Patrick W. Keys

The Precipitationshed – Methods, Concepts, and Applications Patrick W. Keys

The Precipitationshed

Methods, Concepts, and Applications

Patrick W. Keys c Patrick W. Keys, Stockholm 2016

ISBN 978-91-7649-464-6

Printed in Sweden by Holmborgs, Malmö, Stockholm 2016 Distributor: Stockholm Resilience Centre, Stockholm University

Papers I, II, and III and corresponding ﬁgures are reprinted here with permission from the publishers.

Cover Image: The Thunderstorm, c Tez Goodyear (Creative Commons, CC BY-ND 2.0) Images preceding Papers I, III, IV, and V: c Patrick W. Keys 2016, All Rights Reserved. Image preceding Paper II: c Brokentaco (Creative Commons, CC BY 2.0) Abstract

Human societies are reliant on the functioning of the hydrologic cycle. The atmospheric branch of this cycle, often referred to as moisture recycling in the context of land-to-land exchange, refers to water evaporating, traveling through the atmosphere, and falling out as precipitation. Similar to the surface water cycle that uses the watershed as the unit of analysis, it is also possible to consider a ‘watershed of the sky’ for the atmospheric water cycle. Thus, I explore the precipitationshed - defined as the upwind surface of the Earth that provides evaporation that later falls as precipitation in a specific place. The primary contributions of this dissertation are to (a) introduce the precipitationshed concept, (b) provide a quantitative basis for the study of the precipitationshed, and (c) demonstrate its use in the fields of hydrometeorology, land-use change, social-ecological systems, ecosystem services, and environmental governance. In Paper I the concept of the precipitationshed is introduced and explored for the first time. The quantification of precipitationshed variability is described in Paper II, and the key finding is that the precipitationsheds for multiple regions are persistent in time and space. Moisture recycling is further described as an ecosystem service in Paper III, to integrate the concept into the existing language of environmental sustainability and management. That is, I identify regions where vegetation more strongly regulates the provision of atmospheric water, as well as the regions that more strongly benefit from this regulation. In Paper IV, the precipitationshed is further explored through the lens of urban reliance on moisture recycling. Using a novel method, I quantify the vulnerability of urban areas to social-ecological changes within their precipitationsheds. In Paper V, I argue that successful moisture recycling governance will require flexible, trans-boundary institutions that are capable of operating within complex social-ecological systems. I conclude that, in the future, the precipitationshed can be a key tool in addressing the complexity of social-ecological systems.

Keywords: water, atmosphere, precipitationshed, moisture recycling, variability, ecosystem services, social-ecological systems

Sammanfattning

Våra samhällen är beroende av vattnets kretslopp på planeten. Den atmosfäris- ka delen av detta kretslopp över land kallas för fuktåtercirkulering och refererar till vattnet som avdunstar från land, färdas genom atmosfären som vattenånga, och slutligen faller som nederbörd över land. Analogt med vattendelare som analysenhet för ytvatten, är det också möjligt att tala om vattendelare för det atmosfäriska vattnet. Dessa kallas för nederbördsdelare och deﬁnieras som de uppvindsområden på jorden som utgör avdunstningskällor för en given regions nederbörd. Denna avhandlings främsta bidrag är att (a) introducera konceptet nederbördsdelare, (b) tillhandahålla en kvantitativ grund för studier av nederbördsdelare, och (c) visa hur begreppet kan användas inom ämnesområ- dena hydrometeorologi, markanvändningsförändringar, social-ekologiska system, ekosystemtjänster och miljöstyrning.

I Papper I introduceras och utforskas begreppet nederbördsdelare. Kvantifi- ering av nederbördsdelares variabilitet beskrivs i Papper II, där en viktig slut- sats är att fuktkällorna för en given plats är beständiga i tid och rum. Papper III beskriver fuktåtercirkulering som en ekosystemtjänst, i syfte att förankra konceptet inom miljömässig hållbarhet och miljöstyrning. Papper III kartläg- ger även de områden i världen där vegetationens reglering av fukt i atmosfären är extra betydelsefull, samt de regioner som drar mest nytta av vegetationens förmåga att reglera avdunstning. Papper IV undersöker nederbördsdelare ur ett urbant vattenförsörjningsperspektiv. Med hjälp av en ny metod kvantifie- rar jag olika stadsområdens sårbarhet för social-ekologiska förändringar inom deras nederbördsdelare. I Papper V argumenterar jag för att en framgångsrik styrning av fuktåtercirkulering kommer att kräva flexibla, gränsöverskridan- de institutioner som är i stånd att arbeta inom ett komplext social-ekologiskt system. Jag drar slutsatsen att nederbördsdelare i framtiden kan bli ett viktigt verktyg för att hantera social-ekologiska systems komplexitet.

Nyckelord: vatten, atmosfär, nederbördsdelare, fuktåtercirkulering, variabilitet, ekosystemtjänst, social-ekologiska system This thesis is dedicated to my family. List of Papers

Papers included in this thesis

The following papers and manuscripts, referred to in the text by their Roman numerals, are included at the end of this thesis. PAPER I: Analyzing precipitationsheds to understand the vulnerability of rainfall dependent regions. Keys, P.W., van der Ent, R.J., Gordon, L.J., Hoff, H., Nikoli, R., and Savenije, H.H.G. Biogeosciences (9), 733-746 (2012). DOI: 10.5194/bg-9-733-2012

PAPER II: Variability of moisture recycling using a precipitationshed framework. Keys, P.W., Barnes, E.A., van der Ent, R.J. and Gordon, L.J. Hydrology and Earth System Sciences, 18 (10) (2014). DOI: 10.5194/hess-18-3937-2014

PAPER III: Revealing Invisible Water: Moisture Recycling as an Ecosys- tem Service. Keys, P.W., Wang-Erlandsson, L., and Gordon, L.J. PLoS One, 11 (3) (2016). DOI: 10.1371/journal.pone.0151993

PAPER IV: Megacity precipitationsheds reveal reliance on regional evaporation for moisture supply. Keys, P.W., Wang-Erlandsson, L. and Gordon, L.J. Landscape and Urban Planning (under review)

PAPER V: Approaching Moisture Recycling Governance. Keys, P.W., Wang-Erlandsson, L., Galaz, V., Ebbesson, J., and Gordon, L.J. manuscript Relevant co-authored publications not included herein.

i Conditional Relationships between Drought and Civil War in sub- Saharan Africa. Bell, C.M. and Keys, P.W. Foreign Policy Analysis. (2016). DOI: 10.1093/fpa/orw002

ii Global root zone storage capacity from satellite-based evaporation. Wang-Erlandsson, L, Bastiaanssen, W. G. M., Gao, H., Jagermeyr, J., Senay, G.B., Gordon, L.J., Keys, P.W., van Dijk,A. I. J. M., Guerschman, J.P., and Savenije, H. H. G. Hydrology and Earth System Sciences, 20,1459- 1481 (2016). DOI: 10.5194/hess-20-1459-2016

iii Contrasting roles of interception and transpiration in the hydrological cycle - Part 2: Moisture recycling. van der Ent, R. J., Wang-Erlandsson, L., Keys, P. W., and Savenije, H. H. G. Earth System Dynamics, 5, 471-481. (2014). DOI: 10.5194/esd-5-471- 2014

iv Food Policy and Civil War in Africa’s Drought-Suffering States. Bell, C.M., Keys, P.W., and Stossmeister, R. in: Conﬂict-Sensitive Adapta- tion to Climate Change in Africa, eds. Bob, Urmilla and Salome Bronkhorst. Berliner Wissenschafts-Verlag (Berlin Science Publishers), Germany (2014)

v Releasing the pressure: water resource efﬁciencies and gains for ecosystem services. Keys, P.W., Barron, J., and Lannerstad, M. United Nations Environment Programme, ISBN: 978-92-807-3340-5 (2012)

vi Chapter 12: Watershed management through a resilience lens. Barron, J. and Keys, P.W. in: Integrated Watershed Management in Rain- fed Agriculture. eds. Wani, S.P., Rockström, J., and Sahrawat, K.L. CRC Press. (2011) vii Precipitation extremes and the impacts of climate change on stormwa- ter infrastructure in Washington State. Rosenberg, E.A., Keys, P.W., Booth, D.B., Hartley, D., Burkey, J., Steine- mann, A.C., and Lettenmaier, D.P. Climatic Change, 102 (1-2), 319-349. (2010). DOI: 10.1007/s10584-010-9847-0 Contents

Abstract v

Sammanfattning vii

List of Papers ix

Acknowledgements xiii

1 Introduction 17 1.1 Moisture Recycling as a Socially-Relevant Geophysical Process 17 1.2 The Precipitationshed is a Management Entry Point ...... 22 1.3 Summary of this work ...... 23

2 Methods 27 2.1 WAM-2layers ...... 27 2.2 STEAM ...... 29 2.3 STEAM+WAM ...... 29

3 Contributions 31 3.1 Concepts ...... 31 3.2 Methods ...... 32 3.3 Applications ...... 35 3.4 Trans-disciplinary Impact ...... 36

4 Limitations and Considerations 39

5 Next Steps 41

6 Advice to Future PhD Students 43

7 Concluding Remarks 47

8 References 49

Acknowledgements

This PhD would not have been possible were it not for the trust that my advisor Line Gordon granted me as we embarked four years ago on what has been a true test in tele-commuting. Line trusted that with periodic visits to the Stockholm Resilience Centre, and regular guidance via Skype and email, that we could develop a successful research relationship. I’m grateful for that opportunity, and have learned a tremendous amount along the way.

To my colleagues Ruud van der Ent and Lan Wang-Erlandsson, thank you for being critics, advisors, and guides during my PhD. Its surprising to have found a group of people to whom I only felt gratitude when you told me that my methods were incorrect, that my logic was mistaken, or that the underlying relevance was questionable. I could not have asked for a better group of people with whom to share this journey.

To the many reviewers of this thesis, and the articles herein, I appreciate your patience and encouragement. I’m especially grateful for those of you who offered feedback on papers that were perhaps out-of-bounds of your own research. Have comfort. They were out-of-bounds for me, too.

To my fellow PhD students, thank you for the philosophical debate, the ideal- ism, and the fun. I can’t tell you all how much your intellectual and personal connections have meant to me, and I hope that as we spread out around the world that we can continue to be waypoints for one another, academic or oth- erwise.

Thank you to Professor Scott Denning, and his BioCycle research group at Colorado State University, who provided an enthusiastic and welcoming academic home-away-from-home.

Finally, to Elizabeth, my fulcrum for the last four years, without whose support this PhD would not have been possible - Thank You.

1. Introduction

The global water cycle is the bloodstream of the biosphere, ﬂowing in and through every living organism and ecosystem, and connecting the disparate ecologies of our planet to one another. The liquid water that is visible in rivers, lakes, and oceans is a critical part of the water cycle - yet only a part. Along with the reservoirs of water hidden beneath the Earth, the atmospheric branch of the water cycle contains vast ﬂows of water.

Each year, between 400,000 and 500,000 cubic kilometers of water evaporates from the global oceans and flows into the atmosphere (Trenberth et al., 2011). While most of this evaporation returns to oceans as rainfall, between 30,000 and 45,000 cubic kilometers of water vapor flows back toward land. Furthermore, between 70,000 and 100,000 cubic kilometers of water evaporates from the land into the atmosphere where it combines with oceanic evaporation. Then, between 105,000 and 130,000 cubic kilometers of water falls as precipitation on land, providing the water that will eventually flow off the land into the rivers, and back to the ocean. This process of evaporation from the ocean and land, traveling through the atmosphere as water vapor, and return- ing to the surface as precipitation is often referred to as moisture recycling.

1.1 Moisture Recycling as a Socially-Relevant Geophysi- cal Process

Until very recently, the study of moisture recycling was limited to the ﬁelds of hydrology, hydrometeorology, and atmospheric science. The interest was primarily rooted in a traditional viewpoint of understanding the geophysical process of how much evaporation enters the atmosphere, where it ﬂows as water vapor, and where it falls out as precipitation. Along the way, some researchers began referring to moisture recycling as a local process, where evaporation is returned immediately to the same local area as precipitation (e.g. Lettau et al. 1979). Others referred to moisture recycling as a tele-connected process, that brought evaporation from one location to another, distant location as precipita-

17 Schematic of Atmospheric Branch of Water Cycle

water vapor transport water vapor transport water vapor transport evapora,on evapora,on precipita,on evapora,on precipita,on evapora,on precipita,on precipita,on

ocean land ocean

Figure 1. Two-dimensional schematic of the atmospheric branch of the water cycle, emphasizing terrestrial moisture recycling relationships. Within the orange dashed box, precipitation that is primarily of oceanic origin falls onto land, and then the water that evaporates is from the land surface. This water is trans- ported through the atmosphere, and falls out as precipitation again, and is then re-evaporated on land again. tion (e.g. Koster et al. 1986). And still other researchers include both of these perspectives in their deﬁnition of moisture recycling (e.g. van der Ent et al. 2010). I use this last more inclusive deﬁnition throughout this dissertation.

Theories of land-atmosphere interactions, particularly the consequences of land-use change on precipitation, have been around for centuries. The pop- ular 19th century notion that the ‘rain-follows-the-plow’ developed separately among both North American and Australian farmers (Reisner 1993). Clima- tologists of that time believed the mechanism of land-use change driving more precipitation involved, among other things, the stirring up of soil moisture which increased local rainfall. Though these are intriguing ideas, this theory is now understood as mostly wrong, especially since modern climatological reconstructions show the Midwestern US was experiencing a wetter than nor- mal period contemporary with the ‘rain-follows-the-plow’ theory. Nonethe- less, land-use change does have very real consequences for the atmospheric branch of the water cycle, and the ‘rain-follows-the-plow’ theory highlighted an early effort to link land-use change to the atmospheric water cycle, and to further social impacts.

In the 1970’s, a theoretical explanation for land-use change impacting precipitation was proposed, particularly in the context of desertiﬁcation. Charney et al. (1973) and Otterman (1974) both suggested that as vegetation is removed

18 from the land-surface, and brighter soils are exposed to the sky, that this bare ground would reflect more energy (rather than absorb it). As a result of this net reduction in surface energy, there would be less energy emitted to the atmosphere to induce convective cloud formation and subsequent precipitation. This theory, which was also empirically demonstrated, remains one of the cor- nerstones of land-atmosphere interactions. Recent work has highlighted the role that oceans also play in governing surface drying (or wetting), especially in the Sahel (e.g. Giannini et al., 2003; Giannini, 2016). The early work led by Charney and Otterman complements this more recent work, by identify- ing local and proximate land-atmosphere interactions, and how they fit with the more distant and ultimate causes of desertification. Given that the mechanisms that drive land-atmosphere interactions appear to operate at different spatial and temporal scales, it may be revealing to compare how atmospheric, hydrologic, and land surface processes overlap in space and time (Figure 2).

Comparing Moisture Recycling to Other Atmospheric and Surface Processes Global

10,000 km Climate Atmospheric LEGEND El Niño Change Transport and e.g. Aerosols, SST La Niña Meteorological CO2 Variability Regional 1,000 km Hydrological Mesoscale Large River Glaciers Convective Basins systems Land Surface Spatial scale Ground

100 km Local

Mesoscale Water Small River Basins Irrigation or Moisture Recycling Land-Use Soil Moisture and Change Convective Vegetation Variability Large scale Cells Local 10 km

Hour Day Season Year Decade

Timescale

Figure 2. Comparing moisture recycling (orange circle) with the spatial and temporal scales of meteorological, hydrological and surface processes. This is adapted from Tuinenburg, 2013.

Though there is limited overlap between ‘irrigation and land-use change’ (right- most green circle) and ‘moisture recycling’ (dashed, orange circle), they are strongly connected via ‘soil moisture and vegetation variability’ (left-most green circle). In other words, land-use changes are likely to occur on time- scales that are mostly disconnected from moisture recycling, but the conse-

19 quences of land-use change for moisture recycling manifest at the seasonal and annual scale.

There are several mechanisms by which land-use change can modify moisture recycling. Using ‘deforestation for the purpose of cropland expansion’ as an example of land-use change, there are many possible interactions that can lead to changes in rainfall. At the local scale, the interaction between the new cropland and the new forest edge can induce a signiﬁcant temperature gradient that can induce moist static instabilities (e.g. Lawrence and Vandecar, 2015), which can lead to local increases in precipitation. These energy gradient effects appear to diminish as spatial and temporal scales increase, with corresponding increases in the importance of regional moisture budgets, including the magnitude and timing of evaporation to the atmosphere (i.e. moisture recycling). Finally, as spatial and temporal scales increase further, regional and global drivers start to dominate, such as sea-surface temperatures, interannual climate oscillations (e.g. El Niño Southern Oscillation), and perhaps even climate change. In my work, I focus primarily on the regional and sub-regional scale, where moisture recycling effects are most prominent (e.g. van der Ent et al., 2013).

In recent decades, the ability to represent the processes of moisture recycling in computer simulations has improved, and consequently advanced our understanding tremendously. Some of the early modeling work that explicitly sought to relate moisture recycling studies with socially-relevant implications were the studies that aimed to understand the role of moisture recycling and regional flooding, particularly in the USA (Dirmeyer and Brubaker, 1999; Dominguez et al. 2006). This type of research was expanded to large river basins globally (Bosilovich et al., 2006). These, and other studies, primarily sought to understand how the large-scale atmosphere influenced specific human societies via moisture recycling. Conversely, early empirical work examining the impact that human societies have on moisture recycling focused on deforestation of the Amazon as a potential cause of changes in downwind precipitation (e.g. Salati et al., 1971).

Several studies have explicitly examined the consequences of anthropogenic land-use change on moisture recycling processes, and how these altered processes further affect precipitation (see Table 1). All these studies use model simulations, and can be categorized according to those representing actual changes, i.e. land-use changes that have already happened, such as irrigation of the Central Valley of California, and theoretical changes, i.e. potential changes such as extensive Amazonian deforestation. Likewise, the stud-

20 ies generally explore either irrigation scenarios (generally leading to increased evaporation), or de-vegetation scenarios (generally leading to decreased evaporation). As a note, I use the word ‘evaporation’ to refer to all evaporative ﬂuxes, including vegetation interception, transpiration, soil interception, soil evaporation, and open water evaporation. My use of evaporation ought to be understood as referring to what others refer to as evapotranspiration (Savenije 2004).

Table 1. Summary of literature values for land-use changes and the associated impact to downwind precipitation. This table is adapted from Paper IV.

AUTHOR REGION LAND-USE- TYPE-OF CHANGE-IN-PRECIPITATION CHANGE-(LUC) LUC absolute % Bagley&et&al.& East&Asia Difference&between&& Theoretical K&K&K& K11.89% (2012)& Central&Asia natural&vegetation&and& K&K&K& K16.70% North&America bare&soil K&K&K& K8.34% South&America K&K&K& K16.90% West&Africa K&K&K& K9.64% Salih&et&al.& Sudan&(Sahel) Deforestation,&replaced& Theoretical grass:&K1mm/day&to&+0.5&mm/day grass:&K25%&to&+5% (2013) by&grass&or&desert desert:&K2.1&mm/day&(or&more)&to& desert:&K52.5%&to&+5% +0.5&mm/day Lo&and& California Irrigation&replacing& Observed from:&+2mm/month&to& about&+15% Famiglietti& grassland& 12mm/month (2013) Tuinenburg&et& India,&Ganges& Irrigation Observed from:&K200&mm/yr&in&Eastern&India& from:&K15%&in&Eastern&India&to& al.&(2013) plain to&+200&in&W.&India,&N.&India,&and& +15&to30%&in&W.&India,&N.& Pakistan India,&and&Pakistan Wei&et&al.& India Irrigation&replacing& Observed from&120mm/yr&to&10mm/yr from&+22%&to&+2% (2013) variety&of&different& China landKuses from&28&mm/yr&to&2&mm/yr from&+5%&to&+0.4%

US from&5&mm/yr&to&0.4&mm/yr from&+1.1%&to&+0.1&%

Sahel from&4.5&mm/yr&to&0.4&mm/yr from&+3%&to&+0.2%

Spracklen&et&al.& Amazon Deforestation&(replaced& Observed&and& K&K&K& from:&K0%&(0%&deforestation)& (2015) by&variety&of&landKuses& Theoretical to&~20%&(with&100%& depending&on& deforestation) simulation) Swann&et&al.& Amazon Deforestation&(replaced& Theoretical from&+1mm/day&Kto&K3mm/day& from&+17%&to&K25%& (2015) by&cropland) (+365mm/yr&to&K900mm/yr) (Howver,&most&changes&=&0%) but,&most&changes&are&0 Badger&and& Amazon Deforestation&(replaced& Theoretical from&K8mm/day&to&K2mm/day&in& K&K&K Dirmeyer& by&heterogenous& NW&Amazon&(K1460&mm/yr& (2015) cropland) to&K365&mm/yr);& from&K1&mm/day&to&+1&mm/day&in& southern&and&eastern&Amazon& (K365&mm/yr&to&+365mm/yr) Halder&et&al.& South&Asia Deforestation&(replaced& Observed from&K16&mm/yr&to&+16mm/yr K&K&K (2015) by&cropland) Keys&et&al.& Amazon& Deforestation,&replaced& Theoretical from:&K80mm/yr&to&K10mm/yr from:&K6%&to&K1% (2016) (Mato&Grosso) by&desert ! The wide range in impacts that land-use change can have on rainfall are visible in Table 1. Though its generally true that increased transpiration (via irrigation) leads to more precipitation downwind, there are exceptions, as with the ﬁndings of Tuinenburg et al. (2013), where increased irrigation led to less precipitation downwind due to changes in atmospheric circulation. Likewise, though its generally true that decreased vegetation (e.g. via deforestation) leads to less precipitation downwind, the range of this decrease can be very large, even for the same study area (see the difference between Badger and

21 Dirmeyer (2015) and Swann et al. (2015) in the Amazon). The key message here is that though land-use change appears to have a signiﬁcant impact on downwind precipitation, the magnitude (and sometimes the sign) of this impact is less certain.

Research that examines the role of land-use change in modifying moisture recycling patterns implicitly has social relevance. However, until recently the ﬁeld of moisture recycling research lacked a speciﬁc spatial heuristic with which to structure the discussion of ‘regions that provide moisture (e.g. evaporation sources) and regions that receive moisture (e.g. precipitation sinks)’. The precipitationshed is a response to this gap.

1.2 The Precipitationshed is a Management Entry Point

As far as people are concerned, the core insight of moisture recycling research to date is that land-use change in one place may affect precipitation in another. One reason this is so central to social systems is that humanity as a whole is overwhelmingly reliant on rainfall for its food production. 80% of global food production is produced in rainfed, rather than irrigated, agricultural systems (Molden, 2007). Much of this food production is reliant on rainfall that arises as evaporation over other land surfaces, so-called terrestrial moisture recycling (e.g. Bagley et al., 2012; Paper I). Furthermore, this rainfed crop production does not include the rainfed woodlands, rangelands, and myriad other uses of rainfall for human well-being. As a result of humanity’s reliance on terrestrial moisture recycling, it would be useful to have a tool that (a) can assist in the identification of areas that provide rainfall to a given location, and (b) is flexible enough to incorporate multiple datasets, models, and spatial and temporal scales. The precipitationshed is this tool. The precipitationshed is defined as the upwind surface (both water and land) that contributes evaporation to a given location’s precipitation (Figure 2; also see Paper I). As depicted, evaporation rises up from the surface (thin, wavy arrows pointing up), flows through the atmosphere (large arrows), and falls as precipitation downwind (stippled arrows pointing down). The parallel concept to the precipitationshed is the evaporationshed, which is essentially the reverse of the precipitationshed concept: the area that receives water as precipitation from a given location’s evaporation (see Paper III). The reason that a unit of analysis like the precipitationshed is necessary for engaging the social dimensions of moisture recycling is that there needs to be a way to identify the specific social and ecological systems that provide, or receive, moisture.

22 Relevant social systems can vary, ranging from small (e.g. agricultural communities) to large (e.g. large countries). Ecological systems can also vary, ranging from small (e.g. a small forest patch) to large (the Amazon basin). However, given that humanity is currently modifying more than 50% of the Earth’s land surface (Ellis et al., 2010), it is important to recognize that the social and ecological systems on which I focus are tightly coupled, denoted as social-ecological systems, with both fast and slow feedbacks generating unex- pected behavior. Importantly, these social-ecological systems are also complex adaptive systems, which makes it all the more important to recognize the in- herent potential for nonlinear and surprising behavior in the regions that are experiencing change (Folke et al., 2005).

1.3 Summary of this work

In this dissertation, I investigate: (a) the concept of the precipitationshed, (b) how robust the patterns are in space and time, (c) how the concept may be

Figure 3. Conceptual diagram of a precipitationshed. Originally published in Paper I in Biogeosciences, and reproduced here under the Creative Commons Attribution 3.0 License.

23 integrated into existing discourses of human-environment interactions, and (d) how the concept might connect more broadly with the governance and management of the Earth system. Paper I is the pilot study for this research, and introduced the concept of the precipitationshed by exploring the vulnerability of rainfall-dependent dryland agricultural systems. In that research, I identified the spatial domain of each region’s precipitationshed, and qualitatively explored the vulnerability of each precipitationshed to patterns of land-use change. The key findings of Paper I are as follows: 1. The precipitationshed is a novel concept in hydrology that can provide new insights into understanding how evaporation in an upwind area is related to precipitation downwind. 2. There are key hotspots globally where rainfed agriculture in drylands is highly reliant on precipitation that comes from terrestrial sources. 3. Moisture recycling is vulnerable to the interconnected pressures of land- use change, geopolitical complexity, and population pressure; this is especially true in China. In Paper II, I answer a key reservation about the utility of the precipitationshed concept, namely whether or not the precipitationshed has low or high variability. In general, the spatial extent and shape of precipitationshed boundaries are persistent in time. Additionally, the primary source of variation for the precipitationshed is a pulsing of evaporation, rather than a spatially shifting pattern. These two findings legitimize subsequent exploration, by confirming that the source regions providing moisture (i.e. evaporation sources) persis- tently contribute moisture to their sink regions (i.e. precipitation sinks). The key findings of Paper II are as follows: 1. Precipitationsheds persist in space and time. 2. The leading pattern of variability is a pulsing (increase or decrease) from the core precipitationshed area. 3. Precipitationshed shapes generally agree when using different input datasets. Paper III extends the concept of terrestrial moisture recycling; wherein I calculate the specific fraction of moisture recycling that is regulated by vegetation. I wanted to understand how much of current evaporation is due to the current vegetation (relative to a hypothetical bare-soil condition). In this paper, I quantified the global benefits provided by current vegetation for moisture recycling by comparing the amount of moisture recycling that takes place in the current world versus a world where all land surfaces have been turned to bare soil surfaces. I defined this vegetation-regulated moisture recycling as an

24 ecosystem service, and found that many parts of the world are either important providers of this service, or important beneficiaries of this service. Likewise, a case study in the Mato Grosso region of Brazil demonstrates how the concept could be operationalized in a more specific location, and provided a path forward for integrating the concept into future analyses of ecosystem service trade-offs. The key findings of Paper III are: 1. Vegetation-regulated Moisture Recycling (VMR) ecosystem services are a significant source of rainfall for many regions. 2. Some regions rely on current vegetation for equal to or greater than 50% of annual rainfall. 3. VMR ecosystem services can provide an important seasonal benefit, shortening the dry season. Several previous studies had suggested that the relative importance of terrestrial moisture recycling increased during dry years. I explored this idea in Paper IV through the lens of urban water supplies, and used a previously identified group of megacities globally that were reliant on surface or groundwater for their municipal water. Of the 29 megacities globally, 19 of them were reliant on terrestrial sources of moisture for a third or more of their precipitation. Likewise, many of the megacities were significantly more reliant on terrestrial sources of moisture recycling during dry years. Using an integrated moisture recycling vulnerability analysis, 6 megacities were found to be highly vulnerable to potential changes in the precipitationshed. The key insights of Paper IV are: 1. Of the 29 megacities globally, 19 rely on evaporation from land for more than 1/3rd of the annual rain entering their watersheds. 2. Many of the 19 megacities are more dependent on land evaporation sources (rather than oceans), during dry years. 3. The vulnerability analysis reveals that 6 megacities are highly vulnerable to changes in moisture recycling. In Paper V, I examine how to approach the governance of moisture recycling, particularly through the lens of international law and institutions. I focus on the nation as the unit of analysis (similar to the examination of the import and export of moisture from Dirmeyer et al. 2009), and thus examine geopolitical interactions, as well as international legal perspectives. I first calculated the precipiationsheds of 30 nations, and then developed a typology based on these moisture recycling patterns. Using the results of the typological classifi- cation, I examine how current legal and institutional perspectives might inform

25 moisture recycling governance. The key insights of Paper V are: 1. There is a broad range of geopolitical complexity in the case study nations, with the number of nations overlapping the precipitationshed ranging from 2 to 13. 2. For the less complex precipitationsheds, legal principles such as “do no harm” and “equitable sharing of resources” may be most appropriate strategies for governance. However, as the number of actors increases and the moisture recycling relationships become more complex, other approaches may be necessary including: strengthening existing (or cre- ating) transnational collaborations, forming multinational scientific assessments, diffusing of norms with the aim to reduce harm, and estab- lishing common but differentiated responsibilities. 3. The complexity of moisture recycling means institutional fit will be difficult to generalize for all precipitationsheds, and that for the most complex precipitationsheds, institutions will likely be emergent rather than prescribed.

26 2. Methods

I use two models to explore moisture recycling: the Water Accounting Model (WAM-2layers) (Van der Ent, R. J., 2014), and the Simple Terrestrial Evapora- tion to Atmosphere Model (STEAM) (Wang-Erlandsson et al., 2014). WAM- 2layers is a model that is able to keep a detailed account of the atmospheric water cycle, and STEAM simulates evaporation from the land-surface. Both of these models require input data, which I detail below.

2.1 WAM-2layers

Research that aims to track the flow of water through the atmosphere can use a variety of methods, including: (a) isotopic tracers, (b) a model that can simulate the pathway of different parcels of moisture through the atmosphere (i.e., Lagrangian approach), or (c) an accounting procedure that can keep track of where moisture travels (i.e., Eulerian approach). I use the latter in my research, since the Water Accounting Model 2layers (i.e. WAM-2layers) keeps track of the vertical flux of evaporation and precipitation, as well as the horizontal fluxes of water vapor around the planet. The WAM-2layers does not simulate anything, as does a weather or climate model, but rather the model tracks where water moves around the Earth, based on input data. I used a single- atmospheric-layer version of the WAM in Paper I, and the WAM-2layers (i.e. two atmospheric layers, see Figure 4) in all other papers in this dissertation. The results from the two versions of the model are, in general, comparable, and analyzed in van der Ent et al. (2013). Hereafter, I will refer to the moisture tracking procedure I used as the WAM-2layers. Imagine a single piece of land, and the column of atmosphere that exists above that land (Figure 4). The WAM-2layers first tracks evaporated water (i.e. water vapor) that rises up into the lower layer of the atmosphere. Next, this water vapor is mixed with water that arrives into that column from all directions. At the same time, the water vapor in the lower layer of the atmosphere is verti- cally mixed with the water vapor in the upper layer of the atmosphere, which is also mixed with water vapor that arrives from all directions. The precipitation

27 Overview of a Single Column of Atmosphere in the WAM-2layers

1 = Evaporation rises up from the surface of the Earth as water vapor into the atmosphere. Upper layer 2 = The evaporation enters the lower layer of the 4 atmosphere, and the water is mixed with water vapor coming from north, south, east, and west.

3 = Water vapor is exchanged between the lower and upper layers of the atmosphere, based on vertical winds.

Atmosphere 3 4 = The water vapor from the lower layer enters Lower layer the upper layer of the atmosphere and is mixed with water vapor coming from north, south, east, 2 and west.

5 = Precipitation falls on the surface of the Earth as a mixture of the evaporation that entered the atmosphere, as well as the water vapor that had traveled into the upper and lower layers of the 1 5 atmosphere.

Earth’s surface

Figure 4. Diagram of moisture exchange in a single column of atmosphere for the WAM-2layers. that falls out of the bottom of the column of atmosphere is thus a combina- tion of the evaporation that entered that column of air, as well as other water vapor that was mixed into that column of air from outside the column. This process is completed simultaneously at all locations on the planet, and moisture is tracked from where it enters and exits the atmosphere, as evaporation and precipitation, respectively. Thus, in stepping forward in time, it is possible to identify when and where a speciﬁc region’s precipitation entered the atmosphere as evaporation.

Since the WAM-2layers does not simulate anything, it requires speciﬁc input data for evaporation, precipitation, wind, speciﬁc humidity, and surface pressure. With these data, it is possible to reconstruct the atmospheric branch of the water cycle, and calculate both precipitationsheds and evaporationsheds. In my research I have used a type of data called reanalysis data. These data combine output from weather forecast models that is combined with observed data from satellites, weather balloons, surface measurements, and other sources. I have used two reanalysis products, including the European Reanalysis Interim product (i.e. ERA-Interim) from the European Centre for Medium-Range Weather Forecasting (Dee et al., 2011), and the Modern Era Retrospective Reanaly-

28 sis product (i.e. MERRA) from the National Aeronautics and Space Agency (NASA) (Bosilovich et al., 2011). The data have several parameters, including how frequently data are available (i.e. the timestep) and the spatial resolution. The WAM-2layers uses precipitation and evaporation data at the 3-hour timestep, and humidity, surface pressure, and wind data at the 6-hour timestep. I use data that are at the 1.5 x 1.5 degree grid resolution, and span the period 1979 to 2014. These parameters were slightly different for the MERRA dataset, with the same timestep and data period, but with a 1.0 x 1.25 degree grid resolution. The ERA-Interim data were used in Papers I thru V, and the MERRA data were used speciﬁcally in my analysis of moisture recycling variability in Paper II.

2.2 STEAM

In addition to the WAM-2layers model, some of my research uses the Sim- ple Terrestrial Evaporation to Atmosphere Model (STEAM), which realisti- cally simulates evaporation from the land surface. STEAM simulates evaporation based on a variety of input climatological variables (such as precipitation, temperature, humidity, surface wind). STEAM also includes data on soil type (Harmonized World Soil Database), land cover type (using Moderate Resolu- tion Imaging Spectroradiometer satellite data; Friedl et al., 2010), and irrigated croplands (from the MIRCA2000 dataset, Portmann et al., 2010).

STEAM intercepts incoming precipitation into ﬁve different stocks. First, the vegetation canopy intercepts incoming precipitation. Second, the soil surface intercepts the precipitation that makes it through the canopy. Third, water is taken up into the soil and can evaporate from the soil column. Fourth, water can be taken up into the roots of plants and into plant tissues. Fifth, surplus water can pool in open bodies of water. STEAM, then, simulates evaporation in the following order: vegetation interception, transpiration, soil interception, soil moisture evaporation, and ﬁnally open water evaporation.

2.3 STEAM+WAM

Performing the coupling procedure whereby STEAM and WAM-2layers are linked enables the exploration of the changes in the atmospheric moisture budget that result from land-use change, and its impacts on evaporation. The coupling is performed by initially allowing the evaporation from STEAM to

29 modify the input evaporation for the WAM-2layers. This, in turn, modiﬁes the water vapor in the atmosphere, and eventually the precipitation. Then, for a second run of STEAM, the altered precipitation is used as the input for STEAM to modify the amount of moisture available for the evaporation parti- tioning. This process is repeated for several iterations, until the changes in the moisture budget converge, such that the difference at every gridcell between runs is below 1%.

I use a coupled version of STEAM+WAM in Paper III, and incorporate the output from Paper III into Paper IV and Paper V.

30 3. Contributions

This dissertation constitutes a major advance in the ﬁeld of moisture recycling, not least because it develops new conceptual tools that can be used for empirically understanding the hydrological cycle, but also because it explores new methods and applications associated with the precipitationshed. The contributions of my dissertation can be separated into three broad categories: concepts, methods, and applications.

3.1 Concepts

There are three key conceptual contributions made in this dissertation: Introduction and description of the precipitationshed • Moisture recycling as an ecosystem service • Typology of moisture recycling among nations • First, the introduction of the precipitationshed concept (Paper I), including the initial deﬁnition and description, represents the foundation for this dissertation. Given that the precipitationshed is like a watershed, it transforms the atmospheric branch of the water cycle from a purely geophysical phenomenon, to a phenomenon with traceable ﬂows of water, many of which people can change. It bears mentioning that since Paper I was published in 2012, the precipitationshed concept has been actively discussed in the land-atmosphere journal literature (Paper I has 33 citations, and Paper II has 6 citations, according to Web of Science as of August 2016). Several articles on moisture recycling processes and anthropogenic land cover change impacts to moisture recycling have discussed the precipitationshed concept (Gimeno et al., 2012; Goessling and Reick, 2013; Mahmood et al., 2013; Savenije et al., 2013, Wheater and Gober, 2013). Other researchers studying how land cover change can enhance precipitation downwind have acknowledged the utility of the concept for help- ing envision the upwind to downwind connections related to moisture transport (Makarieva et al. 2013; Keenan et al., 2013; Gerten 2013; van Noordwijk et al., 2014). As an extension of the original concept, the core precipitationshed

31 (Paper II) provides the conceptual basis for persistent moisture sources regions, and the quantitative justiﬁcation for precipitationshed science.

The second conceptual contribution regards the significant impact that ecosystems can and do have on moisture recycling processes, discussed in Paper III. As such, the conceptual framework of Vegetation-regulated Moisture Re- cycling (VMR) ecosystem services enables the explicit linking of the atmospheric water cycle to the biosphere. Though this work has only recently been published, this concept may end up being a significant contribution to the field of ecosystem services, since this kind of systemic connection between ecosystem functions, specifically in the context of moisture recycling, has hitherto not been included in ecosystem service assessments, inventories, and comparisons.

Thirdly, Paper V contains an approach for describing how countries exchange moisture with one another, allowing for a clearer integration of moisture recycling (and the precipitationshed speciﬁcally) into existing international governance. This typology of moisture recycling exchange among nations make it possible to quickly identify the types of governance approaches that may be best suited for a given region.

3.2 Methods

The three categories of methodological contribution I make in this dissertation are in: modeling advances • moisture recycling science • precipitationshed analytics • In terms of the modeling, my work has provided an approach for comparing two different datasets in the WAM-2layers, and improved the coupling between the STEAM and WAM-2layers models in a land-use change context.

First, I expanded the functionality of the WAM-2layers (Paper II) to use a different driving dataset (i.e. MERRA) from the one it was designed for (i.e. ERA-Interim). This was an important step in my own learning about the model, as well as for understanding how the WAM-2layers could be used in other research, including global climate model output. This process has also been constructive for subsequent partnerships and collaborations with other institutions that are applying the WAM-2layers in entirely new data contexts.

32 Second, I coupled the WAM-2layers to the STEAM model to simulate how land-use change affects moisture recycling. The original single-atmospheric- layer version of the WAM had been used in conjunction with output from another land-surface model to simulate how irrigation affects moisture recycling (Nikoli 2011). Van der Ent et al. (2014) used evaporation data generated by STEAM, and then used WAM-2layers to track this moisture. Finally, the initial coupling procedure for the STEAM+WAM was ﬁrst published in Paper III. Thus, my research built off of previous efforts to link land surface modeling to the WAM-2layers, by further developing scripts for the land-use change coupling procedure, as well as protocols for propagating the changes in moisture through the model.

The second methodological advance I made was the quantification of Vegetation- regulated Moisture Recycling (VMR) ecosystem services. This quantification is truly novel in that no other researcher has suggested anything similar to the calculations in Paper III. Likewise, the definition of VMR ecosystem services is independent of either data or model. Thus, other researchers who aim to analyze VMR ecosystem services may do so with their own data and their own moisture tracking procedure, providing that they have multiple land-use scenarios and can track both evaporation and precipitation on a spatially-explicit grid.

The third category of methods I developed were various precipitationshed analytics, including: 1. Multiple precipitationshed boundary calculations 2. Precipitationshed variability quantification 3. Seasonal comparison approach between dry and wet years A major component of the precipitationshed is the specification of the boundary, which defines a spatially explicit source region of evaporation. In Paper I, the boundary was defined as the surface area that contributed 70% of sink region precipitation. This approach was based on the fact that I wanted to include the sources that contributed more than two thirds of sink region precipitation. In Paper II, I modified this definition from a fraction of total precipitation, to a depth of contribution. This change was based on a desire to relate the precipitationshed to actual depths of evaporation that could be theoretically measured as precipitation in the sink region. For example, the 1-millimeter (mm) boundary was used in Paper II, which means that every surface that contributed 1mm of evaporation (calculated as the volume of evaporation di-

33 vided by the surface area of the 1.5 x 1.5 degree gridcell) was included within the precipitationshed boundary. This definition was used in the definition of the core precipitationshed referring to the region that contributes this minimum amount during every year of analysis. So, in Paper II, the boundary referred to any gridcell that contributed 1mm of evaporation during every year of analysis. In Paper V, I moved away from the gridcell based boundary definition, and adopted a ‘national contribution’ boundary, where the relevant precipitationshed is defined as including the nations that contributed 1% or more of sink nation precipitation. I want to emphasize that the different boundary definition methods each have their own strengths and weaknesses, and that the selection is important but also ought to be tailored to the questions being asked.

Along with the core precipitationshed being included in the boundary definition, it was also a component of the variability analysis in Paper II. The approach used to quantify the core precipitationshed, notably the persistent contribution of evaporation from a given region, was also a means to explore variability, since it helps identify the frequency of contribution for a given sink region’s precipitationshed (for more on this, see Figure 4 in Paper II). In addition to the simple metric of persistence, empirical orthogonal function (EOF) analysis was used to understand the key patterns of variability in the precipitationsheds. This was primarily done so that myself and others exploring the precipitationshed phenomena could be confident that the core precipitationshed boundaries were corroborated by the statistical variation in the precipitationshed. Indeed, the primary reason for variation in evaporation contribution within a precipitationshed was due to interannual pulsing of more (or less) evaporation, rather than a shifting from one location to another. Both the persistence and EOF analyses provided the justification for continuing precipitationshed science (i.e. Papers III, IV, and V).

The third precipitationshed analytic I developed, was the approach for comparing dry and wet year precipitationsheds (Paper IV). This method identiﬁes whether there are signiﬁcant differences between the evaporation contribution that was coming from land during either dry or wet years. The steps were: (1) identify the years that corresponded to dry, neutral, and wet years in the sink region, (2) calculate the average evaporation contribution associated with each of those categories, and (3) calculate the anomalous evaporation that occurred during either the dry or wet years. This method provided insights into which parts of the precipitationshed were particularly important during dry (or wet) rainfall years, which helped clarify whether there was increased vulnerability to land-use change during dry or wet years. This analysis fed directly into the quantitative vulnerability analysis, which I discuss in the next section.

34 3.3 Applications

Finally, there are three key applications I develop in this dissertation: Qualitative and quantitative precipitationshed vulnerability analyses • Vegetation-regulated Moisture Recycling (VMR) ecosystem service de- • scription and quantification Analysis of moisture recycling exchange among nations • First, the identification of a precipitationshed provides two key outputs: a spatially-explicit region of evaporation contribution to a specific location; and, a boundary that allows for a finite analysis of various factors, including vulnerabilities. Using the boundary I develop two kinds of vulnerability indices in this work, specifically: 1. An index based on a literature review of how vulnerable a given location’s precipitation is to various pressures (Paper I). 2. An index capturing how a given location’s moisture recycling might be affected by current and potential stresses (Paper IV). The qualitative index created for Paper I was initially meant as a proof-of- concept approach for how the precipitationshed might be used. This index, and the inclusion of land-use change impacts to the precipitationshed, has been highlighted as one of the primary contributions of Paper I (e.g. Gimeno et al., 2012; Lele et al., 2013; Mahmood et al., 2013; Savenije et al., 2014). The quantitative index created in Paper IV expands on the original index in Paper I, and highlights (a) the current (i.e. a priori) vulnerability of water resources for the sink region, (b) the current status of the precipitationshed, and (c) the potential for land-use change to impact moisture recycling. In this way, the index in Paper IV represents a quantification of moisture recycling vulnerability for a given location. The index was developed for use in the context of urban surface watersheds, but it could be applied more broadly assuming the specific driving datasets are available.

The Vegetation-regulated Moisture Recycling (VMR) ecosystem service analysis, from Paper III, that provides a method for both global and local calculations, could be a signiﬁcant contribution of this dissertation. In recent years, several different authors and research groups have converged on the idea that the evaporation arising from the land-surface, particularly from forests, ought to be considered an ecosystem service in terms of climate regulation and tenta- tively in terms of atmospheric moisture regulation (e.g. van Noordwijk et al.,

35 2014; Ellison et al., 2012). My work is, however, the first to provide (a) a clear definition of this service, (b) a method for its quantification, (c) a global and regional (i.e. case study) analysis, and (d) a conceptual framework for how this might integrate into broader ecosystem service assessments. Thus, the work in Paper III could be a clear departure point for moisture recycling to integrate with new disciplines.

Finally, the analysis of moisture recycling exchange among nations, described conceptually and quantified in Paper V, provides an approach for integrating moisture recycling science into international law and policy. Specifically, I provide a complete method of how to classify a given set of moisture recycling relationships among nations (or for a single nation) into a typology, which is then matched to different approaches for moisture recycling governance. Ultimately, there is much to be done to bridge science and policy regarding moisture recycling, but Paper V is a key first step on that path.

3.4 Trans-disciplinary Impact

Along with the specific conceptual, methodological, and applied contributions I make in my dissertation, many disciplines are impacted by my work. Table 2 below highlights these impacts. In surface hydrology, the watershed (a.k.a. catchment or basin) is the single common spatial unit that is used for distin- guishing related areas. In moisture recycling there has not been a similar concept that the scientific community has used. In general, though, when previous work has referred to ‘sources of evaporation’ they have used the same ideas, but different language to describe the spatial aspects. The precipitationshed (and the evaporationshed) could be a common platform or heuristic for the moisture recycling community to use. This would be beneficial for many rea- sons including: a common descriptive unit would make it easier to interrelate and compare different research; it would be more straightforward to follow or search out research on the topic; and, a scientific body of work could begin building around the precipitationshed. There could be challenges to the broader use of the precipitationshed, depending on the moisture recycling method being used, such as isotope tracking or Lagrangian modeling (e.g. Stohl et al., 1998), since the results of these analyses do not necessarily produce the same type of spatial data. However, there are many examples of similar types of results presented in other work, that would lend themselves quite easily to using the precipitationshed framework, notably much of the work by the widely used quasi-isentropic back trajectory

36 model (QIBT) that has been used in many other moisture recycling studies (e.g. Dirmeyer and Brubaker, 1999; Bagley et al., 2012; Wei et al., 2013). Furthermore, the number of people that are actively using the WAM-2layers is growing, which suggests there is a broader community that could also po- tentially use the precipitationshed (e.g. Zemp et al., 2014; Zhao et al., 2016; Duerinck et al., 2016).

Perhaps more important than providing a heuristic for the hydrology, atmo-

Table 2. Integrated insights from this dissertation for relevant ﬁelds.

DISCIPLINE NEW INSIGHTS a) Better understanding of atmospheric moisture source and sink dynamics b) New insights about what drives moisture recycling variability c) Refined appreciation of the role of vegetation for moisture recycling Hydrometeorology, regulation Atmospheric Science, d) Connection between large-scale geophysical phenomena and specific and Hydrology impacts e) Awareness of a broader research and practitioner community in which there exists an appetite for this science a) A new unit of analysis for exploring water stress & vulnerability (i.e. the precipitationshed) Water Resources b) An improved sense of the criticality of upwind land-cover and land-use Management c) Improved sense of dry season (and dry year), vulnerabilities for inflows to reservoirs a) Clearly defined moisture recycling as an ecosystem service related to vegetation-regulation (expanding on loose definitions related to climate regulation, etc.) Ecosystem Services b) New (or alternative) representations of how ecosystems provide ecosystem services c) Awareness of new, and quantified specific trade-offs that emerge (that were either unknown or ignored in the past a) Identified that the consequences of land-use change expand well beyond surface runoff effects, and extend into changes that can affect downwind rainfall Land-Change Science b) Evaporation is not a loss from the system, if the system domain is expanded c) Land-use change impacts not just the magnitude of moisture recycling, but also the timing of moisture recycling a) Water resources could be vulnerable to more than watershed characteristics and general climate variability, specifically land-use change in the precipitationshed Urban Studies b) There may be surprising opportunities to influence precipitation within the city and within the watershed via policies aimed at trade, land-use, conservation incentives, etc. that target activities in the precipitationshed a) There exist specific, trackable connections between hitherto separate actors b) Though regulation of activities, proof of harm, or enforcement of International Law and activities may be difficult to establish, other legal means may be possible Politics for justly and legally addressing how region's activities that affect one- another c) Institution-building and finding the 'right fit' will be key to successful management

37 spheric science, and climate communities, is the bridge that my work can build between social-ecological systems, land-use change, and moisture recycling. Given that the key socially-relevant insight of moisture recycling is that land- use change can modify rain elsewhere, it is critical to provide useful analytical tools to understand how land-use change interacts with moisture recycling. Similarly, since land-use change is a spatially explicit process, it is useful to have a consistent, spatially explicit method for evaluating subsequent impacts on moisture recycling. The precipitationshed and evaporationshed are these methods.

38 4. Limitations and Considerations

There are several aspects of the research that warrant a brief explanation in terms of limitations and additional considerations. First, I want to emphasize that the moisture tracking procedure that I use in this dissertation relies on input data and is integrated into one atmospheric layer in Paper I, and two atmospheric layers in Paper II thru Paper VI. The fact that the WAM-2layers uses input data means that it is not simulating any emergent features of the atmosphere, i.e. changes in circulation, changes in humidity, or changes in the energy budget. Other studies that have used fully dynamic and fully coupled land and atmosphere models reveal that for certain parts of the world these changes and impacts in the atmosphere matter more than others. For example, in the monsoon regions of south Asia land cover change has a demonstrated impact on the atmospheric circulation, and subsequently impacts precipitation downwind (e.g. Tuinenburg et al., 2014). However, other studies have found that the impact on circulation from land cover change can be modest in other regions, such as in the Amazon where changes to the atmospheric circulation from deforestation appear to be between 5 and 10% of mean annual winds (e.g. Bagley et al., 2014).

Also, an aspect of the climate system that is not incorporated into my analysis is how changes in land cover might influence the overall energy budget of the region that is being studied. Evaporation, which can also be understood as latent heat flux, has a counterpart in sensible heat flux. And as latent heat flux goes down (as a fraction of total heat loss), sensible heat flux increases. These changes in the fraction of heat flux that is either latent or sensible can and do have impacts for how precipitation forms and how the atmosphere cir- culates. And as before, these changes can be more or less important depending on where you are on the planet, and the spatial extent of the change. All this being said, several comparisons between our work and others’ have found that the modeling set-up I use continues to be robust to both qualitative and quantitative scrutiny (e.g. Bagley et al., 2012; van der Ent et al 2013; Zhao et al., 2016). This could be due to the fact that I tend to look at large spatial scales, and over long periods of time. The data on which I base my moisture recycling analyses (i.e. the ERA-Interim and MERRA reanalysis datasets, discussed ear-

39 lier in the Methods section) tend to represent the broad features of the climate system reasonably well. There are some problems associated with accurate representation of precipitation, and this is most apparent in terms of very small amounts of precipitation or intense precipitation. However, since my analyses are primarily based on large-scale, mean features of both evaporation and precipitation I am conﬁdent my results are robust.

Another important consideration is that the climate system is likely to be strongly inﬂuenced by global climate change both in the coming decades, and especially towards the latter half of the 21st century. As a result, in the context of discussions related to management or governance, it is critical that the stake- holders and scientists involved in such questions carefully discern the scope of their questions, to identify the best models and data to be used. This includes both spatial and temporal scope, but also the granularity and precision of the processes they wish to represent. I note that there are certain scales that are simply inappropriate for use with moisture recycling or precipitationshed analysis, such as the scale of a farmer’s ﬁeld or even that of a city, since other meteorological processes will dominate any moisture recycling effects.

40 5. Next Steps

There are many directions that this research can go, depending on the discipline. From the perspective of traditional atmospheric and hydrometeoro- logical research, there continue to be dozens of moisture recycling studies every year. Thus, there is unlikely to be a shortage of foundational science on the understanding of moisture recycling processes in the coming decades. Furthermore, there has been a steady increase during the last decade in high- resolution, coupled land and atmosphere modeling, with a particular focus on moisture recycling processes (e.g. Bagley et al., 2014; Swann et al., 2015). Likewise the studies have focused on understanding the impacts of various land cover change scenarios on moisture recycling. A research area that remains relatively open is understanding how future climate change will impact moisture recycling. Since climate change is projected to have signiﬁcant impacts on the very large-scale features of the atmosphere, such as where the storm tracks are located, this will have fundamental impacts on moisture recycling patterns. Understanding how climate change interacts with land-use changes, and subsequently moisture recycling will be a key area of research.

There remains, a large gap in understanding how land cover change (or climate change) induces moisture recycling changes that impact the stability of downwind and tele-coupled societies. That is to say, although there is a great deal of research and activity understanding the bio-geophysical processes, there is much less emphasis on the social-ecological consequences that arise from systematic changes in the atmospheric water cycle. In this way, a future direc- tion of this research is to better understand the interaction of changes in moisture recycling with social-ecological processes that are the recipients of these changes. As I have found in my own work in this dissertation, working across disciplines will be necessary to answer these types of questions.

One pathway forward in this area of research would be to use the STEAM + WAM modeling framework to simulate how land cover change impacts moisture recycling processes, and precipitation downwind. Likewise, this analysis could be complemented by a dynamic land and atmosphere model that can simulate the interactions between the surface of the earth and the climate sys-

41 tem. This comparative multi-method approach would assist in quantifying the validity of STEAM+WAM more effectively than has already been done, and could reveal new insights about which processes are most important in a moisture recycling context. With the output of this comparative study, the ranges of precipitation impacts that are likely to result from the prescribed changes in land-use could be identiﬁed. The truly novel component of this next step would be to integrate these changes and precipitation with a systems model that represents the feedbacks and interactions in the recipient social-ecological systems. By using the range of precipitation impacts that result from land-use change, it may be possible to identify whether or not tipping points emerge in the recipient social-ecological systems across the spectrum of precipitation impacts.

42 6. Advice to Future PhD Students

During my PhD, I have gleaned some important insights into conducting research on moisture recycling... but also into research in general. Here are some of those insights. 1. Maintain steady communication with your advisor on the goals of your research. It’s easy to get side-tracked, especially early on in your PhD. Your advisor has a sense of where the research ought to go, and how the individual PhD fits into the broader field. 2. Listen to your advisor, as well as yourself. If you think there’s a promis- ing research path try and pursue it. There are few instances in the future where you’ll have that opportunity. In my own research, Paper II would not have happened without this self-motivation, and ultimately the variability analysis and approach represents a very important justification for precipitationshed science as a whole. Once I convinced my advisor of the value of Paper II she was onboard, but it took a little bit of my own self-motivation to get us there. 3. Pursue interesting ideas, and explore new methods, since you never know where they’ll end up. The PhD is a rare time where you develop enough skills to (a) ask interesting questions, (b) conduct really interesting work, and (c) have very few other responsibilities. Thus, take advan- tage of this time and dive deeply into your topic. Likewise, take advan- tage of the opportunity to learn new things. In the context of my own research, that meant learning a lot more about Matlab, interpreting new datasets, coupling models together, and working in trans-disciplinary teams. Papers III and IV were simply interesting ideas that I pursued, which both eventually evolved into complete pieces of work. In addition to these three thoughts, my PhD ended up being very trans-disciplinary. This was somewhat by design, given that from the outset I wanted to write the final paper about moisture recycling governance. However, in practice this was not a natural path, at least for most PhD students. In terms of my own background (natural science) and moisture recycling (geophysics), I did not have a familiarity with many of the topics that are important when engaging in

43 transdisciplinary research, perhaps most importantly an individual philosophy of science. I credit much of my own advancement in transdisciplinary work to the community at the Stockholm Resilience Centre (SRC), which fostered (at least in me) an appreciation for new ideas and broad thinking. The tips I would give to anyone who wants to engage in trans-disciplinary work are the following: 1. Test your ideas often, particularly with people that will probably disagree with you. It’s uncomfortable to share things with your peers, and even more so when you might be told that the idea is wrong, ill- conceived, or silly. Deal with it. As a PhD student you must learn how to develop new ideas, communicate these ideas, and handle legiti- mate (and illegitimate) criticism. At the SRC, the best opportunities for this are speedtalks, brown-bag lunches, and informal seminars. In the context of transdisciplinary work, it is important to get used to feeling uncomfortable. 2. Listen to other’s ideas, especially if their methods or topics are different from your own. This is a direct corollary of the first tip, however it relates more to strategically broadening your research worldview. This means making time to attend speedtalks, brown-bag lunches, and informal seminars that others are giving, on a wide range of topics. Likewise, both conferences and visiting other institutions are critical to having your academic worldview challenged. 3. Surround yourself with people who are smarter than you, especially in the fields you know little about. I heard once that you are the average of the five people with whom you spend the most time”, but I can’t remember who said it. I think this can be extended to research, and in the context of trans-disciplinary research it means that you must seek out the people who you want to be like, and then wiggle your way into their domain of influence. This might mean strategic advisor selection, joining committees, or asking your advisor to make connections for you. 4. Read. A lot. Of course as a PhD student, you must read a lot of journal literature. This is not what I mean here. I mean that as a transdisciplinary scientist, you must broaden your horizons by reading about topics that are outside your immediate research area, but nonetheless represent excellent writing. This can be nonfiction or fiction, print or digital, news or opinion. Whatever it is you do read, make sure it’s high quality writing.

44 5. Find people with whom you disagree and learn why. One of the key rea- sons I was able to get so much from my PhD, was the oft times heated debates in which myself and my PhD colleagues would engage. Ulti- mately, we ended up agreeing about many things, but the road to this agreement was paved with lots of introspection, exploration, and discussion. I would expect anyone who wishes to attain a Doctor of Philosophy in any discipline to be able to fruitfully engage with those they disagree with. In transdisciplinary research, a plasticity of mind is essential.

45 46 7. Concluding Remarks

The concept of the precipitationshed has been explored in great detail and in many different ways within this dissertation. Since the initial publication of the precipitationshed concept, it has been critiqued and discussed in the academic literature. Along with that, there has emerged an appetite in the practitioner community to have a better understanding of the tools and resources that are available for understanding the relationship between changes in land cover and eventual precipitation. Thus, I hope that the study of the precipitationshed, and how social-ecological systems interacting with moisture recycling more broadly, persists within the academic community and leads to ever improved understanding. Specifically, the scientific community ought to continue to investigate the quantitative nature of the precipitationshed such as the variability and seasonal differences, as well as a qualitative features of the precipitationshed since it is an inherently socially relevant phenomena. Ultimately for this concept to be expanded and incorporated into broader hydrological thinking, hydrologists and water resources scientists who are typically concerned with surface water will need to start connecting their branch of the water cycle to the atmospheric branch. Finally, this work is critical given that global environmental and anthropogenic forces are leading to increasingly unpredictable changes. The precipitationshed is a key contribution to a scientific toolkit for moving towards a sustainable future for humanity in the Anthropocene.

47 48 8. References

Badger, A. M., & Dirmeyer, P. A. (2015). Climate response to Amazon forest replacement by heterogeneous crop cover. Hydrology and Earth System Sci- ences, 19(11), 4547-4557.

Bagley, J. E., Desai, A. R., Dirmeyer, P. A, & Foley, J. A. (2012). Effects of land cover change on moisture availability and potential crop yield in the world’s breadbaskets. Environmental Research Letters, 7(1), 014009.

Bagley, J. E., Desai, A. R., Harding, K. J., Snyder, P. K., & Foley, J. A. (2014). Drought and deforestation: has land cover change inﬂuenced recent precipitation extremes in the Amazon? Journal of Climate, 27(1), 345-361.

Bosilovich, M. G., & Chern, J. D. (2006). Simulation of water sources and precipitation recycling for the MacKenzie, Mississippi, and Amazon River basins. Journal of Hydrometeorology, 7(3), 312-329.

Bosilovich, M. G., Robertson, F. R., & Chen, J. (2011). Global energy and water budgets in MERRA. Journal of Climate, 24(22), 5721-5739.

Charney, J., Quirk, W. J., Chow, S. H., & Kornﬁeld, J. (1973). A comparative study of the effects of albedo change on drought in semi-arid regions. Journal of the Atmospheric Sciences, 34(9), 1366-1385.

Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., ... & Bechtold, P. (2011). The ERA-Interim reanalysis: Conﬁguration and performance of the data assimilation system. Quarterly Journal of the Royal Meteorological Society, 137(656), 553-597.

Dirmeyer, P. A., & Brubaker, K. L. (1999). Contrasting evaporative moisture sources during the drought of 1988 and the ﬂood of 1993. Journal of Geophys- ical Research: Atmospheres, 104(D16), 19383-19397.

Dirmeyer, P. A., Brubaker, K. L., & DelSole, T. (2009). Import and export

49 of atmospheric water vapor between nations. Journal of Hydrology, 365(1), 11-22.

Dominguez, F., Kumar, P., Liang, X. Z., & Ting, M. (2006). Impact of atmospheric moisture storage on precipitation recycling. Journal of Climate, 19(8), 1513-1530.

Duerinck, H. M., van der Ent, R. J., van de Giesen, N. C., Schoups, G., Babovic, V., & Yeh, P. J. F. (2016). Observed Soil Moisture-Precipitation Feedback in Illinois: A Systematic Analysis over Different Scales. Journal of Hydrometeorology, (2016).

Ellis, E. C., Klein Goldewijk, K., Siebert, S., Lightman, D., & Ramankutty, N. (2010). Anthropogenic transformation of the biomes, 1700 to 2000. Global Ecology and Biogeography, 19(5), 589-606.

Ellison, D., N. Futter, M., & Bishop, K. (2012). On the forest cover-water yield debate: from demand- to supply-side thinking. Global Change Biology, 18(3), 806-820.

Folke, C., Hahn, T., Olsson, P., & Norberg, J. (2005). Adaptive governance of social-ecological systems. Annual Review of Environmental Resources, 30, 441-473.

Friedl, M. A., Sulla-Menashe, D., Tan, B., Schneider, A., Ramankutty, N., Sibley, A., & Huang, X. (2010). MODIS Collection 5 global land cover: Al- gorithm reﬁnements and characterization of new datasets. Remote Sensing of Environment, 114(1), 168-182.

Gerten, D. (2013). A vital link: water and vegetation in the Anthropocene. Hydrology and Earth System Sciences, 17(10), 3841-3852.

Giannini, A., Saravanan, R., & Chang, P. (2003). Oceanic forcing of Sahel rainfall on interannual to interdecadal time scales. Science, 302(5647), 1027- 1030.

Giannini, A. (2016). 40 Years of Climate Modeling: The Causes of Late-20th Century Drought in the Sahel. In The End of Desertiﬁcation? (pp. 265-291). Springer Berlin Heidelberg.

Gimeno, L., Stohl, A., Trigo, R. M., Dominguez, F., Yoshimura, K., Yu, L., ...

50 & Nieto, R. (2012). Oceanic and terrestrial sources of continental precipitation. Reviews of Geophysics, 50(4).

Goessling, H., & Reick, C. H. (2013). On the" well-mixed" assumption and numerical 2-D tracing of atmospheric moisture. Atmospheric Chemistry and Physics, 13, 5567-5585.

Halder, S., Saha, S. K., Dirmeyer, P. A., Chase, T. N., and Goswami, B. N. (2016). Investigating the impact of land-use land-cover change on Indian sum- mer monsoon daily rainfall and temperature during 1951-2005 using a regional climate model, Hydrology and Earth System Science, 20, 1765-1784.

Keenan, T. F., Hollinger, D. Y., Bohrer, G., Dragoni, D., Munger, J. W., Schmid, H. P., & Richardson, A. D. (2013). Increase in forest water-use efﬁciency as atmospheric carbon dioxide concentrations rise. Nature, 499(7458), 324-327.

Koster, R., Jouzel, J., Suozzo, R., Russell, G., Broecker, W., Rind, D., & Ea- gleson, P. (1986). Global sources of local precipitation as determined by the NASA/GISS GCM. Geophysical Research Letters, 13(2), 121-124.

Lawrence, D., & Vandecar, K. (2015). Effects of tropical deforestation on climate and agriculture. Nature climate change, 5(1), 27-36.

Lele, U., Karsenty, A., Benson, C., Féstiveau, J., Agarwal, M., & Goswami, S. (2012). Background Paper 2: Changing Roles of Forests and their Cross- Sectorial Linkages in the Course of Economic Development. In Draft. United Nations Forum for Forests.

Lettau, H., Lettau, K., & Molion, L. C. B. (1979). Amazonia’s hydrologic cycle and the role of atmospheric recycling in assessing deforestation effects. Monthly Weather Review, 107(3), 227-238.

Linsley Jr, R. K., Kohler, M. A., & Paulhus, J. LH(1982). Hydrology for Engineers. McGraw-Hill Series in Water Resources and Environmental Engi- neering (pp. 508): McGraw-Hill, Inc.

Lo, M. H., & Famiglietti, J. S. (2013). Irrigation in California’s Central Valley strengthens the southwestern US water cycle. Geophysical Research Letters, 40(2), 301-306.

Mahmood, R., Pielke, R. A., Hubbard, K. G., Niyogi, D., Dirmeyer, P. A.,

51 McAlpine, C., ... & Baker, B. (2014). Land cover changes and their bio- geophysical effects on climate. International Journal of Climatology, 34(4), 929-953.

Makarieva, A. M., Gorshkov, V. G., & Li, B. L. (2013). Revisiting forest impact on atmospheric water vapor transport and precipitation. Theoretical and applied climatology, 111(1-2), 79-96.

Molden, D. (2007). Water for food, water for life. Earthscan, London, and International Water Management Institute, Colombo.

Nikoli, R. Moisture recycling and the effect of land-use change. Masters The- sis. TU Delft, 2011.

Otterman, J. (1974). Baring high-albedo soils by overgrazing- A hypothesized desertiﬁcation mechanism. Science, 186(4163), 531-533.

Portmann, F. T., Siebert, S., & Döll, P. (2010). MIRCA2000-Global monthly irrigated and rainfed crop areas around the year 2000: A new high-resolution data set for agricultural and hydrological modeling. Global Biogeochemical Cycles, 24(1).

Reisner, M. (1993). Cadillac desert: The American West and its disappear- ing water. Penguin.

Salati, E., Dall’Olio, A., Matsui, E., & Gat, J. R. (1979). Recycling of water in the Amazon basin: an isotopic study. Water Resources Research, 15(5), 1250-1258.

Salih, A. A. M., Zhang, Q., & Tjernström, M. (2015). Lagrangian tracing of Sahelian Sudan moisture sources. Journal of Geophysical Research: Atmo- spheres, 120(14), 6793-6808.

Savenije, H. H. G., Hoekstra, A. Y., & Van der Zaag, P. (2014). Evolving water science in the Anthropocene. Hydrology and earth system sciences, 18(1), 319-332.

Savenije, H. H. (2004). The importance of interception and why we should delete the term evapotranspiration from our vocabulary. Hydrological Pro- cesses, 18(8), 1507-1511.

52 Spracklen, D. V., & Garcia-Carreras, L. (2015). The impact of Amazonian deforestation on Amazon basin rainfall. Geophysical Research Letters, 42(21), 9546-9552.

Stohl, A., Hittenberger, M., & Wotawa, G. (1998). Validation of the lagrangian particle dispersion model FLEXPART against large-scale tracer experiment data. Atmospheric Environment, 32(24), 4245-4264.

Swann, A. L., Longo, M., Knox, R. G., Lee, E., & Moorcroft, P. R. (2015). Future deforestation in the Amazon and consequences for South American climate. Agricultural and Forest Meteorology, 214, 12-24.

Trenberth, K. E., Fasullo, J. T., & Mackaro, J. (2011). Atmospheric moisture transports from ocean to land and global energy ﬂows in reanalyses. Journal of Climate, 24(18), 4907-4924.

Tuinenburg, O. A. (2013). Atmospheric effects of irrigation in monsoon climate: the Indian subcontinent. PhD Thesis, Wageningen University, Wagenin- gen, NL

Tuinenburg, O. A., Hutjes, R. W. A., Stacke, T., Wiltshire, A., & Lucas-Picher, P. (2014). Effects of irrigation in India on the atmospheric water budget. Jour- nal of Hydrometeorology, 15(3), 1028-1050.

Van der Ent, R. J., Savenije, H. H. G., Schaeﬂi, B., & Steele-Dunne, S. C. (2010). Origin and fate of atmospheric moisture over continents. Water Re- sources Research, 46(9).

Van der Ent, R. J., Tuinenburg, O. A., Knoche, H.-R., Kunstmann, H., & Savenije, H. H. G. (2013). Should we use a simple or complex model for moisture recycling and atmospheric moisture tracking? Hydrology and Earth System Sciences, 17(12), 4869-4884.

Van der Ent, R. J. (2014), A new view on the hydrological cycle over continents, Ph.D. thesis, 96 pp, Delft University of Technology, Delft. http://dx.doi.org/10.4233/uuid:0ab824ee-6956-4cc3-b530-3245ab4f32be.

Van der Ent, R. J., Wang-Erlandsson, L., Keys, P. W., & Savenije, H. H. G. (2014). Contrasting roles of interception and transpiration in the hydrological cycle-Part 2: Moisture recycling. Earth System Dynamics, 5(2), 471.

53 Van Noordwijk, M., Namirembe, S., Catacutan, D., Williamson, D., & Ge- brekirstos, A. (2014). Pricing rainbow, green, blue and grey water: tree cover and geopolitics of climatic teleconnections. Current Opinion in Environmental Sustainability, 6, 41-47.

Wang-Erlandsson, L., Van Der Ent, R. J., Gordon, L. J., & Savenije, H. H. G. (2014). Contrasting roles of interception and transpiration in the hydrological cycle-Part 1: Temporal characteristics over land. Earth System Dynamics, 5(2), 441.

Wei, J., Dirmeyer, P. A., Wisser, D., Bosilovich, M. G., & Mocko, D. M. (2013). Where Does the Irrigation Water Go? An Estimate of the Contribution of Irrigation to Precipitation Using MERRA. Journal of Hydrometeorology, 14(1), 275-289.

Wheater, H., & Gober, P. (2013). Water security in the Canadian Prairies: science and management challenges. Philosophical Transactions of the Royal So- ciety of London A: Mathematical, Physical and Engineering Sciences, 371(2002), 20120409.

Zemp, D. C., Schleussner, C. F., Barbosa, H. M. J., Van der Ent, R. J., Donges, J. F., Heinke, J., ... & Rammig, A. (2014). On the importance of cascading moisture recycling in South America. Atmospheric Chemistry and Physics, 14(23), 13337-13359.

Zhao, T., Zhao, J., Hu, H., & Ni, G. (2016). Source of atmospheric moisture and precipitation over China’s major river basins. Frontiers of Earth Science, 10(1), 159-170.