese Himalaya. Spec. Pap. Geol. Soc. Am. 2398(20). 321–338. doi:10.1130/2006.2398(20). Brookfi eld, M.E., (1998). The evolution of the great river systems of southern Asia during the Cenozoic India-Asia collision: rivers draining southwards. Geomor- phology 22, 285–312, doi:10.1016/S0169-555X(97)00082-2. Brown, R. (1991). Backstacking apatite _ssion track “stratigraphy”: a method for re- solving the erosional and isostatic rebound components of tectonic uplift histo- ries. Geology, 19:74-77.

Burbank D.W., (1992). Causes of recent Himalayan uplift deduced from deposited patterns in the Ganges basin. Nature 357, 680 – 683. doi:10.1038/357680a0. Burg, J.P., Nievergelt, P., Oberli, F., Seward, D., (1998). The NB syntaxis: evidence for exhumation related to compressional crustal folding. J. Asian Earth Sci. 16, 239–252, doi:10.1016/S0743-9547(98)00002-6. Carrapa, B., A. Di Giulio, and J. Wijbrans (2004). The early stages of the Alpine collision: An image derived from the upper Eocene-lower Oligocene record in the Alps-Apennines junction area, Sediment. Geol., doi:10.1016/j.sed- geo.2004.05.015. Carrapa, B., (2009). Tracing exhumation and orogenic wedge dynamics in the Eu- ropean Alps with detrital thermochronology. Geology 37(12). 1127–1130. doi:10.1130/G30065A.1 Carrapa, B., Orme, D.A., DeCelles, P.G., Kapp, P., Cosca, M.A., Waldrip, R., (2014). Miocene burial and exhumation of the India-Asia collision zone in southern Ti- bet: Response to slab dynamics and erosion. Geology 42, 443–446, doi:10.1130/ G35350.1. Carrapa, B., Di Giulio, A., Mancin, N., Stockli, D., Fantoni, R., Hughes, A., Gupta,

166 S., (2016). Tectonic signifi cance of Cenozoic exhumation and foreland basin evolution in the Western Alps. Tectonics, 35. Doi:10.1002/2016TC004132 Cerveny, P.F., Naeser, N.D., Zeitler, P.K., Naeser, C.W. & Johnson, N.M., (1988). History of uplift and relief of the Himalaya during the past 18 million years; evidence from sandstones of the Siwalik Group. In: New Perspectives in Basin Analysis (Ed. by K.L. Kleinspehn & C. Paola). pp. 43-61. NewYork: Springer. Challandes, N., Marquer, D., Villa, I.M., (2003). Dating the evolution of C-S mi- crostructures: A combined 40Ar/39Ar step-heating and UV laserprobe analysis of the Alpine Roffna shear zone. Chem. Geol. 197(1-4). 3–19. doi:10.1016/ S0009-2541(02)00354-6 Champagnac, J.D., van der Beek, P., Diraison, G., Dauphin, S., (2008). Flexural iso- static response of the Alps to increased Quaternary erosion recorded by foreland basin remnants, SE France. Terra Nov. 20(3). 213–220, doi:10.1111/j.1365- 3121.2008.00809.x. Chirouze, F., Huyghe, P., Chauvel, C., van der Beek, P., Bernet, M., Mugnier, J.-L., (2015). Stable drainage pattern and variable exhumation in the western Hima- laya since the middle Miocene. J. Geol. 123, 1–20, doi:10.1086/679305. Chirouze, F., Huyghe, P., van der Beek, P., Chauvel, C., Chakraborty, T., Dupont- Nivet, G., Bernet, M., (2013). Tectonics, exhumation, and drainage evolution of the eastern Himalaya since 13 Ma from detrital geochemistry and thermo- chronology, Kameng River Section, Arunachal Pradesh. Geol. Soc. Am. Bull. 125, 523–538, doi:10.1130/B30697.1. Clark, M.K., Schoenbohm, L.M., Royden, L.H., Whipple, K.X., Burchfi el, B.C., Zhang, X., Tang, W., Wang, E., Chen, L., (2004). Surface uplift, tectonics, and erosion of eastern Tibet from large-scale drainage patterns. Tectonics 23, TC1006, doi:10.1029/2002TC001402. Clark, M.K., Bilham, R., (2008). Miocene rise of the Shillong Plateau and the begin- ning of the end for the Eastern Himalaya. Earth Planet. Sci. Lett. 269, 336–350, doi:10.1016/j.epsl.2008.01.045. Condon, D. J., and M. D. Schmitz (2013). One Hundred Years of Isotope Geochronol- ogy, and Counting, Elements, 9, 15–17, doi:10.2113/gselements.9.1.15. Copeland, P. & Harrison, M.T., (1990). Episodic rapid uplift in the Himalaya revealed by 40Ar/39Ar analysis of detrital K-feldspar and muscovite, Bengal fan. Geol- ogy, 18, 354^359.

Copeland, P., Bertrand, G., France-Lanord, C., Sundell, K., (2015). 40Ar/39Ar ages of muscovites from modern Himalayan rivers: Himalayan evolution and the relative contribution of tectonics and climate. Geosphere 11, 1837–1859, doi:10.1130/GES01154.1. Copeland, P., Harrison, T.M., Yun, P., Kidd, W., (1995). Thermal evolution of the

167 Bibliography Gangdese batholith, southern Tibet: A history of episodic unroofi ng. Tectonics 14, 223–236. Coutand, I., Whipp, D.M., Grujic, D., Bernet, M., Fellin, M.G., Bookhagen, B., Landry, K.R., Ghalley, S.K., Duncan, C., (2014). Geometry and kinematics of the Main Himalayan Thrust and Neogene crustal exhumation in the Bhutanese Himalaya derived from inversion of multithermochronologic data. J. Geophys. Res. 119, 1446–1481, doi:10.1002/2013JB010891. Dal Piaz, G. V., Bistacchi, A., Massironi, M., (2003). Geological outline of the Alps. Episodes 26(3). 175–180, doi:10.18814/epiiugs/2003/v26i3/62456. Dallmeyer, R. D., Handler, R., Neubauer, F., F.H., (1998). Sequence of Thrusting Within A Thick-Skinned Tectonic Wedge: Evidence From 40Ar/39Ar and Rb-Sr Ages from the Austroalpine Nappe Complex of the Eastern Alps. The journal of Geology 106(1). doi:10.1086/516008.

Darry, L.A., France-Lanord, C., (1996). Neogene Himalayan weathering history and river87Sr86Sr: impact on the marine Sr record, Earth Planet. Sci. Lett.s 142, 1-2. doi.org/10.1016/0012-821X(96)00091-X.

Deino, A. L., P. R. Renne, and C. C. Swisher, (1998). 40Ar/39Ar Dating in Paleoanthro- pology and archeology, Evol. Anthropol., 63–75. Dodson, M. H. (1973). Closure temperatures in cooling geological and petrological systems., Contrib. Miner. Pet., 40, 259–274. Efron, B., (1981). Nonparametric estimates of standard error: The jackknife, the boot- strap and other methods. Biometrika. 68 (3): 589–599. Engi, M., Lanari, P., Kohn, J., (2017). Signifi cant ages- An introduction to petrochro- nology. In: Petrochronology: Methods and Applications, review in Min. and Geochem., Geolog. Soc. Of America, Vol., 83.

Enkelmann, E., and T. A. Ehlers (2015). Evaluation of detrital thermochronology for quanti fi cation of glacial catchment denudation and sediment mixing, Chem. Geol., 411, 299–309, doi:10.1016/j.chemgeo.2015.07.018. Enkelmann, E., Ehlers, T.A., Zeitler, P.K., Hallet, B., (2011). Denudation of the Namche Barwa antiform, eastern Himalaya. Earth Planet. Sci. Lett. 307, 323– 333, doi:10.1016/j.epsl.2011.05.004. Farley, K.A., (2000). Helium diffusion from apatite: General behavior as illus- trated by Durango fl ourapatite: Journal of Geophysical Research, v. 105, p. 2903– 2914. Faure, G., (1986). Principles of Isotope Geology, John Wiley & Sons. Fin. Geodin- amica, SELCA Firenze. Finnegan, N.J., Hallet, B., Montgomery, D.R., Zeitler, P.K., Stone, J.O., Anders, A.M., Yuping, L., (2008). Coupling of rock uplift and river incision in the

168 Namche Barwa-Gyala Peri massif, Tibet. Geol. Soc. Am. Bull. 120, 142–155, doi:10.1130/B26224.1. Fitzgerald, P. G., and A. J. W. Gleadow (1990). New approaches in fi ssion track geo- chronology as a tectonic tool: examples from the transantarctic mountains, NucL Tracks Radiat. &leas. Int. J. Radiat. Appl. Instrum., Part D, 17(3). 351–357. Fleischer, R.L., Price, P.B., and Walker, R.M., (1975). Nuclear tracks in solids; Prin- ciples and applications: Berkeley, California, University of California Press, 605 pp. Fodor, L. I., A. Gerdes, I. Dunkl, B. Koroknai, Z. Pécskay, M. Trajanova, P. Horváth, M. Vrabec, B. Jelen, K. Balogh, W. Frisch,(2008). Miocene emplacement and rapid cooling of the Pohorje pluton at the Alpine-Pannonian-Dinaridic junction, Slovenia, Swiss J. Geosci., 101, 255 – 271 doi:10.1007/s00015-008-1286-9. Fox, M., Herman, F., Willett, S.D., Schmid, S.M., (2016). The exhumation history of the European alps inferred from linear inversion of thermochronometric data. Am. J. Sci. 316(43). 505–541. doi:10.2475/06.2016.01. Frisch, W., Gawlick, H.-J., (2003). The nappe structure of the central Northern Calcar- eous Alps and its disintegration during Miocene tectonic extrusion - a contribu- tion to understanding the orogenic evolution of the Eastern Alps. Int. J. Earth Sci. (Geologische Rundschau) 92(5). 712–727. doi:10.1007/s00531-003-0357- 4. Frisch, W., Kuhlemann, J., Dunkl, I., Brügel, A., (1998). Palinspastic reconstruction and topographic evolution of the Eastern Alps during late Tertiary tectonic ex- trusion. Tectonophysics 297, 1–15. doi:10.1016/S0040-1951(98)00160-7. Froitzheim, N., Schmid, S.M., Conti, P., (1994). Repeated change from crustal short- ening to orogen-parallel extension in the Austroalpine units of Graubünden. Eclogae Geol. Helv. 87(2). 561–607. doi:10.5169/seals-167471. Fugenschuh, B., Seward, D., Mancktelow, N., (1997). Exhumation in a convergent orogen?: the western Tauern window. Terra Nov. 9(5-6). 213–217. Galbraith, R. F., and P. F. Green (1990). Estimating the component ages in a fi - nite mixture, Int. J. Radiat. Appl. Instrumentation. Part, 17(3). 197–206, doi:10.1016/1359-0189(90)90035- Gallagher, K., R. Brown, and C. Johnson (1998). Fission track analysis and its ap- plications to geological problems, Annu. Rev. Earth Pl. Sci., 26, 519–572, doi:10.1146/annurev.earth.26.1.519. Garver, J. I., (2004). Fission-track Laboratory Procedures at Union College. Union. Garver, J. I., M. T. Brandon, M. Roden-Tice, and P. J. J. Kamp (1999). Exhumation history of orogenic highlands determined by detrital fi ssion-track thermochro- nology, Geol. Soc. London, Spec. Publ., 154(1). 283–304, doi:10.1144/GSL. SP.1999.154.01.13.

169 Bibliography Garver, J.I., Brandon, M.T., Roden-tice, M., Kamp, P.J.J., (1999). Exhumation history of orogenic highlands determined by detrital fi ssion-track thermochronology. Geol. Soc. London, Spec. Publ. 154, 283–304. Garzanti, E., Andò, S., (2007). Heavy mineral concentration in modern sands: Im- plications for provenance interpretation, in: Heavy Minerals in Use, Develop- ments in Sedimentology, 58, 517–545. Garzanti, E., Resentini, A., Andò, S., Vezzoli, G., Pereira, A., Vermeesch, P., (2015). Physical controls on sand composition and relative durability of detrital miner- als during ultra-long distance littoral and aeolian transport (Namibia and south- ern Angola). Sedimentology 62, 971–996, doi:10.1111/sed.12169. Garzanti, E., Vezzoli, G., Andò, S., France-Lanord, C., Singh, S.K., Foster, G., (2004). Sand petrology and focused erosion in collision orogens: the Brahmaputra case. Earth Planet. Sci. Lett.s 220, 157–174, doi:10.1016/S0012-821X(04)00035-4.

Gemignani L., Xilin Sun, Braun, J., Wijbrans, J., 2017. A new detrital mica 40Ar/39Ar dating approach for provenance and exhumation of the Eastern Alps. Tectonics 36, no. 8 (2017): 1521-1537. doi/10.1002/2017TC004483/ Gleadow, A.J.W., and C. K. Brooks (1979). Fission Track Dating, Thermal Histories and Tectonics of Igneous Intrusions in East Greenland, Contrib. Miner. Pet., 71, 45–60. Gleadow, A.J.W., (1981). Fission track dating methods: what are the real alternatives? Nuclear Tracks and Radiation Measurements, v. 5, p. 3–14. Glotzbach, C., Bernet, M., van der Beek, P., (2011). Detrital thermochronology re- cords changing source areas and steady exhumation in the Western European Alps. Geology 39(3). 239–242. doi:10.1130/G31757.1. Govin, G., Najman, Y., van der Beek, P., Huyghe, P., Millar, I., Bernet, M., Dupont- Nivet, G., Wijbrans, J., Gemignani, L., Vögeli, N., (2016). Constraining the timing of exhumation of the Eastern Himalayan syntaxis from a study of the palaeo-Brahmaputra deposits, Siwalik Group, Arunachal Pradesh, India. Paper presented at the 2016 Himalayan-Karakorum-Tibet Workshop, Aussois, France, 9-12 May 2016. Gradstein, F. M., Ogg, J. G., (2012). The Chronostratigraphic Scale. In: Gradstein et al., The Geologic Time Scale 2012. Elsevier Publ. Co. Grujic, D., Coutand, I., Bookhagen, B., Bonnet, S., Blythe, A., Duncan, C., (2006). Climatic forcing of erosion, landscape, and tectonics in the Bhutan Himalayas. Geology 34, 801–14, doi:10.1130/G22648.1. Hallet, B., Molnar, P., (2001). Distorted drainage basins as markers of crust- al strain east of the Himalaya. J. Geophys. Res. 106, 13697–13709, doi:10.1029/2000JB900335.. Hames, W.E., Bowring, S.A., (1994). An empirical evaluation of the argon diffusion

170 geometry in muscovite. Earth and Planet. Sci. Lett.s 124, 161-169. Handy, M.R., Herwegh M., Kamber, B.S., Tietz, R., V.I.M., (1996). Geochronologic, petrologic and kinematic constraints on the evolution of the Err-Platta bound- ary, part of fossil continent-ocean suture in the Alps (Eastern Swizzerland). Schweiz. Miner. Petrogr. Mitt 76, 453–474. Handy, M.R., Schmid S. M., Bousquet, R., Kissling, E., Bernoulli, D., (2010). Recon- ciling plate-tectonic reconstructions of Alpine Tethys with the geological–geo- physical record of spreading and subduction in the Alps. Earth-Sci. Rew., 102, 121-158. doi:10.1016/j.earscirev.2010.06.002 Harland, W.B., Smith, A.G. & Wilcock, B., (1964). The Phanerozoic time-scale. Geol. Soc. Land., Quart. J., 120S, 1-458. Harrison, T.M., Célérier, J., Aikman, A.B., Hermann, Herman, J., Heizler, M.T., (2009). Diffusion of 40Ar in muscovite. Geochim. Cosmochim. Acta 73(4). 1039–1051. doi:10.1016/j.gca.2008.09.038. Herman F. , Seward D., Valla P. G., Carter A., Kohn B., Willett S.D., Ehlers T. (2013). Worldwide acceleration of mountain erosion under a cooling climate. Nature 504, 423–426, doi:10.1038/nature12877

Hodges, K.V., Ruhl, K.W., Wobus, C.W., Pringle, M.S., (2005). 40Ar/39Ar Thermo- chronology of Detrital Minerals, , 58, 239–257, doi:10.2138/rmg.2005.58.9. Hodges, K.V. (2005). 40Ar/39Ar Thermochronology of Detrital Minerals, Rev. Min- eral. Geochemistry, 58(1). 239–257, doi:10.2138/rmg.2005.58.9. Hodges, K.V., (2000). Tectonics of the Himalaya and southern Tibet from two perspectives. Geol. Soc. Am. Bull. 112, 324–350, doi:10.1130/0016- 7606(2000)112<324:TOTHAS>2.0.CO;2. Holmes, A., (1911). The association of lead with uranium in rock-minerals and its application to the measurement of geological time. Proceedings of the Royal Society of London 85: 248–256. Huntington, K.W., Blythe, A.E., Hodges, K.V., (2006). Climate change and Late Plio- cene acceleration of erosion in the Himalaya, Earth and Planet. Sci. Lett.s, 252, 107–118. Hunziker, J.C., Desmons, J., Hurford A.J., (1992). Thirty-two years of geochronologi- cal work in Central and Western Alps: a review on seven maps. Memor.de Geol. (Lausanne).,No. 13, ISSN 1015-3578. Hurford A.J., Green, P.F., (1983). The zeta age calibration of fi ssion-track dating. Chem. Geol., 188: 1-22 Hurford, A.J., and Carter, A., (1991). The role of fi ssion track dating in discrimination of provenance, in Morton, A.C., Todd, S.P., and Haughton, P.D., W., eds., De- velopments in sedimentary provenance studies: Geological Society [London],

171 Bibliography Special Publication v. 57, p. 67–78. Hurley P.M., (1954). The helium age method and the distribution and migration of helium in rocks. In: Nuclear Geology, Wiley & Sons, 301-329. Kapp, P., DeCelles, P.G., Leier, A.L., Fabijanic, J.M., He, S., Pullen, A., Gehrels, G.E., Ding, L., (2007). The Gangdese retroarc thrust belt revealed. GSA Today 17, 4–6, doi:10.1130/GSAT01707A.1. Kellett, D., Grujic, A., Coutand, I., Cottle, J., Mukul, M., (2013). The South Tibetan detachment system facilitates ultra-rapid cooling of granulite-facies rocks in Sikkim Himalaya, Tectonics, 32(2). 252–270, doi:10.1002/tect.20014. Kelley, S. (2002). Excess argon in K – Ar and Ar – Ar geochronology, , 188, 1–22, PII: S0009-2541(02)00064-5. King, G.E., Herman, F., Guralnik, B., (2016). Northward migration of the eastern Hi- malayan syntaxis revealed by OSL thermochronometry. Science 353, 800–804, doi:10.1126/science.aaf2637.

Kim, J., Su-in Jeon, (2015). 40Ar/39Ar age determination using ARGUS VI multiple- collector noble gas mass spectrometer: performance and its application to geo- sciences, Journal of Analytical Science and Technology 6:4, DOI 10.1186/ s40543-015-0049-2 Koons, P. O., Zeitler, P.K, Hallet, B., (2013). in Treatise on Geomorphology ed. Shro- der J. F. 318–349 Academic. Koornneef, J.M., Bouman, C., Schwieters, J.B., Davies, G.R., (2014). Measurement of small ion beams by thermal ionisation mass spectrometry using new 1013 Ohm resistors. Anal. Chim. Acta. 819, 49–55. Koornneef, J.M., Nikogosian, I., van Bergen, M.J., Smeets, R., Bouman, C., Davies, G.R., (2015). TIMS analysis of Sr and Nd isotopes in melt inclusions from Ital- ian potassium-rich lavas using prototype 1013 Ω amplifi ers, Chemical Geology 397 14–23. Koornneef, J. M., Gress, M.U., Chinn, I.L., Jelsma, H.A., Harris, J.W., Davies, G.R., (2017). Archaean and Proterozoic diamond growth from contrasting styles of large-scale magmatism. Nature commun. 8:648, DOI: 10.1038/s41467-017- 00564-x. Koppers, A.A.P., (2002). ArArCALC-software for 40Ar/39Ar age calculations. Com- put. Geosci. 28(5). 605–619. doi:10.1016/S0098-3004(01)00095-4. Krummenacher, D., (1961). Determination d’age isotopique faites sur quelques roches de I’Himalayan du Nepal par la methode potassium-argon. Schweiz. Mineral. Petrogr. Mitt., 41: 273-283. Kuhlemann, J., Dunkl, I., Brugel, A., Spiegel, C., Frisch, W. (2006). From source terrains of the Eastern Alps to the Molasse Basin: Detrital record of non-

172 steady-state exhumation. Techtonoph., 413(3-4). 301-316, doi:10.1016/j.tec- to.2005.11.007. Kuiper, K. F., A. Deino, F. J. Hilgen, W. Krijgsman, P. R. Renne, and J. R. Wijbrans (2008b). Synchronizing Rock Clocks of Earth History, Science, 320(5875). 500–504, doi:10.1126/science.1154339. Kurz, W., Handler, R., Bertoldi, C., (2008). Tracing the exhumation of the eclogite zone (Tauern window, Eastern Alps) by 40Ar/39Ar dating of white mica in eclogites. Swiss J. Geosci. 101 (supl.1). 191–206. doi:10.1007/s00015-008- 1281-1. Lang, K. A., Huntington, K. W., Burmester, R., and Housen, B., (2016). Rapid exhu- mation of the eastern Himalayan syntaxis since the late Miocene. Geol. Soc. Am. Bull., 128:1403-1422.

Lanphere, M.A., Dalrymple, G.B., (1976). Identifi cation of Excess 40Ar by the 40Ar/39Ar age spectrum technique. Earth Planet. Sc.i Lett., 32:141-148. Lee, J. Y., K. Marti, J. P. Severinghaus, K. Kawamura, H. S. Yoo, J. B. Lee, and J. S. Kim, (2006). A redetermination of the isotopic abundances of atmo- spheric Ar, Geochim. Cosmochim. Acta, 70(17). 4507–4512, doi:10.1016/j. gca.2006.06.1563. Leloup, P.H., Mahéo, G., Arnaud, N., Kali, E., Boutonnet, E., Liu, D., Xiaohan, L., Haibing, L., (2010). The South Tibet detachment shear zone in the Dinggye area. Earth Planet. Sci. Lett. 292, 1–16, doi:10.1016/j.epsl.2009.12.035. Li, G., Kohn, B., Sandiford, M., Xu, Z., Tian, Y., Seiler, C., (2016). Synorogenic mor- photectonic evolution of the Gangdese batholith, South Tibet: Insights from low-temperature thermochronology. Geochem. Geophys. Geosyst. 17, 101– 112, doi:10.1002/2015gc006047. Li, G., Tian, Y., Kohn, B.P., Sandiford, M., Xu, Z., Cai, Z., (2015). Cenozoic low temperature cooling history of the Northern Tethyan Himalaya in Zedang, SE Tibet and its implications. Tectonophysics 643, 80–93, doi:10.1016/j.tec- to.2014.12.014. Liu, Yongjiang , Johan Genser, Robert Handler, G.F. and F.N., (2001). 40Ar/39Ar muscovite ages from the Penninic-Austroalpine plate boundary, Eastern Alps. Tectonics 20(4). 526-547. doi: 10.1029/2001TC900011. Luth, S.W., Willingshofer, E., (2008). Mapping of the post-collisional cooling history of the Eastern Alps. Swiss J. Geosci. 101 (2008) Supplement 1, S207–S223. DOI 10.1007/s00015-008-1294-9. Malusà, M. G., A. Resentini, and E. Garzanti (2016). Hydraulic sorting and mineral fertility bias in detrital geochronology, Gondwana Res., 31, 1–19, doi:10.1016/j. gr.2015.09.002. Mark, D.F., Barfod, D., Stuart, F. M., Imlach, J., (2009), The ARGUS multicollector noble

173 Bibliography gas mass spectrometer: Performance for 40Ar/39Ar geochronology, G3,Technical Brief, Volume 10, Number 2, Q0AA02, doi:10.1029/2009GC002643, ISSN: 1525-2027 McDougall, I., Harrison, T.M., (1999). Geochronology and Thermochronology by the 40Ar/39Ar method. Oxford University Press, New York, 212 p. Merrihue, C., and Turner, G., (1966). Potassium-argon dating by activation with fast neutrons, Journal of Geophysical Research, v. 78, p. 3216–3221. Misra, D. K. (2009). Litho-tectonic sequence and their regional correlation along the Lohit Dibang valleys, eastern Arunachal Pradesh, J. Geol. Soc. India, 73(2). 213–219, doi:10.1007/s12594-009-0077-x. Mitchell, J. G., (1968). The argon-40/argon-39 method for potassium-argon age deter- mination. Geochim. Cosmochim. Acta, 32, 781-790. Molnar, P., England P. (1990). Late Cenozoic uplift of mountain ranges and global climate change: Chicken or egg? Nature 346(6279). DOI: 10.1038/346029a0. Monster, M., (2016). Multi-Method paleointensity data of the geomagnetic fi eld dur- ing the past 500 kyrs from European volcanoes, Retrived from: Utrecht Studies in Earth Sciences. Sci., No. 117. IssN 2211-4335. Najman, Y.M.R., M. S. Pringle, M. R. W. Johnson, a. H. F. Robertson, and J. R. Wijbrans, (1997). Laser 40Ar/39Ar dating of single detrital muscovite grains from early foreland-basin sedimentary deposits in India: Implications for early Himalayan evolution, Geology, 25(6). 535–538, doi:10.1130/0091- 7613(1997)025<0535:LAADOS>2.3.CO;2. Najman, Y., Pringle, M., Godin, L, Oliver, G., (2001). Dating of the oldest continental sediments from the Himalayan foreland basin. Nature 410(6825):194-197. Najman, Y., Appel, E., Bou Dagher-Fadel, M., Bown, P., Carter, A., Garzanti, E., Go- din, L., Han, J., Liebke, U., Oliver, G., Parrish, R., Vezzoli, G., (2010). Timing of India-Asia collision: Geological, biostratigraphic, and palaeomagnetic con- straints. J. Geophys. Res. 115, B12416, doi:10.1029/2010JB007673. Neubauer, F., Dallmeyer, R.D., Dunkl, I., Schirnik, D., (1995). Late Cretaceous exhu- mation of the metamorphic Gleinalm dome, Eastern Alps: kinematics, cooling history and sedimentary response in a sinistral wrench corridor. Tectonophysics 242, 79–98. doi:10.1016/0040-1951(94)00154-2. Neubauer, F., Genser, J., Kurz, W. & Wang, X., (1999). Exhumation of the Tauern window, Eastern Alps. Phys. Chem. Earth Part A: Solid Earth and Geodesy, 24(8). 675–680. Nier, A.O., (1935). Evidence for existence of an isotope of potassium of mass 40. Phys. Rev., 48, 283-284. Pfi ffner, O.A., Ellis, S., (2001). Collision tectonics in the Swiss Alps: Insight from

174 geodynamic modeling. Tectonics 19(6). 1065–1094. Ratschbacher, L., Frisch, W., Linzer, H.-G., Merle, O., (1991). Lateral Extrusion in the Eastern Alps, Part 2: Structural analysis. Tectonics 10(2). 257–271. Ratschbacher, L., Dingeldey, C., Miller, C., Hacker, B.R., McWilliams, M.O., (2004). Formation, subduction, and exhumation of Penninic oceanic crust in the East- ern Alps: Time constraints from 40Ar/39Ar geochronology. Tectonophysics 394(3-4). 155–170. doi:10.1016/j.tecto.2004.08.003. Raymo M.E., Ruddimond W.F., (1992). Tectonic forcing of late Cenozoic climate, Nature, 356, 117-122. Reiners, P. W. (2005). Past, Present, and Future of Thermochronology, Rev. Mineral. Geochemistry, 58(1). 1–18, doi:10.2138/rmg.2005.58.1. Reiners, P. W., and M. T. Brandon (2006). Using Thermochronology To Understand Orogenic Erosion, Annu. Rev. Earth Planet. Sci., 34(1). 419–466, doi:10.1146/ annurev.earth.34.031405.125202. Reiter, W., Elfert, S., Glotzbach, C., Bernet, M., Spiegel, C., (2013). Relations be- tween denudation, glaciation, and sediment deposition: Implications from the Plio-Pleistocene Central Alps. Basin Res. 25, 659–674. doi:10.1111/bre.12023. Renne, P.R., Swisher, CC.C., Deino A.L., Karner, D.B., Owens T.L., De Paolo D. J., (1998). Intercalibration of standards, absolute ages and uncertainties in 40Ar/ 39Ar dating. Chemical Geology 145, 1 I7- I52. Rahl, J.M., T.A. Ehlers, B.A. van der Pluijm, (2007). Quantifying transient erosion of orogens with detrital thermochronology from syntectonic basin deposits. Earth Planetary Sci. Lett., 256, pp. 147-161. Robert, X., van der Beek, P., Braun, J., Perry, C., Mugnier, J.-L., (2011). Control of detachment geometry on lateral variations in exhumation rates in the Himalaya: Insights from low-temperature thermochronology and numerical modeling. J. Geophys. Res. 116, B05202, doi:10.1029/2010JB007893. Robl, J., K. Stüwe, S. Hergarten, and L. Evans, (2008). Extension during continental convergence in the Eastern Alps: The infl uence of orogeny scale strike slip faults, Geology, 36(12). 963–966, doi:10.1130/G25294A.1. Rosenberg, L.C., Berger, A., (2009). On the causes and modes of exhumation and later- al growth of the Alps. Tectonics, vol. 28, TC6001, doi:10.1029/2008TC002442. Ruhl, K.W., Hodges, K. V., (2005). The use of detrital mineral cooling ages to evalu- ate steady state assumptions in active orogens: An example from the central Nepalese Himalaya. Tectonics 24(4). 1–14. doi:10.1029/2004TC001712. Rutherford, E., Soddy, F., (1902). The Cause and Nature of Radioactivity I, II, Philo- sophical Magazine IV, 370–96, 569–85.

175 Bibliography Salvi, D., Mathew, G., Kohn, B., (2016). Rapid exhumation of the upper Siang Valley, Arunachal Himalaya since the Pliocene. Geomorphology, in press, doi:10.1016/j. geomorph.2016.09.032.

Santato, A. Hamilton, D., Bouman, C., Wijbrans, J.R., (2015). New Design 1013 Ω Amplifi ers for the Analysis of Noble Gases. Thermo Fisher Scientifi c Inc. Tech- nical Note 30304 Scharf, A., Handy, M.R., Favaro, S., Schmid, S.M., Bertrand, A., (2013). Modes of orogen-parallel stretching and extensional exhumation in response to micro- plate indentation and roll-back subduction (Tauern Window, Eastern Alps). Int. J. Earth Sci. 102(6). 1627–1654. doi:10.1007/s00531-013-0894-4. Schmid, S.M., Fügenschuh, B., Kissling, E., Schuster, R., (2004). Tectonic map and overall architecture of the Alpine orogen. Eclogae Geol. Helv. 97(1). 93–117. doi:10.1007/s00015-004-1113-x. Schmid, S.M., Bernoulli, D., Fügenschuh, B., Matenco, L., Schefer, S., Schuster, R., Tischler, M., Ustaszewski, K., (2008). The Alpine-Carpathian-Dinaridic oro- genic system: Correlation and evolution of tectonic units. Swiss J. Geosci. 101, 139–183. doi:10.1007/s00015-008-1247-3. Schmid, S.M., Scharf, A., Handy, M.R., Rosenberg, C.L., (2013). The Tauern Win- dow (Eastern Alps, Austria): A new tectonic map, with cross-sections and a tectonometamorphic synthesis. Swiss J. Geosci. 106(1). 1–32. doi:10.1007/ s00015-013-0123-y. Schmitz, M. D., and K. F. Kuiper (2013). High-Precision Geochronology, Elements, (9). 25-30, doi:10.2113/gselements.9.1.25. Schneider, S., Hammerschmidt, K., Rosenberg, C.,L., (2013). Dating the longevity of ductile shear zones: Insight from 40Ar/39Ar in situ analyses, Earth and Planetary Sci. Let., Vol. 369-370, 43-58, http://dx.doi.org/10.1016/j.epsl.2013.03.002i. Schultz, M.H., Hodges, K.V., Ehlers, T.A., van Soest, M., Wartho, J.-A., (2017). Ther- mochronologic constraints on the slip history of the South Tibetan detachment system in the Everest region, southern Tibet. Earth Planet. Sci. Lett. 1–13, doi:10.1016/j.epsl.2016.11.022. Seward, D., Burg, J.-P., (2008). Growth of the NB Syntaxis and associated evolution of the Tsangpo Gorge: Constraints from structural and thermochronological data. Tectonophysics 451, 282–289, doi:10.1016/j.tecto.2007.11.057. Shen, T., Wang, G., Leloup, P.H., van der Beek, P., Bernet, M., (2016). Controls on Cenozoic exhumation of the Tethyan Himalaya from fi ssion-track thermo- chronology and detrital zircon U-Pb geochronology in the Gyirong basin area, southern Tibet. Tectonics 35, 1713–1734, doi:10.1002/2016TC004149. Singh, S.K., (1993). Geology and tectonics of the eastern syntaxial bend, Arunachal Himalaya. J. Himal. Geol. 4, 149-163.

176 Singh, S.K., France-Lanord, C., (2002). Tracing the distribution of erosion in the Brahmaputra watershed from isotopic compositions of stream sediments. Earth Planet. Sci. Lett. 202, 645–662, doi:10.1016/S0012-821X(02)00822-1. Soddy, F., (1913). Intra-atomic charge, Nature, 92: 399–400. Spiegel, C., Kuhlemann, J., Dunkl, I., Frisch, W., Von Eynatten, H., Balogh, K., (2000). The erosion history of the Central Alps: evidence from zircon fi ssion track data of the foreland basin sediments. Terra Nov., 12(4). 163-170. doi: 10.1046/j.1365-3121.2000.00289.x. Spiegel, C., Siebel, W., Kuhlemann, J., Frisch, W., (2004). Toward a comprehensive provenance analysis: A multi-method approach and its implications for the evo- lution of the Central Alps. Geol. Soc. of Am. 378, 37-50. doi: 10.1130/0-8137- 2378-7.37. Stampfl i, G.M., Mosar, J., Marquer, D., Marchant, R., Baudin, T., Borel, G., (1998). Subduction and obduction processes in the Swiss Alps. Tectonophysics 296, 159–204. doi:10.1016/S0040-1951(98)00142-5. Stewart, R.J., Brandon, M.T., (2004). Detrital-zircon fi ssion-track ages for the “Hoh Formation”: Implications for late Cenozoic evolution of the Cascadia subduc- tion wedge. Geol. Soc. Am. Bull. 116, 60–16, doi:10.1130/B22101.1. Stewart, R.J., Hallet, B., Zeitler, P.K., Malloy, M.A., Allen, C.M., Trippett, D., (2008). Brahmaputra sediment fl ux dominated by highly localized rapid erosion from the easternmost Himalaya. Geology 36, 711–39, doi:10.1130/G24890A.1. Stock, G. M., Ehlers, T. A., and Farley, K. A., (2006). Where does sediment come from? Quantifying catchment erosion with detrital apatite (U-Th)/He thermo- chronometry, Geology, 34(9):725-728.20. doi:10.1130/G22592.1. Stockli, D. F., K. A. Farley, and T. A. Dumitru (2000). Calibration of the apatite (U- Th)/He thermochronometer on an exhumed fault block, White Mountains, Cal- ifornia, Geology, (11). 983–986.

Stüwe, K., Foster, D., (2001). 40Ar/39Ar, pressure, temperature and fi ssion track con- straints on the age and nature of metamorphism around the main central thrust in the eastern Bhutan Himalaya. J. Asian Earth Sci. 19, 85–95, doi:10.1016/ S1367-9120(00)00018-3. Surian, N., (2002). Downstream variation in grain size along an Alpine river: analysis of controls and processes. Geomorphology 43, 137 – 149. Timmerman, S., Koornneef, J. M., Chinn, I. L. & Davies, G. R., (2017). Dated eclog- itic diamond growth zones reveal variable recycling of crustal carbon through time. Earth Planet Sci. Lett. 463, 178–188 x. Turner, G., (1968). the distribution of potassium and Argon in chondrites. In: L.H. Ahrens, ed., Origin and distribution of the elements, 387-398. Pergamum Press, London, 1178 p.

177 Bibliography Van der Beek, P., M. A. Summerfi eld, J. Braun, R. W. Brown, and A. Fleming (2002). Modeling post breakup landscape development and denudational history across the southeast African (Drakensberg Escarpment) margin, J. Geophys. Res., 107(B12). 2351, doi:10.1029/2001JB000744. Vance D., and Harris N. (1999). Timing of prograde metamorphism in the Zanskar Himalaya, Geology, 27(5). 395–398. Vermeesch, P., (2004). How many grains are needed for a provenance study? Earth Planet. Sci. Lett., 224:441-451. Vermeesch, P. (2007). Quantitative geomorphology of the White Mountains (Cali- fornia) using detrital apatite fi ssion track thermochronology, J. Geophys. Res. Earth Surf., 112(3). 1–11, doi:10.1029/2006JF000671. Vermeesch, P. (2012). On the visualisation of detrital age distributions, Geochemical Geol., 312–313, 190–194, doi:10.1016/j.chemgeo.2012.04.021. Viola, G., N. S. Mancktelow, and D. Seward, (2001). Late Oligocene-Neogene evolu- tion of Europe-Adria collision: New structural and geochronological evidence from the Giudicarie fault system (Italian Eastern Alps). Tectonics, 20, 999 – 1020,doi:10.1029/2001TC900021.

Von Eynatten, H., Gaupp, R., and Wijbrans, J.R. (1996). 40Ar/39Ar laser-probe dating of detrital white micas from Cretaceous sedimentary rocks of the eastern Alps: Evidence for Variscan high-pressure metamorphism and implications for Al- pine orogeny. Geology, 24(8). 691-694. Von Eynatten, H., and J. R. Wijbrans (2003). Precise tracing of exhumation and provenance using 40Ar/39Ar geochronology of detrital white mica: the ex- ample of the Central Alps, Geol. Soc. London, Spec. Publ., 208(1). 289–305, doi:10.1144/GSL.SP.2003.208.01.14. Wagner, G.A., Reimer G.M., Jäger, E., (1977). Cooling ages derived by apatite fi s- sion-track, mica Rb-Sr and K-Ar dating: the uplift and cooling history of the Central Alps, Società cooperativa tipografi ca Padova. Wagner, G. A., D. S. Miller, R. P. Instttute, and U. S. A. Ny (1979). Fission track ages on apatite of bergell rocks from central alps, Earth Planet. Sci.Lett., 45, 355–360. Wagner, G. A. (1989). The signifi cance of the partial annealing zone in apatite fi ssion- track analysis?: Projected track length measurements and uplift chronology of the Transantarctic Mountains, Chemic. Geol., 79, 295–305. Wang, P., Scherler, D., Liu-Zeng, J., Mey, J., Avouac, J.P., (2014). Tectonic control of Yarlung Tsangpo Gorge revealed by a buried canyon in Southern Tibet. Science 346, 978–981, doi:10.1126/science.1259041. Wänke, H., and König, (1959). Eine neue Methode zur Kalium-Argon-Alterinculun- zum und ihre Anwendung auf Steinmeteorite. Z. Naturforschung, 14a, 860-866.

178 Warren, C.J., Smye, A.J., Kelley, S.P., Sherlock, S.C., (2012). Using white mica 40Ar/39Ar data as a tracer for fl uid fl ow and permeability under high-P con- ditions: Tauern Window, Eastern Alps. J. Metamorph. Geol. 30(1). 63–80. doi:10.1111/j.1525-1314.2011.00956.x. Wiederkehr, M., Sudo, M., Bousquet, R., Berger, A., Schmid, S.M., (2009). Alpine orogenic evolution from subduction to collisional thermal overprint: The 40Ar/39Ar age constraints from the Valaisan Ocean, central Alps. Tectonics 28(6). doi:10.1029/2009TC002496. Wijbrans, J.R., Pringle, M.S., Koppers A.A.P., Scheveers, R., (1995). Argon geochro- nology of small samples using the Vulkaan argon laserprobe. Proc. Kon. Ned. Akad. Wetensch. 98, 185-218. Willett, S.D., Brandon, M.T., (2013). Some analytical methods for converting thermo- chronometric age to erosion rate. Geochem. Geophys. Geosyst. 14, 209–222, doi:10.1029/2012GC004279. Wobus, C. W., K. V. Hodges, and K. X. Whipple (2003). Has focused denudation sus- tained active thrusting at the Himalayan topographic front?, Geology, 31(10). 861–864, doi:10.1130/G19730.1. Wobus, C. W., K. X. Whipple, and K. V. Hodges (2006). Neotectonics of the cen- tral Nepalese Himalaya: Constraints from geomorphology, detrital40Ar/39Ar thermochronology, and thermal modeling, Tectonics, 25(4). 1–18, doi:10.1029/2005TC001935. Wölfl er, A., W. Kurz, H. Fritz, and K. Stüwe, (2011). Lateral extrusion in the Eastern Alps revisited: Refi ning the model by thermochronological, sedimentary, and seismic data, Tectonics, 30, TC4006, doi:10.1029/2010TC002782. Wölfl er, A., Kurz, W., Fritz, H., Glotzbach, C., Danišík, M., (2016). Late Miocene in- creasing exhumation rates in the eastern part of the Alps - implications from low temperature thermochronology. Terra Nov. 28(5). 1–9. doi:10.1111/ter.12221. Yang R., Herman, F., Fellin, M.,G., Maden, C. 2017. Exhumation and topograph- ic evolution of the Namche Barwa Syntaxis, eastern Himalaya. Tectophisics, doi:10.1016/j.tecto.2017.10.026 Yin, A., Harrison, T.M., (2000). Geologic evolution of the Himalayan-Tibetan orogen. Ann. Rev. Earth Planet. Sci. 28, 211–280, doi:10.1146/annurev.earth.28.1.211. Yin, A., Harrison, T.M., Murphy, M.A., Grove, M., Nie, S., Ryerson, F.J., Feng, W.X., Le, C.Z., (1999). Tertiary deformation history of southeastern and southwest- ern Tibet during the Indo-Asian collision. Geol. Soc. Am. Bull. 111, 1644–1664, doi:10.1130/0016-7606(1999)111<1644:TDHOSA>2.3.CO;2. Yin, A., Dubey, C.S., Kelty, T.K., Webb, A.A.G., Harrison, T.M., Chou, C.Y., Celerier, J., (2010). Geologic correlation of the Himalayan orogen and Indian craton: Part 2. Structural geology, geochronology, and tectonic evolution of the Eastern

179 Bibliography Himalaya. Geol. Soc. Am. Bull. 122, 360–395, doi:10.1130/B26461.1. Zachos, J., Pagani, M., Sloan, L., Thomas, E., Billups, K., (2001). Trends, Rhythms, and Aberrations in Global Climate 65 Ma to Present. Science, Vol. 292, Issue 5517, pp. 686-693 DOI: 10.1126/science.1059412 Zeitler, P.K., Johnson, M.N., Briggs, N.D., and Naeser, C.W., (1986). Uplift history of the NW Himalaya as recorded by fi ssion-track ages of detrital Siwalik zircons, in Jiqing, H., ed., Proceedings of the Symposium on Mesozoic and Cenozoic Geology: Beijing, Geological Publishing House, p. 481–49. Zeitler, P.K., Meltzer, A.S., Koons, P.O., Craw, D., Hallet, B., Chamberlain, C.P., Kidd, W.S.F., Park, S.K., Seeber, L., Bishop, M.P., Shroder, J.F., (2001). Ero- sion, Himalayan geodynamics, and the geomorphology of metamorphism. GSA Today 11, 4–9. Zeitler, P.K., Meltzer, A.S., Brown, L., Kidd, W.S.F., Lim, C., Enkelmann, E., (2014). Tectonics and topographic evolution of NB and the easternmost Lhasa block, Tibet. Geol. Soc. Am. Spec. Paper 507, SPE507–02–58, doi:10.1130/2014.2507(02).

Zhang, P., Molnar P., Downs, W.R., (2001). Increased sedimentation rates and grain sizes 2–4 Myr ago due to the infl uence of climate change on erosion rates. Na- ture 410, 891-897. doi:10.1038/35073504. Zhang, X., Honda M., Hamilgton D., (2016). Performance of the High Resolution, Multi-collector Helix MC Plus Noble Gas Mass Spectrometer at the Australian National University, Am. Soc. Mass Spectrom. 27:1937Y1943. DOI: 10.1007/ s13361-016-1480-3. Zimmermann, R., Hammerschmidt, K., Franz, G., (1994). Eocene high pressure meta- morphism in the Penninic units of the Tauern Window (Eastern Alps): evidence from 40Ar-39Ar dating and petrological investigations. Contrib. to Mineral. Petrol. 117(2). 175–186. doi:10.1007/BF00286841.

180 Aknowledgements The work presented in this manuscript is the result of the research that I conducted at the Ar- laboratory of the Vrjie Universiteit of Amsterdam in collaboration with the iTECC network. A special thanks go to the members of the reading committee that accepted to spend time in reading this work. I would like to thanks my supervisor Jan Wijbrans that decided to work with me in the last four years. Jan gave me a total freedom of movement in his laboratory. I learned from you many aspects of the 40Ar/39Ar method. Jan never neglected to me his role of intellectual guidance even during some diffi cult past periods, naturally occurring during a PhD project. I would like to thanks him es- pecially to have learned me how to be rigorous and precise while dealing with Science especially in writing and communicating it to the community. I felt always welcome to jump in his offi ce with the most various ideas, questions, and problems. I would like to thanks, Rodolfo Carosi and Margherita Ferrero that involved me in the “Mineral” project giving me the freedom and fi nancial support to accomplish my Ph.D project. Yani Najman helped me a lot at the beginning of this project. She gave me always a great welcoming at LAC and we shared some challenging moment and discussion during the fi eld work in India. A fundamental part of this research project has started thanks to Peter vd Beek and Jean Braun. Thanks Peter for your scientifi c inputs and for having stress my strength and week points in a constructive way. I want to thanks, Jean to introduce me to the basic of the coding, I always had a great feeling in discussing with you since the be- ginning. I still remember a long discussion about the “Channel fl ow model” in Central Nepal while crossing the MCT. Furthermore, you both have been always involved and supportive while I was presenting our work at the international meetings. Klaudia Kuiper has helped me while working with the Helix +. She explained to me how to run experiments and how to look at data in a “proper way”. She also gave me inputs and comments during the writing of this thesis. Thanks to our Lab techni- cians. Onno Potsma helped me during the experiments with the AGES, we identifi ed many leaks at once and he had the patience of working out how to repair the sample house before the arrival of the new piece. Onno also shared quite a lot of good music and inspiration in the lab. Thanks to Roel for helping me in preparing a large number of samples of different grain sizes and species. I worked with pleasure in testing “Pic- colino” the latest generation of mineral vacuum picking tool. You are the example that good manner and effi cient work are always the best combinations to produce high standard research. I think you should have some ancestral from the Mediterranean area... that is why your pizza is so good! Thanks to Fenny Bosse, you where always ready in helping me and all the new PhD students knocking at your door. It has been a pleasure to share those years on a daily base relationship with Ber- tram, we have slowly known each other and we have found much in common. You introduced me to the Borrel at the Tuinzaal together with Christel , Lara, and Thomas. We had the pleasure to share scientifi c interests, politics, and parties. Christel, we had such lovely time at the beginning of my experience at VUA. The last arrival, Cas (lo splendido), entered like a meteorite in our offi ce. He woke up our calm beings with a lot of energy, Thanks for the good vibes! Xilin has been a good friend and a supportive colleague in the last years, I missed you in the last period at the VUA. I will never forget our fi eld work in the Alps and in Yunnan, and how your blood pressure went high after your fi rst espresso in an Ital- ian “Autogrill” in a hot summer afternoon! I would like also to remember all the nice lunch we have shared with Janne, Jurien, Melissa, Christoph, Steph, Alice, Paolo, Ina, Shaolong, Onno and many others. Janne, you helped me in many situations, from fi nding a temporary location in the Pijp to suggest me some helpful comments while writing papers. We have also shared nice moments in Amsterdam parting with music, It has been fun! I would also like to thanks all the staff of the department that created a proper location for working in a challenging environment. Working in a network like iTECC gave me the opportunity to meet many friend s and colleagues. We were lucky to share such an ambitious project. We had the plea- sure to share many unforgettable moments together. An extremely fulfi lling experience was shared between a team of three Belgians, one Italian and one France, but describing this would require another chapter... mercie le gars pour l’ávventura Nepalese. We learned a lot from that experience. Natalie, pressionnnnnn! Zazá, Ruben, Ale (Giannetto), Eric and all the other iTECC mates, we had extremely funny and boring moments together. Some of us are not here anymore... to Gwladys once more all my love. Ai miei genitori e alla mia famiglia dedico infi ne questa tesi. La mia famiglia e costellata di fraterni amici che mi hanno sempre sopportato! Mamma, babbo mi avete sempre spronato alla scoperta e all’áccettazione della diversitá culturale. Mi avete navigato alla vita offendomi ogni mezzo per poter di- ventare un esploratore. Ma la cosa piu’importante é che mi avete donato un’infanzia felice e serena. Siete sempre il mio approdo sicuro. A mia sorella, Camilla, grazie per avermi supportato nei momenti diffi cili e durante le gioie della mia esistenza. Sei stata un bell’ésempio come sorella minore. Nonno, zio, so che sareste stati fi eri di me e che non sareste riusciti a non intervenire durante la mia discussione. Nonno insieme alla ni’e alla vostra magia avete fatto proprio un bel casino. Dani tu mi hai sopportato e sostenuto negli ultimi anni, la tua calma e ragione danzano all’únisono con la mia impulsivitá incontrollata. Dalla nostra unione é nata Sofi a. Per noi é iniziata una strada ricca di bellezza, poesia ed avveniere. Il resto sono solo pratiche di vita da dover diligentemente affrontare senza mai perdere lo stupore e la sorpresa di vivere assieme. A voi che ci siete stati, grazie. Bedank!