40 SCIENCE HIGHLIGHTS: Future of past global change research

doi.org/10.22498/pages.29.1.40 The enigma and complexity of landscape dynamics in Chinese : From case studies to big data Peng Liang, H. Li, Y. Zhou, X. Fu, L. Mackenzie and D. Zhang

We analyze the recent progress in eolian surface processes and landscape dynamics in Chinese deserts. These diverse eolian studies, including paleoenvironmental reconstructions, advances in dating techniques, and clarification of sediment provenance, highlight the complexity of landscape evolution.

Drylands occupy ~41% of the global land stabilization processes are usually a pathway is more complex in western Chinese surface and are home to more than two bil- landscape response to increased precipi- deserts (Feng and Yang 2019). lion people (Reynolds et al. 2007). In recent tation associated with enhanced summer decades, global warming and intensified monsoons, while increased eolian activity Spatial heterogeneity of the landscape human activities have exacerbated the en- and dune activation occurs during periods of leads to an unavoidable uncertainty when vironmental degradation or drought. These alternating sequences are a interpreting geological signals from eolian in drylands, threatening nations' economies direct and sensitive record of past monsoon stratigraphic sequences. Liang and Yang and sustainable development. Additionally, variability. Yang et al. (2019) found that a (2016) investigated drivers of landscape deserts are an important yet poorly under- dark brown (Munsell soil color 10YR 4/3) heterogeneity at different scales in the stood component of the Earth system, with paleosol began to develop at 14.5 thousand Maowusu Sandy Land, northern dust released from these regions affect- years before present (kyr BP) and lasted until (Fig. 1). They found that climate and large- ing the global biogeochemical cycle and 2 kyr BP in the Hulunbeier Sandy Land, while scale agricultural reclamation affect regional climate. Early-career researchers (ECRs) at stronger pedogenesis (i.e. soil formation) oc- landscape patterns in the Maowusu Sandy Zhejiang University and elsewhere in China curred during 9-5 kyr BP. Eolian sequences Land, whereas microtopography and river are currently investigating sedimentary from the Hunshandake Sandy Land recorded networks drive landscape heterogeneity at archives from the interior of Chinese deserts dune stabilization processes from 9.6 to a local scale. The landscape response to de- in the eastern Asian desert belt (Fig. 1) to ex- 3.0 kyr BP, though localized eolian events clining wind strength in the Maowusu Sandy plore complex landscape dynamics through occurred at the same time (Fig. 2). Although Land from 1981 to 2016 shows a significantly time. the paleolandscape in the eastern sandy out-of-step pattern between the western lands shows a high degree of spatial het- and eastern regions, arising from different Paleoenvironmental signals archived erogeneity, were generally stabilized regional climates and land-use histories in the eastern Chinese deserts and eolian activity was suppressed from 7.5 (Liang and Yang 2016), highlighting the com- The large semi-stabilized sandy lands in to 3.5 kyr BP in the mid-Holocene (Fig. 2; plexity of landscape dynamics in drylands. the deserts of eastern China are located Yang et al. 2019). These geological records near the northern extent of the East Asian from dune sequences are generally consis- The eolian processes at the dune scale also summer monsoon (Fig. 1). Stratigraphic tent with paleoclimate simulations, which play an important role in the interpreta- sequences from the sandy lands record indicate that northern China received higher tion of paleoenvironmental records from repeated periods of dune activation and sta- summer monsoon precipitation during the dune deposits. The dune stabilization bilization through alternating eolian (wind- mid-Holocene than during the pre-industrial process is assumed to be mainly caused by blown) sands and paleosols (buried soils). period, although the moisture transport precipitation-induced vegetation expansion.

Figure 1: Distribution of Asian deserts, major rivers, and published data in the growing database. The thick dashed line indicates the boundary of the Asian summer monsoon. Active sand seas: Taklamakan (TK); Kumtagh (KT); Chaidamu (CD); Badain Jaran (BJ); Tengger (TG); Wulanbuhe (WB). Semi-stabilized sandy lands: Gurbantunggut (GT); Maowusu (MS); Kubuqi (KB); Hunshandake (HD); Horqin (HQ); Hulunbeier (HB).

PAGES MAGAZINE ∙ VOLUME 29 ∙ NO 1 ∙ May 2021 CC-BY SCIENCE HIGHLIGHTS: Future of past global change research 41

reconstructions based on single-site dune deposits or sequences alternating between lacustrine, eolian sands, and paleosols in deserts inevitably contain uncertainties. These uncertainties mainly arise from the spatial heterogeneity of the eolian land- scape (Liang and Yang 2016), the episodic/ discontinuous eolian sand deposition fea- tures with possible eolian erosion (Forman 2015), and the generally non-linear response between eolian depositional processes and climate fluctuations (Yang et al. 2019). A big-data concept incorporating a conti- nental-scale database using the substantial paleoenvironmental records from Chinese deserts could be introduced to overcome these difficulties and complexities. This data- base is currently under construction (Fig. 1) Figure 2: Eolian events represented by the frequency of optically stimulated luminescence (OSL) ages from the eastern sandy lands of northern China. The gray rectangles are OSL age histograms, and the bin size is 0.5 kyr. and will include sedimentary sequences that Data are from Li and Yang (2016) and Yang et al. (2019) with new data added. contain well-vetted geomorphic context information, stratigraphic descriptions, prox- However, a new case study investigating and late Holocene but are mostly dormant ies (such as grain size, magnetic susceptibil- the transition from barchan (crescent- during the mid-Holocene (Fig. 2). However, ity), and relevant ages. A comprehensive and shaped sand dune) to parabolic dunes in the our understanding of eolian activity at the well-organized database that includes mul- Maowusu Sandy Land demonstrated that glacial-interglacial timescale is still unclear tiple physical and chemical indices of surface the reduction of wind strength can lower the due to the lack of well-preserved and statisti- dune sand, such as grain size, geochemical sand flux rate and dune height, which allows cally meaningful archives older than 20 kyr composition, and petrology, is also required for vegetation establishment and dune trans- BP from desert interiors, limiting regional to advance sand provenance studies. These formation (Zhang et al. 2020). This research multi-site paleoenvironmental reconstruc- increasingly large and high-dimensional suggests that some mismatches between tion (Li and Yang 2016). datasets and data-driven computations dune activity and moisture variability could are a promising avenue to enhance our be reconciled through a better understand- Sediment provenance and understanding of eolian processes and the ing of past wind regimes. surface processes Earth system from a big-data perspective, Identifying sediment provenance can yield especially with the aid of machine learning Improved dating techniques for insights into understanding the complexity algorithms. However, more in-depth field paleoenvironmental reconstructions of past and present landscape dynamics in studies are still indispensable. Over the past 20 years, optically stimulated deserts. Sediment sources in the Taklamakan luminescence (OSL) dating has become Desert, , , ACKNOWLEDGEMENTS a well-established Quaternary geochro- and Maowusu Sandy Land were investigated Our current work was supported mainly by the Ministry of nometer, particularly for eolian sediments, by combining geochemical compositions Science & Technology of China (2017FY101001) and the and is arguably the most important tool of the sand with geomorphic analysis (Hu National Natural Science Foundation of China (41672182; for desert paleoenvironmental research. and Yang 2016; Liu and Yang 2018; Jiang 42001003; 41430532). Sincere thanks are extended to A series of luminescence dating proce- and Yang 2019; Zhou et al. 2020). Results the early-career researchers whose work we have cited, dures for K-feldspar, a ubiquitous mineral show that the dust fraction (<63 μm) in dune including Dr. Fangen Hu, Qianqian Liu, Qida Jiang and in natural sediments, has recently been sands from the varies Yingying Feng. We also thank Dr. Ziting Liu and Wancang Zhao for enriching the ongoing database. developed using eolian and fluvial samples from 0.44% to 21.7% and can be traced to from Chinese deserts (Fu et al. 2015; Fu et the Kunlun and Tianshan Mountains by their AFFILIATION al. 2018). These techniques considerably geochemical and sedimentological charac- School of Earth Sciences, Zhejiang University, extend upper and lower dating limits and teristics. However, the sand particles (>63 Hangzhou, China improve dating accuracy. The ability to μm) were predominantly sourced from the CONTACT analyze raw luminescence dating data has Kunlun Mountains in the south and trans- Peng Liang: [email protected] also been enhanced by a newly developed ported via fluvial processes (Jiang and Yang software "Luminescence Dose and Age 2019; Zhou et al. 2020). These results sug- REFERENCES Calculator (LDAC v1.0)" (Liang and Forman gest that dust particles within deserts have Adams SM, Soreghan GS (2020) Geology 48: 1105-1109 2019; https://github.com/Peng-Liang/LDAC), independent provenance, which is consis- Feng Y, Yang X (2019) J Geogr Sci 29: 2101-2121 which can maintain, archive, and synthesize tent with the low dust-generation potential Forman SL (2015) Front Earth Sci 3: 3 basic OSL data, apply appropriate statistical from sand saltation and wind abrasion found Fu X et al. (2015) Quat Geochronol 30: 161-167 models, calculate environmental dose rate, in wind-tunnel experiments (Adams and Fu X et al. (2018) Quat Geochronol 47: 1-13 and render statistically significant final ages. Soreghan 2020). Similarly, fluvial processes This self-contained tool for luminescence provide sand to the Badain Jaran Desert (Hu Hu F, Yang X (2016) Quat Sci Rev 131, Part A: 179-192 dating allows for inter-laboratory OSL age and Yang 2016), the Kubuqi Desert, and the Jiang Q, Yang X (2019) J Geophys Res Earth Surf 124: comparisons and promotes more robust Maowusu Sandy Land (Liu and Yang 2018) by 1217-1237 datasets for landscape-dynamics research in transporting loose sediments from nearby Li H, Yang X (2016) Quat Int 410: 58-68 drylands and beyond. OSL dating of eolian mountains to the desert basins. These Liang P, Forman SL (2019) Ancient TL 37: 21-40 sands has produced over 300 age records primary dune-building sediments are then Liang P, Yang X (2016) Catena 145: 321-333 from dunes in China, which were recently further mixed via local eolian processes (Liu Liu Q, Yang X (2018) Geomorphology 318: 354-374 compiled as part of the INQUA Dunes Atlas and Yang 2020). chronologic database (Li and Yang 2016; Liu Q, Yang X (2020) J Desert Res 40: 158-168 https://www.dri.edu/inquadunesatlas/). Future work: moving to big data Reynolds JF et al. (2007) Science 316: 847-851 This primary dataset collated by ECRs has The aforementioned case studies have Yang X et al. (2019) Sci China Earth Sci 62: 1302-1315 helped to build a picture of the eolian greatly enriched our understanding of the Zhang D et al. (2020) Earth Surf Process Landf 45: history in Chinese deserts since the Last landscape dynamics and surface processes 2300-2313 Glacial Maximum, showing that the eolian in deserts of China and beyond, but a more Zhou Y et al. (2020) Geomorphology 375: 107560 events identified by the frequency of OSL comprehensive and continental-scale ages increased during the last deglaciation picture is still lacking. Paleoenvironmental

CC-BY PAGES MAGAZINE ∙ VOLUME 29 ∙ NO 1 ∙ May 2021