{ "id": "loess-plateau", "title": "Loess Plateau", "lang": "en", "source_lang": "en", "summary": "A major upland region of northern China, distinguished by deep windblown sediments, dissected terrain, early settlement, and extensive ecological restoration.", "category": "geography", "subcategory": "", "type": "place", "aliases": [ "Chinese Loess Plateau", "Huangtu Plateau", "Huangtu Gaoyuan" ], "generated_at": "2026-10-22", "model": "pro/gpt-6.1-sol", "url": "https://aiwiki.page/en/loess-plateau/", "translations": { "zh-Hans": { "url": "https://aiwiki.page/zh/loess-plateau/", "json": "https://aiwiki.page/zh/loess-plateau.json" } }, "sections": [ { "heading": "", "text": "The Loess Plateau is an extensive upland region of north-central China, associated with the upper and middle reaches of the Yellow River. In a widely used broad delimitation, it covers approximately 640,000 square kilometres. Its defining material is loess, a predominantly silt-sized sediment deposited by wind, which blankets much of the underlying landscape. Thick sediment sequences and deeply dissected tablelands, ridges, and hills make the region one of the world’s most extensive and distinctive loess landscapes. (documents1.worldbank.org)\n\nThe plateau is important both as a record of environmental change and as a landscape shaped by human occupation. Its alternating loess and ancient soil layers preserve evidence of Asian monsoon variability and glacial–interglacial cycles, while archaeological discoveries document a human presence extending into the Early Pleistocene. Its readily eroded sediments also contribute substantially to the Yellow River’s sediment load, linking upland land use to downstream river conditions. (nature.com)\n\nSince the late twentieth century, the plateau has become a major setting for ecological restoration, combining watershed engineering, changes in agricultural land use, and vegetation recovery. These interventions have reduced land degradation in project areas, but their wider effects require attention to water availability, vegetation type, and differences between local and regional processes. The region consequently connects questions of geological history, settlement, rural livelihoods, and the long-term management of water-limited landscapes. (documents1.worldbank.org)" }, { "heading": "Geographic extent and regional divisions", "text": "The Loess Plateau occupies north-central China, principally within the upper and middle reaches of the Yellow River drainage basin. Its broad regional extent is commonly given as approximately 640,000 square kilometres. This figure describes a geographical region containing loess-covered uplands, mountains, valleys, and plains, rather than an uninterrupted sheet of loess. Three spatial units therefore need to be distinguished: the physical plateau region, the mapped distribution of loess deposits, and the areas selected for watershed rehabilitation or other administrative programmes. These boundaries are not interchangeable. (documents1.worldbank.org)\n\nThe Taihang Mountains provide the principal eastern reference boundary, beyond which the terrain descends towards the North China Plain; the Qinling Mountains delimit the region to the south. Western limits are commonly placed near the Riyue Mountains of Qinghai and the Helan Mountains farther north, adjoining the northeastern margins of the Tibetan Plateau. Northern limits vary more substantially. A Chinese Academy of Sciences expedition conducted in 1955–1958 described a predominantly geomorphological region extending north to the Great Wall, whereas broader regional definitions extend to the Yinshan Mountains and incorporate adjoining northern uplands. Thus, “north of the Qinling and west of the Taihang” identifies the main setting but does not establish a uniquely agreed perimeter. (geodoi.ac.cn)\n\nDifferences in delineation account for some of the variation in published area estimates. A county-based boundary dataset associated with the Chinese Academy of Sciences’ 1984–1990 integrated regional survey covers 624,641 square kilometres, compared with the rounded 640,000-square-kilometre figure used by the World Bank. The former follows selected county boundaries and therefore represents a regional planning unit rather than a precisely traced landform edge. Across these broad definitions, the plateau extends through Shanxi, northern and central Shaanxi, eastern Gansu, and Ningxia, with adjoining areas in Inner Mongolia, eastern Qinghai, and western Henan. Provincial membership does not imply inclusion of an entire province: administrative borders cross physical transitions between uplands, mountain ranges, and basins. (geodoi.ac.cn)\n\nRelief generally descends from northwest to southeast, but the plateau is neither a single level surface nor a uniformly loess-covered upland. Elevations of approximately 1,000–2,000 metres are characteristic of extensive upland districts, while river valleys lie lower and mountain ridges rise above them. Within this regional framework, relatively broad, dissected tablelands contrast with densely gullied hills and ridges. The World Bank’s 2004 project description, for example, identifies gullied tableland landscapes in Gansu and rounded, deeply dissected hill landscapes in Shanxi, Shaanxi, and parts of Inner Mongolia. Broader regional classifications also distinguish river-valley plains, sandy northern margins, and rocky mountain districts; these are constituent landscapes, not exceptions that must be excluded from the plateau boundary. (sciencedirect.com)\n\nRestoration boundaries select much smaller areas within this geographical mosaic. The first World Bank-supported Loess Plateau Watershed Rehabilitation Project, implemented from 1994 to 2002, covered approximately 1.56 million hectares—15,600 square kilometres—in nine Yellow River tributary watersheds. Its sites lay in Shanxi, Shaanxi, Gansu, and Inner Mongolia, rather than throughout all seven provincial-level jurisdictions included in the broader plateau region. Consequently, a map of project sites depicts selected intervention catchments, not the geographical limits of the Loess Plateau. (documents1.worldbank.org)" }, { "heading": "Loess materials and the construction of the plateau", "text": "Primary loess is sediment deposited directly from airborne dust, predominantly as silt-sized particles, with smaller proportions of clay and fine sand. “Silt” describes particle size rather than origin: river mud can be silty without being loess. Likewise, material washed from a loess-covered slope or redeposited in a channel is reworked loess, not a primary wind deposit. These distinctions matter because similar-looking sediments may record different transport processes; studies of Chinese deposits use sedimentary fabric, grain-size distributions, and other evidence to distinguish wind deposition from subsequent water transport. (dbc.library.uu.nl)\n\nThe mineral framework contains quartz and feldspars, accompanied by clay minerals and carbonate minerals, principally calcite and dolomite. Composition and porosity vary between sites and stratigraphic units rather than having a single plateau-wide value. A 2024 microstructural study of Malan loess collected two metres below an exposed surface at Yan’an, for example, measured 36.5% quartz, 22.3% feldspars, 14.8% calcite, 1.2% dolomite, and 24.8% clay minerals, with a porosity of approximately 45%. Imaging revealed silt-sized framework particles and a hierarchy of pores between grains and aggregates. Such measurements describe a particular deposit, not a standard composition for all Chinese loess. (sciencedirect.com)\n\nThe plateau’s construction involves two different geological histories: the development of its underlying bedrock, mountains, and basins, and the later accumulation of a discontinuous dust mantle. The Ordos Block provided relatively stable ground for extensive late Cenozoic deposition, but this does not make its tectonic age the age of the loess. Dust accumulation additionally required abundant sediment, exposed source surfaces, transporting winds, and suitable ground on which deposits could survive. Consequently, the age of an individual basal deposit, the expansion of widespread dust accumulation, and the development of the present landscape are not interchangeable “formation dates.” (pmc.ncbi.nlm.nih.gov)\n\nSediment production and final deposition were commonly separated by several transport stages. Physical weathering, glacial grinding, and erosion in mountain regions produced particles that rivers carried into piedmont basins, floodplains, and desert sediment stores. Wind subsequently removed exposed fines and deposited them on the plateau. Research published in 2015 identified matching provenance signatures in upper Yellow River sediment, the western Mu Us desert, and plateau loess, supporting a pathway from the northeastern Tibetan Plateau through river transport and temporary storage before wind redistribution. This mechanism makes a desert a possible sediment reservoir rather than necessarily the place where its grains originated. Nevertheless, provenance is spatially heterogeneous: a 2021 study distinguished central-western, eastern, and northeastern source zones, while eastern-plateau research published in 2023 found substantial contributions from nearby North China Craton rocks. No single source or pathway accounts uniformly for every locality and interval. (nature.com)\n\nThe chronological record extends well beyond the familiar Quaternary yellow loess. In 2002, Zhengtang Guo and colleagues reported nearly continuous windblown deposits at Qinan in Gansu spanning approximately 22–6.2 million years ago, dated using palaeomagnetic measurements and fossil evidence. This establishes early Miocene dust accumulation at that locality, not a 22-million-year age for the entire plateau. Younger red-clay sequences document extensive pre-Quaternary deposition: the Lingtai succession, for example, spans about 7–2.6 million years ago. “Red clay” here denotes a weathered sedimentary succession with evidence of windblown origin, not simply river-deposited clay. Around 2.6 million years ago, widespread accumulation shifted toward the generally coarser, faster-accumulating Quaternary loess–paleosol succession, in which dust deposits alternate with buried soil horizons. These successive records represent different stages and geographical footprints of accumulation rather than one synchronous beginning. (nature.com)" }, { "heading": "Tablelands, ridges, gullies, and ground behaviour", "text": "Three Chinese terms distinguish characteristic upland forms: yuan are broad loess tablelands with relatively level surfaces bounded by incised valleys; liang are elongated ridges separating drainage lines; and mao are rounded or dome-shaped hills. Dongzhiyuan, near Qingyang in Gansu, exemplifies a tableland whose extensive flat surface contrasts with deeply dissected margins. These terms describe landform geometry, not necessarily successive ages. Although dissection can break a tableland into ridges and isolated hills, the relief beneath the loess and the episodic accumulation of windblown sediment also influence their shape. A universal yuan–liang–mao evolutionary sequence therefore oversimplifies the landscape. (dspace.library.uu.nl)\n\nDrainage incision fragments upland surfaces through several interacting processes. Concentrated surface runoff cuts channels downward, while erosion at their heads lengthens them into previously unchannelled ground. Water plunging over a headcut excavates a scour hole and undercuts its face; the unsupported material then collapses, shifting the headcut upslope. Experiments on vegetated loess gully heads document repeated cycles of scour, overhang collapse, temporary infilling, and renewed excavation. Meanwhile, undercutting and failures along gully walls cause lateral slope retreat. The combined effects narrow interfluves, enlarge tributary networks, and progressively isolate remnants of formerly continuous surfaces. (sciengine.com)\n\nVertical joints help explain both the steepness and the instability of loess cliffs. These fractures divide the sediment into columns or blocks and provide preferential routes for infiltrating water. Weakening or erosion at their bases can release blocks along the joint surfaces. Subsurface piping adds another route of landscape development: flowing water removes particles through cracks, root channels, and animal burrows, enlarging them into tunnels. Roof failure produces surface depressions and openings that may subsequently become connected to gullies. Field observations and experiments using loess from Wanrong County, Shanxi, show that tunnel geometry, initial moisture, and sediment density affect this internal erosion. (onlinelibrary.wiley.com)\n\nWetting-induced collapse is distinct from the removal of sediment by piping. In susceptible loess, wetting weakens interparticle support and reduces matric suction, allowing an open soil structure to contract under load. Contrasting layers further complicate water movement: relatively low-permeability paleosols can impede downward infiltration and create perched wet zones above their contacts. At the Zaoling landslide in Shanxi on 15 March 2019, boreholes identified a softened zone near a loess–paleosol boundary approximately 40 metres deep. Investigators attributed the failure to accumulated deep soil water rather than an immediately preceding storm or earthquake, demonstrating that substantial failures need not coincide with an obvious short-term trigger. (doi.org)\n\nRainfall commonly triggers shallow landslides by wetting near-surface loess and reducing its shear strength; strong earthquake shaking can additionally mobilize much larger, deeper masses. This is a distinction between common mechanisms, not a strict division by trigger or depth. The 1556 Huaxian earthquake affected loess terrain around the Wei River basin as well as the basin itself. A 2018 investigation by Xu Yueren and colleagues mapped two concentrations of landslides attributed to the earthquake, containing 1,515 and 2,049 interpreted failures at opposite ends of the Huashan piedmont fault zone. Their reconstruction connected slope failures with destruction of houses and excavated cave dwellings in settlements outside the basin floor. (sciencedirect.com)\n\nThe 1920 Haiyuan earthquake produced highly mobile loess landslides, some of which crossed and dammed valleys. At Dangjiacha, field investigation reconstructed failure from a slope of about \\(20^\\circ\\) and approximately 3.2 kilometres of travel by the displaced mass. In a 2014 study, Gonghui Wang and colleagues combined field evidence with laboratory shear tests and inferred that liquefaction of water-saturated loess explained this exceptional mobility. Tests on air-dried samples did not produce a comparable loss of resistance, arguing against trapped pore air as the principal explanation for this particular landslide. Valley blockage extended the affected area beyond the original slope, exposing settlements to both moving debris and impounded water. (hub.hku.hk)" }, { "heading": "Monsoon climate, soils, and vegetation", "text": "The plateau lies near the inland limit of the East Asian monsoon, between the comparatively humid climate of eastern China and the dry continental interior. Summer circulation carries warm, moisture-bearing air inland, while winter circulation brings cold, dry northwesterly winds associated with the Siberian–Mongolian high-pressure system. Dust transport is strongly seasonal but does not coincide exactly with midwinter conditions: a 2003 study of modern dust collection identified spring and early summer as the principal dust-storm seasons. Winter–spring winds and summer rainfall therefore represent distinct, seasonally alternating influences on the plateau’s surface environment. (agupubs.onlinelibrary.wiley.com)\n\nPrecipitation generally decreases from the southeastern districts towards the northwest, producing a transition from subhumid to semiarid and locally arid conditions. An analysis using monthly observations from 100 meteorological stations within and around the plateau for 1957–2018 reported a regional mean annual precipitation of approximately 440 millimetres, with a pronounced southeast-to-northwest gradient. This moisture pattern should not be treated as a simple temperature gradient. Latitude and elevation also influence temperature: a separate analysis for 1961–2011, based on gridded observations from 299 stations, found generally cooler conditions towards the northwest but also relatively low temperatures in the elevated southwest. Higher ground can consequently be cooler without occupying the driest part of the region. (mdpi.com)\n\nRainfall is concentrated in the warm season and frequently arrives in intense storms rather than as evenly distributed, sustained precipitation. For the 1957–2018 station analysis, approximately 70–80% of annual precipitation fell between June and September. Such concentration makes annual rainfall totals an incomplete measure of growing-season water supply: dry intervals can interrupt plant growth even during the nominal rainy season, while intense rainfall can generate surface runoff instead of replenishing the root zone uniformly. The 1961–2011 analysis found a longer temperature-defined growing season across much of the plateau, but also declining precipitation on wet days over extensive areas. A longer thermally suitable season therefore did not necessarily mean a longer period with adequate soil moisture. (mdpi.com)\n\nThe loess parent material is not itself a single soil type. Moisture availability, vegetation, surface stability and cultivation influence the development of organic-rich surface horizons and the redistribution of carbonate. Infiltrating water dissolves carbonate and moves it down the profile; subsequent drying can favour its precipitation at depth. Less effective leaching leaves soils more strongly calcareous. Texture also varies regionally, with generally coarser soils in the northwest and finer soils in the southeast. A 2018 investigation along an approximately 400-kilometre transect in Shaanxi described predominantly silty-loam Calcic Cambisols, illustrating one widespread combination of loess-derived texture and carbonate-bearing soil development rather than a classification applicable to every plateau district. (pmc.ncbi.nlm.nih.gov)\n\nPotential vegetation broadly follows the moisture gradient through forest, forest-steppe, steppe and desert-steppe, but these zones are not maps of uninterrupted surviving vegetation. A field transect described warm-temperate deciduous oak vegetation around Lantian and Binxian, forest-steppe around Ningxian, and steppe around Qingyang and Mubo. In the forest-steppe belt, pine, oak and birch were particularly associated with shaded valleys and mountains; drier communities included Stipa grasses, Artemisia and Caragana shrubs. Present land cover must additionally distinguish secondary woodland, shrub communities, cultivated fields and plantations. For example, naturally regenerated oak stands in Ziwuling and planted Robinia pseudoacacia stands represent different vegetation histories, even where both are mapped as tree cover. (researchgate.net)" }, { "heading": "The Yellow River system and sediment transfer", "text": "The Yellow River links the plateau’s drainage networks through its middle reaches. After turning south around the Ordos region, the main stem follows the Shaanxi–Shanxi gorge, receiving the Kuye and Wuding systems from the west and the Fen system from Shanxi to the east. Farther south, the Wei flows east through the Guanzhong basin and joins the Yellow River near Tongguan; the Jing is a tributary of the Wei, rather than a separate main-stem tributary. These nested drainage basins connect small upland gullies to larger tributary valleys and ultimately to the trunk river. Measurements in a small headwater catchment therefore describe a different contributing area and sediment-routing system from those at a main-stem station such as Tongguan. (citeseerx.ist.psu.edu)\n\nDuring intense storms, surface runoff mobilizes loose soil, cuts rills, and concentrates along fields, roads, and gullies. Channel incision and erosion of gully walls add sediment beyond that removed from cultivated surfaces. Material may then be deposited behind check dams or within channels instead of passing immediately downstream: local soil loss is not equivalent to sediment delivered at a basin outlet. A field investigation following the July 26, 2017 storm in Wangmaogou, a small catchment within the Wuding system, illustrates this distinction. Rainfall totalled 148.8 millimetres over eight hours and ten minutes. Investigators estimated approximately 150,900 tonnes of erosion from outlet transport plus surveyed check-dam deposits, but only 35,600 tonnes passed the outlet; about 115,300 tonnes remained behind dams. The resulting sediment-delivery ratio was approximately 0.24 for that event and catchment, not a plateau-wide constant. (html.rhhz.net)\n\nDownstream effects depend on both sediment supply and the river’s capacity to transport it. Deposition in the lower Yellow River has historically raised the channel bed within embankments, reducing flood conveyance and increasing the consequences of levee failure. Reservoir trapping reduces onward sediment delivery but consumes storage capacity. At Xiaolangdi, researchers reported approximately 2.833 billion cubic metres of accumulated sediment between October 1999 and October 2010, while the lower river underwent net erosion. The water–sediment regulation scheme, initiated in 2002, uses controlled reservoir releases to scour downstream deposits and convey sediment through the system. Flood management consequently involves the timing and concentration of sediment-bearing releases as well as control of peak water discharge. (journals.plos.org)\n\nHydrometric comparisons distinguish discharge, a volume per unit time, from sediment concentration and sediment load. Discharge is established from cross-sectional velocity and area measurements, with continuous water-level records converted through calibrated stage–discharge relationships. Suspended-sediment sampling must represent the discharge-weighted concentration across the section. If \\(Q(t)\\) is discharge in cubic metres per second and \\(C(t)\\) is representative concentration in kilograms per cubic metre, suspended-sediment mass over an interval is \\(M=\\int Q(t)C(t)\\,dt\\), in kilograms. This is distinct from bedload moving along the bed. Concentration alone cannot establish annual load, and observations must adequately resolve flood pulses rather than simply multiply annual mean discharge by annual mean concentration. (usgs.gov)\n\nHistorical and more recent station records show that water and sediment have not declined proportionately. A 2018 study reported mean annual water volume at Tongguan of 42.634 billion cubic metres for 1952–1959, compared with 26.838 billion for 2010–2015; corresponding sediment loads were 1.804 billion and 166 million tonnes per year. These are period-specific observations, not estimates of present-day transport. Main-stem load also reflects upstream storage, tributary inputs, channel erosion, and deposition, rather than newly eroded plateau soil alone. Any claimed percentage contribution from the plateau therefore requires a specified period, downstream station, and sediment-accounting method; a historical source proportion cannot be applied unchanged after alterations in catchment management and reservoir operation. (html.rhhz.net)" }, { "heading": "Archaeology and the historical transformation of land use", "text": "At Shangchen, in Lantian County, Shaanxi, a stone-tool sequence reported by Zhaoyu Zhu and colleagues in 2018 extends from approximately 2.12 to 1.26 million years ago. Seventeen artefact-bearing horizons occur within alternating loess and buried soils; their ages were established through magnetic-polarity stratigraphy and correlation with the loess–paleosol sequence, rather than direct dating of the tools. The finds document repeated hominin activity on the southern plateau, not continuous occupation throughout that interval. They are evidence relevant to early human dispersal, but do not identify the toolmakers as Homo erectus or any other species: stone artefacts cannot substitute for diagnostic hominin fossils. (nature.com)\n\nThe emergence of farming was a much later, uneven transition within the wider Wei–Yellow River region. At Dadiwan, in Gansu, the early Neolithic phase conventionally dated to approximately 7,800–7,300 calibrated years before present—roughly 5850–5350 BCE—contained small quantities of domesticated broomcorn millet. Later occupation, approximately 6,500–5,900 calibrated years before present, shows substantially more intensive millet agriculture. Robert Bettinger, Loukas Barton and colleagues interpreted this sequence as a transition from hunting and gathering through low-level food production to agricultural dependence, rather than the sudden appearance of a fully developed farming economy. Dadiwan occupies a terrace of the Qingshui River in the Wei drainage: its location illustrates how settlements within the loess region could combine access to tributary valleys with surrounding uplands, rather than occupying only exposed plateau surfaces. (researchgate.net)\n\nAgricultural intensification involved changing relationships between crops, animals and soil fertility. An archaeobotanical synthesis published in 2022 identified a broad shift towards foxtail millet relative to broomcorn millet in parts of northern China, while emphasizing regional differences. Elevated nitrogen-isotope values in millet grains from the Baishui valley have been interpreted as evidence of animal-manure application from the late Yangshao period; evidence from Dadiwan likewise indicates increasingly integrated crop–livestock production. Such findings distinguish the first presence of cultivated plants from later increases in farming intensity, settlement investment and the recycling of nutrients between fields and domestic animals. (essd.copernicus.org)\n\nHistorical land-use change did not begin with a uniformly forested plateau. Pollen and documentary reconstructions indicate differing mixtures of woodland, grassland and cultivated ground, and disagree over the extent of ancient forest cover. Cultivation and pastoralism repeatedly expanded or contracted across this mosaic. Research on the Mu Us margin for 221 BCE–907 CE identifies agricultural expansion under the Qin and Han, followed by population displacement and increased grazing during political fragmentation, and renewed population and livestock pressures under the Sui and Tang. Settlement also generated demands beyond food: Tang-period Chang’an required construction timber and fuelwood from surrounding mountain districts. These histories describe successive, geographically differentiated pressures, not a single uninterrupted episode of forest clearance. (pmc.ncbi.nlm.nih.gov)\n\nLater reclamation increasingly affected sloping land. A millennium-scale reconstruction focused on Shaanxi and Shanxi associated population growth with expanding cultivation, including steep hillsides; a separate historical study of the Luochuan tableland traced changing reclamation through Qing and Republican-period records. Population pressure, government policy, warfare and variable precipitation influenced both land abandonment and renewed cultivation. Clearing vegetation for fields, collecting woody fuel and maintaining livestock imposed different pressures: loss of trees alone is therefore an incomplete measure of historical land-use intensity or susceptibility to soil erosion. (pmc.ncbi.nlm.nih.gov)\n\nReconstruction depends on comparing independent records. Excavated seeds and animal remains identify subsistence practices; documentary accounts and local chronicles record settlement, cultivation and resource extraction; pollen, charcoal and sediment sequences register vegetation, burning and landscape disturbance. None independently establishes causation: charcoal can reflect natural or human-set fires, while vegetation decline can accompany climatic drying as well as clearance. At Tianchi Lake in the Liupan Mountains, a 6,200-year pollen and charcoal record indicates that drying initiated woodland decline after the mid-Holocene, whereas human activities substantially accelerated deforestation during the last approximately 2,000 years. This local sequence separates overlapping climatic and human influences without assigning its chronology to the entire plateau. (essd.copernicus.org)" }, { "heading": "Settlement patterns and vernacular architecture", "text": "Settlement sites balance access to cultivable land, water, and routes through dissected terrain. Tableland surfaces offer broad fields and space for compact villages, but water supply can require rainwater collection and storage. Valley floors and river terraces bring settlements closer to water and transport corridors, while gully margins provide exposed loess faces suitable for excavation without occupying the centre of a cultivated tableland. These settings produce different layouts rather than a uniform “cave-village” pattern. In eastern Shaanxi, walled settlements known as buzi combine above-ground courtyard houses with defensive terrain. At South Changyi in Heyang County, gullies enclose three sides of the village; its courtyards face east, rather than following the customary southern orientation, to fit the site. (en.wikipedia.org)\n\nYaodong comprise three broad construction families: cliffside excavations, sunken-courtyard dwellings, and independently constructed vaulted houses. Cliffside examples, characteristic of the Yan’an area, extend horizontally into a hillside or cut loess face, with arched chambers opening onto an external yard. The yard supplies daylight, ventilation, circulation, and working space. Sunken forms reverse this relationship: builders excavate a square or rectangular courtyard into relatively level ground, then hollow rooms into its sides. Access is provided by a descending passage or, where adjacent relief permits, a tunnel entering from a slope. Such settlements occur in Sanyuan in Shaanxi, the Shanxian area of western Henan, and Pinglu in Shanxi. Ground between the courtyards can remain available for circulation, threshing, and drying grain. (jstage.jst.go.jp)\n\nIndependent yaodong reproduce the elongated, vaulted interior through construction rather than excavation. Brick or stone barrel vaults carry the roof, sometimes beneath an earthen covering; consequently, these buildings do not require a sufficiently high, thick loess exposure. Excavated and constructed forms can coexist within a settlement or compound. A 2020 study of Mizhi’s old city documented a local transition from cliffside dwellings towards independent brick-built forms, examining houses dating from the Qing dynasty through the Republican period. This is a particular regional development, not an evolutionary sequence applicable to every yaodong settlement. (jstage.jst.go.jp)\n\nConstruction depends on the depth and condition of the loess as well as its ease of excavation. Traditional unlined chambers retain an earthen arch and substantial material between neighbouring rooms; constructed versions instead depend on masonry vaults and supporting walls. Moisture management is integral to this architecture: rain entering a sunken courtyard must be collected or drained, while runoff above excavated rooms must not undermine their covering. Village-scale water arrangements also shape the plan. Lin Xiaodan and Pan Jiang’s 2022 account of Guanzhong tableland villages describes streets and alleys directing surface runoff into retention ponds, commonly one or two per village, which supplied people and livestock in a water-limited setting. (researchgate.net)\n\nThe surrounding earth moderates indoor temperature fluctuations through heat storage and delayed heat transfer, but thermal buffering does not guarantee adequate winter warmth, daylight, or ventilation. Nan Zhou, Masaru Nishida, and Hiroki Kitayama investigated dwellings at Zaoyuan near Yan’an in August 1999; their 2002 paper reported favourable summer thermal conditions. Later research comparing traditional yaodong, redesigned dwellings, and brick houses explicitly noted shortcomings in the traditional indoor environment. Deep rooms with openings concentrated at the front make window design and air exchange important alongside the thickness of the earth envelope. (jstage.jst.go.jp)\n\nChanging housing preferences and urbanization have brought replacement, abandonment, and adaptation rather than a single trajectory. Fang Wang and colleagues’ 2016 investigation of five Shanxian cases linked sunken-courtyard change to dwelling limitations, household structure, residents’ preferences, land-use policy, and management. New housing projects have also retained yaodong forms while changing their services and environmental design. At Zaoyuan, an 85-house pilot built in 1996–2001 incorporated enlarged glazing, sunspaces, improved ventilation, and two-storey layouts. These alterations addressed demands for brighter, more convenient accommodation without simply reproducing the spatial and infrastructural constraints of excavated dwellings. (sciencedirect.com)" }, { "heading": "Agriculture, livelihoods, and economic change", "text": "Rain-fed farming on the plateau’s uplands differs from irrigated agriculture in the major valleys. Dryland crops depend on rainfall and water stored in the soil between growing seasons, making production sensitive to drought and the timing of summer rain. Crop combinations vary locally rather than forming a single plateau-wide system. An FAO account of Mizhi County in northern Shaanxi records maize, sorghum, millet, wheat, and potatoes, alongside livestock supplying manure and draught power. A study published in 2018 of Xifeng and Zhenyuan in eastern Gansu found winter wheat grown principally for household food and spring maize as a cash crop; additional crops included potatoes and millet in both districts, with forage sorghum in Zhenyuan and apples, peaches, and vegetables in Xifeng. By contrast, irrigation systems in the Wei and Fen valleys include gravity-fed river diversions, while some upland schemes require water to be pumped through substantial vertical lifts. Valley location therefore does not imply that every field is irrigated. (fao.org)\n\nThe economic distinction between steep fields and level farmland involves labour and production risk as well as average yield. On untreated slopes, rainfall can leave fields as surface runoff, carrying soil away rather than replenishing crop-water reserves. Terraces retain more water and provide level working surfaces; where accompanied by roads, they also permit machinery and vehicles to replace some manual cultivation and carrying. A World Bank project assessment reported that terraces reduced both yield variability and labour inputs, freeing time for livestock production and non-farm work. In the Xinshuihe basin in Shanxi, the cultivated area covered by the assessment declined from 45,496 hectares in 1994 to 39,938 hectares in 2001, while terraces increased from 24% of cultivated land in 1993 to 60% in 2001. These figures describe a particular basin and intervention, not the plateau as a whole. Valley-bottom land, including sediment-filled land behind dams, offers additional level cultivation, although drainage and a rising water table can constrain its productivity. (documents1.worldbank.org)\n\nLivestock and horticulture provide alternatives to dependence on grain sales. The Mizhi account describes sheep and goats grazing pasture, uncultivated land, and harvested fields, while cattle, donkeys, horses, and mules supplied traction and manure. In watershed project areas, restrictions on free grazing were accompanied by fodder production and changes in livestock management. Orchards likewise formed an explicitly productive component of land-use change: the first World Bank-supported rehabilitation project, implemented in 1994–2002, included approximately 30,890 hectares of orchards. Thus, conversion from annual crops to perennial vegetation did not necessarily mean withdrawal from commercial agriculture. A new orchard changes both land cover and farm production, whereas earnings from work away from the farm can change a household’s livelihood without directly changing its fields. (fao.org)\n\nRural households increasingly combined farming with participation in wider labour markets. A 2010 survey of 50 farm households in Mizhi County found that 70% had migrant workers, demonstrating the importance of employment beyond the village within this small sample rather than establishing a county-wide migration rate. Migration also means that household membership, residence, and place of employment need not coincide. A separate panel study of 600 households in three Loess Plateau counties compared conditions in 1999 and 2006. It found that changes in crop, livestock, and off-farm income differed with programme participation and local economic conditions, cautioning against attributing livelihood diversification solely to cropland conversion. (researchgate.net)\n\nIncome statistics require similarly precise geographical limits. Monitoring reported for the Xinshuihe basin recorded an increase in annual per-capita net income from 671 yuan in 1993 to 1,254 yuan in 2001; net agricultural income rose from 647 to 1,062 yuan over the same period. These are the assessment’s reported monetary figures, not an inflation-adjusted income series. Its comparison in Xixian County found that project-area per-capita net income, initially similar to that outside the project, was about 200 yuan higher from 1996 onward. Such observations document local economic change during a defined intervention; they are not estimates of average rural income growth across the entire Loess Plateau. (documents1.worldbank.org)" }, { "heading": "Watershed conservation and ecological restoration", "text": "Organized soil and water conservation expanded from the 1950s, combining terracing, dam construction, tree planting, and pasture establishment. These measures initially emphasized erosion control and agricultural production; integrated management of small watersheds subsequently linked the treatment of slopes, gullies, farmland, and vegetation. Two World Bank-supported Loess Plateau Watershed Rehabilitation Projects, initiated in 1994 and 1999 and implemented over 1994–2005, developed this approach through coordinated land rehabilitation and livelihood investments. Their defined intervention areas were not equivalent to the entire geographical plateau. (doi.org)\n\nThe national Grain for Green programme, initiated in 1999, added incentives for converting selected erosion-prone cropland to forest or grassland. Participating households received compensation for land-use change, linking vegetation recovery with agricultural restructuring rather than requiring all cultivated land to be abandoned. In the watershed projects, productive terraces, orchards, and livestock facilities complemented the retirement of marginal slopes. Restrictions on open grazing protected recovering vegetation; animal sheds and cut-and-carry feeding provided an alternative means of maintaining livestock production. These arrangements addressed the dependence of conservation on continuing household livelihoods. (catalogue.unccd.int)\n\nEngineering measures operate at different points along the sediment-transfer pathway. Bench terraces replace long cultivated slopes with flatter platforms, shortening uninterrupted flow paths and retaining rainfall and soil; contour ridges and ditches intercept downslope surface runoff without necessarily reshaping the whole hillside. Check dams cross gullies, slowing sediment-laden flows and storing material already eroded upstream; accumulated deposits can eventually support farmland. Gully stabilization additionally treats the sources of channel expansion: runoff-storage structures, boundary ridges, shrub belts, and drainage works reduce direct discharge over vulnerable gully heads. Thus, preventing soil detachment on slopes and trapping sediment within channels are complementary, but different, achievements. (onlinelibrary.wiley.com)\n\nVegetation-based restoration encompasses planted trees, established shrubs, sown grasses, and natural regeneration after cultivation or grazing pressure is removed. These treatments do not produce interchangeable ecosystems. Forest establishment is constrained by moisture availability, while grasses and herbs can restore protective ground cover without creating woodland. Rainfall, slope, soil conditions, and previous cultivation consequently affect the suitability of each treatment and its combination with engineering works. FAO’s 2015 assessment identified high tree mortality and inappropriate species selection among the limitations of earlier planting, and distinguished afforestation from recovery of native grassland. Grazing exclusion can enable spontaneous regeneration rather than requiring planting at every site; the associated water-balance consequences are a separate dimension of restoration performance. (catalogue.unccd.int)\n\nReported project outcomes require explicit geographical and temporal qualification. FAO’s retrospective account of the 1994–2005 projects reported perennial vegetation cover increasing from 17% to 34%, annual income rising from approximately US$70 to US$200 per person in project households, and annual sediment flow to the Yellow River falling by more than 100 million tonnes. These are project-associated evaluation figures, not plateau-wide measurements or a controlled estimate of the projects’ effects alone. The World Bank’s 2007 account separately reported that terracing typically doubled crop yields and reduced their variability during drought. Agricultural production, household income, perennial cover, and sediment delivery therefore describe different outcomes, not a single measure of “restoration.” (catalogue.unccd.int)\n\nWhole-region assessments use different observational units and indicators. A 2020 study reported significantly increasing satellite-derived vegetation trends across 73.5% of its Loess Plateau study area during 1999–2012; this percentage describes the area exhibiting greening, not the fraction covered by vegetation. At a much smaller scale, monitoring of 14 microplots on steep gully slopes in the Zhifanggou watershed during 2018–2020 found erosion reductions of 83.9–99.6% under different plant communities relative to bare ground. Such plot-scale reductions cannot be directly equated with sediment reductions at river gauges, because intervening gullies, dams, and sediment storage modify delivery downstream. (pmc.ncbi.nlm.nih.gov)" }, { "heading": "Water–vegetation trade-offs and long-term environmental change", "text": "Vegetation recovery redistributes water rather than simply increasing its availability. Canopies intercept rainfall, some of which returns directly to the atmosphere through evaporation. Litter and roots can improve infiltration and water retention while reducing surface runoff; shading can also suppress evaporation from exposed soil. Nevertheless, greater leaf area and deeper rooting increase transpiration, potentially raising total evapotranspiration despite these savings. Improved near-surface conditions can therefore coexist with depletion several metres below ground. Research published in 2017 found that planted forests and shrubs could increase infiltration yet consume soil water faster than rainfall replenished it, especially in dense plantations. These persistent dry layers represent depletion of unsaturated soil-water reserves, not necessarily extraction from groundwater. (chinaflux.org)\n\nCatchment water balances demonstrate why modest absolute changes can substantially affect water supply. In a 2011 analysis of 57 basins, Yanhui Wang and colleagues estimated forestland evapotranspiration and runoff at 447 and 16 millimetres annually, respectively, compared with 424 and 39 millimetres for non-forestland, using a common mean precipitation of 463 millimetres. The estimated runoff difference—23 millimetres—amounted to approximately 58% of non-forestland runoff. These were regional water-balance estimates, not measurements from 57 experimentally paired afforestation sites. Such comparisons establish the importance of vegetation water consumption but depend on assumptions about storage change and differences between basins; they do not independently calculate vegetation-induced changes in rainfall. (research.fs.usda.gov)\n\nCoupled land–atmosphere studies allow precipitation to respond to altered vegetation, but their results are not uniform. A 2024 modelling study estimated that revegetation increased annual evapotranspiration by 51.0 millimetres and reduced runoff by 8.6 millimetres: the simulated precipitation increase was insufficient to offset additional water consumption. By contrast, Lei Tian and colleagues’ 2026 study found that greening-induced reductions in surface albedo could increase absorbed energy, convection, and precipitation efficiency sufficiently to sustain or increase streamflow in its Loess Plateau simulations. These findings address different model experiments and cannot be treated as interchangeable estimates. Unlike catchment comparisons or offline land models with prescribed precipitation, coupled climate models incorporate atmospheric feedbacks; their attribution nevertheless remains dependent on model structure, experimental conditions, and spatial scale. (mdpi.com)\n\nVegetation type, density, and restoration age influence whether increased biomass persists. Monitoring in Shenmu County during 2004–2014 found pronounced drying beneath Caragana korshinskii shrubs, including depleted moisture at depths of 1–4 metres after eight years of growth. Studies of black locust (Robinia pseudoacacia) plantations identified an initial drying phase and partial moisture recovery in older stands as canopy density and water consumption declined; recovery did not imply a return to cropland moisture levels. High planting density intensifies competition, while prolonged soil drought can restrict growth and cause branch death. Carbon gains are likewise depth- and site-dependent: Zhilong Lan and colleagues’ 2021 comparison at Gaoxigou, Shaanxi, found approximately 1,140–1,170 millimetres less water stored through 20-metre profiles beneath two forest types and an apple orchard than beneath farmland, without statistically significant increases in whole-profile soil organic carbon. (chinaflux.org)\n\nLong-term uncertainties concern both replenishment and disturbance. Partial recovery in shallow monitored profiles does not establish how quickly depleted reserves at greater depths recover, particularly as warming, drought, and rainfall variability alter plant demand and infiltration. The contrasting coupled-model results also leave the magnitude of future rainfall compensation unresolved. Sediment reductions cannot simply be extrapolated from average conditions: after the 25–26 July 2017 storm in northern Shaanxi, investigations of 12 dam-controlled catchments estimated erosion of approximately 10,300–49,200 tonnes per square kilometre during the event. Such observations show that exceptional storms can mobilize substantial sediment despite conservation measures. Future projections of water availability, carbon storage, or sediment delivery consequently remain conditional on rainfall intensity and seasonality, vegetation survival, and the persistence of conservation infrastructure. (chinaflux.org)" }, { "heading": "The plateau as an archive of past climate", "text": "Alternating loess and buried soils, or paleosols, preserve changes in the balance between dust accumulation and soil development. Pale, weakly weathered loess commonly represents intervals of greater dust supply and limited soil formation; darker, more strongly developed paleosols generally indicate conditions more favourable to weathering and biological activity. These are not mutually exclusive processes: dust can accumulate while a soil develops, and soil formation can modify previously deposited sediment. Consequently, a layer boundary need not mark an instantaneous climatic change. The conspicuous L9 unit at Luochuan and Xifeng, for example, includes two loess layers separated by a weaker paleosol and spans several climatic fluctuations approximately 940,000–860,000 years ago, rather than one uniformly cold episode. (pubs.geoscienceworld.org)\n\nSeveral complementary measurements underpin these paleoclimatic interpretations. Coarser grain sizes often indicate stronger dust-transporting winds, but also respond to source proximity, sediment availability, and changes in transport pathways; they are not direct anemometer readings. Magnetic susceptibility measures how readily sediment becomes magnetized. Soil formation produces ultrafine ferrimagnetic particles, including magnetite and maghemite, commonly raising susceptibility above that of less-weathered loess. This provides evidence for summer-monsoon moisture, although soil-development duration and temperature also affect the signal. Geochemical indicators, including carbonate redistribution and magnesium isotopes, reveal leaching and chemical alteration associated with soil-water movement. Agreement among these indicators is therefore more informative than interpreting any one as an unqualified measure of rainfall or wind strength. (nature.com)\n\nBiological remains provide additional, partly independent constraints. Pollen and phytolith assemblages record vegetation composition, while carbon-isotope ratios in soil organic matter and pedogenic carbonate help distinguish contributions from plants using different photosynthetic pathways. At Weinan, organic-matter isotope profiles extending back approximately 75,000 years have been used to reconstruct changes in vegetation associated with monsoon variability. Fossil land snails offer another test because species differ in their temperature and moisture requirements. In 2011, Baoyin Wu and Naiqin Wu showed that mollusc assemblages within L9 at Luochuan and Xifeng recorded alternating cold–dry and milder–humid conditions, challenging interpretations of the whole unit as an exceptionally severe, uninterrupted glaciation. (sciencedirect.com)\n\nChronology combines methods with different strengths. Magnetostratigraphy matches sedimentary records of geomagnetic polarity with the established reversal timescale. Friedrich Heller and Liu Tung-sheng’s 1982 investigation of a borehole near Luochuan identified polarity zones extending below the Olduvai subchron and assigned the oldest measured loess an age of approximately 2.4 million years under the chronology then used. Luminescence dating instead estimates the time since mineral grains were last sufficiently exposed to light, using accumulated radiation dose and environmental dose rate. It provides independent age constraints, particularly for younger deposits, but signal saturation limits its range: conventional quartz optically stimulated luminescence commonly underestimates ages beyond roughly 50,000–60,000 years in central-plateau loess, while suitable feldspar methods extend further. (nature.com)\n\nCorrelation with marine oxygen-isotope records and astronomical tuning supplies additional age control between independently dated horizons. A 1999 Luochuan chronology, for example, tuned magnetic-susceptibility variations to calculated insolation changes. Such procedures depend on assumptions about which climatic processes a proxy records and how quickly it responds. A tuned correspondence cannot subsequently serve as wholly independent proof of the orbital relationship used to construct it. Uncertainty also arises from selecting correlation points and interpolating through intervals with changing accumulation rates; closely spaced samples do not necessarily provide equally precise ages. (sciencedirect.com)\n\nThese sequences connect Asian monsoon history with Northern Hemisphere ice-sheet development. A 2023 study led by Hong Ao combined new Luochuan and Chaona measurements with existing records covering 1.6 million years and climate-model experiments. Pronounced grain-size increases around 1.25 million and 900,000 years ago were interpreted as intensified aridity and winter-monsoon transport associated with ice-sheet expansion during the mid-Pleistocene transition, when dominant glacial periodicity shifted from approximately 41,000 toward 100,000 years. The Luochuan component comprised 982 samples through 98.3 metres of sediment. That sampling density describes the spacing of observations, not uniform temporal resolution: the duration represented by each sample varies with accumulation, while soil formation can integrate environmental conditions over longer intervals. (eprints.gla.ac.uk)" } ], "links": [ "china", "yellow-river", "ecological-restoration", "drainage-basin", "tibetan-plateau", "world-bank", "porosity", "erosion", "floodplain", "surface-runoff", "permeability", "temperature", "forest", "river-discharge", "human-migration", "isotope", "deforestation", "causation", "architecture", "urbanization", "irrigation", "maize", "household", "labor-market", "evaporation", "transpiration", "evapotranspiration", "groundwater", "climate-model", "paleoclimatology" ], "sources": [ { "title": "The Loess Plateau Watershed Rehabilitation Project", "url": "https://documents1.worldbank.org/curated/en/142661468762366534/pdf/307770CHA0Loess0Plateau01see0also0307591.pdf" }, { "title": "Boundary Data of Loess Plateau Region", "url": "https://www.geodoi.ac.cn/DOIPaper/ENHTML/2017139401/201701139417.htm" }, { "title": "Work Report by the Working Group on Sustainable Agriculture", "url": "https://english.mee.gov.cn/Events/Special_Topics/AGM_1/1999agm/meetingdoc99/201605/t20160524_345055.shtml" }, { "title": "25-year, quarterly land change maps of China's Loess Plateau reveal long-term and substantial water-induced soil erosion mitigation", "url": "https://essd.copernicus.org/articles/18/3303/2026/essd-18-3303-2026.html" }, { "title": "Magnetic fabric from Red clay sediments in the Chinese Loess Plateau", "url": "https://www.nature.com/articles/srep09706" }, { "title": "Large-number detrital zircon U-Pb ages reveal global cooling caused the formation of the Chinese Loess Plateau during Late Miocene", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC9555786/" }, { "title": "Loess Plateau storage of Northeastern Tibetan Plateau-derived Yellow River sediment", "url": "https://www.nature.com/articles/ncomms9511" }, { "title": "Onset of Asian desertification by 22 Myr ago inferred from loess deposits in China", "url": "https://www.nature.com/articles/416159a" }, { "title": "Modelling soil erosion in a small catchment on the Loess Plateau", "url": "https://dspace.library.uu.nl/bitstream/handle/1874/553/full.pdf?isAllowed=y&sequence=1" }, { "title": "Strategies for Gully Stabilization and Highland Protection in Chinese Loess Plateau", "url": "https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2022.812609/full" }, { "title": "Morphology evolution of vegetation-covered loess gully heads induced by hydraulic and gravitational erosion", "url": "https://www.sciengine.com/ISWCR/doi/10.1016/j.iswcr.2025.01.003" }, { "title": "Mechanisms of wetting-induced loess slope failures", "url": "https://doi.org/10.1007/s10346-019-01144-4" }, { "title": "The loess landslide on 15 march 2019 in Shanxi Province, China", "url": "https://link.springer.com/article/10.1007/s10346-019-01342-0" }, { "title": "Spatiotemporal Trends and Variation of Precipitation over China’s Loess Plateau across 1957–2018", "url": "https://www.mdpi.com/2073-4433/14/2/323" }, { "title": "Relationship between multi-scale climate factors and performance of ecological engineering on the Loess Plateau, China", "url": "https://link.springer.com/article/10.1007/s11676-021-01343-7" }, { "title": "Calcium Nodules as a Proxy for Quaternary Paleoclimate Change on China’s Loess Plateau", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC4669161/" }, { "title": "Regional-scale variation and distribution patterns of soil saturated hydraulic conductivities in surface and subsurface layers in the loessial soils of China", "url": "https://www.fceia.unr.edu.ar/seminarios-dh-curiham/Seminario_34_mario_castiglioni/wang2013.pdf" }, { "title": "The Biogeographical Distribution of Soil Bacterial Communities in the Loess Plateau as Revealed by High-Throughput Sequencing", "url": "https://www.frontiersin.org/journals/microbiology/articles/10.3389/fmicb.2018.02456/full" }, { "title": "Responses of soil fungal communities and functional guilds to ~160 years of natural revegetation in the Loess Plateau of China", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC9479326/" }, { "title": "黄土高原暴雨产沙路径及防控—基于无定河流域2017-07-26暴雨认识", "url": "https://html.rhhz.net/sswcc/html/2018/4/2018-4-66.htm" }, { "title": "Temporal Variation of Streamflow, Sediment Load and Their Relationship in the Yellow River Basin, China", "url": "https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0091048" }, { "title": "How Streamflow is Measured", "url": "https://www.usgs.gov/water-science-school/science/how-streamflow-measured?qt-science_center_objects=0" }, { "title": "Suspended-sediment transport measurement", "url": "https://www.usgs.gov/publications/suspended-sediment-transport-measurement" }, { "title": "Computation of Fluvial-Sediment Discharge", "url": "https://pubs.usgs.gov/twri/twri3-c3/pdf/TWRI_3-C3.pdf" }, { "title": "Hominin occupation of the Chinese Loess Plateau since about 2.1 million years ago", "url": "https://www.nature.com/articles/s41586-018-0299-4" }, { "title": "Stone tools put early hominids in China 2.1 million years ago", "url": "https://www.sciencenews.org/article/shangchen-stone-tools-put-early-hominids-china-earlier" }, { "title": "Holocene spatiotemporal millet agricultural patterns in northern China: a dataset of archaeobotanical macroremains", "url": "https://essd.copernicus.org/articles/14/4777/2022/essd-14-4777-2022.html" }, { "title": "Evolution and effects of the social-ecological system over a millennium in China’s Loess Plateau", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC7541069/" }, { "title": "Evolution process of landuse and its impact on soil erosion in Luochuan Tableland from the Qing Dynasty to the Republic of China", "url": "https://www.dlyj.ac.cn/EN/abstract/article/1000-0585/8477" }, { "title": "Loess Plateau", "url": "https://en.wikipedia.org/wiki/Loess_Plateau" }, { "title": "Research on the Development Process of Yaodong in Mizhi Ancient City, China", "url": "https://www.jstage.jst.go.jp/article/aiue/2020/1/2020_aiue2020013/_article/-char/ja/" }, { "title": "Yaodong", "url": "https://en.wikipedia.org/wiki/Yaodong" }, { "title": "Doctoral dissertation on Yaodong courtyard layouts", "url": "https://kitakyu.repo.nii.ac.jp/record/2000256/files/%28KKDE1271SJ%29%E7%94%B2%E7%AC%AC271%E5%8F%B7_%E3%82%A4%E3%82%B7%20%E3%82%AB%E3%82%A4.pdf" }, { "title": "Study on the Thermal Environment of the YaoDong Dwelling in the Loess Plateau of China", "url": "https://www.jstage.jst.go.jp/article/jaabe/1/1/1_1_81/_article" }, { "title": "The New Generation of Yaodong Cave Dwellings, Loess Plateau", "url": "https://world-habitat.org/awards/winners/the-new-generation-of-yaodong-cave-dwellings-loess-plateau/" }, { "title": "The New Generation of Yaodong", "url": "https://world-habitat.org/?acf_field=long_copy&generate_pdf=1&nonce=04bf4817dc&post_id=13471" }, { "title": "301183-219", "url": "https://www.fao.org/4/ar432e/ar432e.pdf" }, { "title": "Adoption of Conservation Tillage on the Semi-Arid Loess Plateau of Northwest China", "url": "https://www.mdpi.com/2071-1050/10/8/2621" }, { "title": "World Bank Document: Yellow River Basin Water Resources", "url": "https://documents1.worldbank.org/curated/en/761171468018630241/pdf/multi-page.pdf" }, { "title": "IDA at WORK: Agriculture", "url": "https://documents1.worldbank.org/curated/en/464451468324545355/pdf/519370WP0IDA1A1Box345544B001PUBLIC1.pdf" }, { "title": "An empirical analysis of the effects of China's Land Conversion Program on farmers' income growth and labor transfer", "url": "https://pubmed.ncbi.nlm.nih.gov/19777292/" }, { "title": "China and the World Bank: A Partnership for Innovation", "url": "https://documents1.worldbank.org/curated/en/823911468241152523/pdf/427960Second0R1x0327366B0001PUBLIC1.pdf" }, { "title": "Global guidelines for the restoration of degraded forests and landscapes in drylands", "url": "https://catalogue.unccd.int/590_a-i5036e.pdf" }, { "title": "Long-Term Gully Erosion and Its Response to Human Intervention in the Tableland Region of the Chinese Loess Plateau", "url": "https://www.mdpi.com/2072-4292/13/24/5053" }, { "title": "Quantifying the Effects of Vegetation Restorations on the Soil Erosion Export and Nutrient Loss on the Loess Plateau", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC7728691/" }, { "title": "Soil moisture decline due to afforestation across the Loess Plateau, China", "url": "https://www.chinaflux.org/upload/accessory/20198/2019871457236913.pdf" }, { "title": "Annual runoff and evapotranspiration of forestlands and nonforestlands in selected basins of the Loess Plateau of China", "url": "https://research.fs.usda.gov/download/treesearch/38552.pdf" }, { "title": "Vegetation Restoration Increases the Drought Risk on the Loess Plateau", "url": "https://www.mdpi.com/2223-7747/13/19/2735/pdf?version=1727668722" }, { "title": "Energy-mediated feedbacks of vegetation greening enhance precipitation efficiency and sustain water yield in global semi-arid regions", "url": "https://www.nature.com/articles/s44221-026-00631-y" }, { "title": "Long-term vegetation restoration increases deep soil carbon storage in the Northern Loess Plateau", "url": "https://www.nature.com/articles/s41598-021-93157-0" }, { "title": "Magnitude of soil erosion in small catchments with different land use patterns under an extreme rainstorm event over the Northern Loess Plateau, China", "url": "https://doi.org/10.1016/j.catena.2020.104780" }, { "title": "Terrestrial mollusc records from Xifeng and Luochuan L9 loess strata and their implications for paleoclimatic evolution in the Chinese Loess Plateau during marine Oxygen Isotope Stages 24-22", "url": "https://cp.copernicus.org/articles/7/349/2011/cp-7-349-2011.pdf" }, { "title": "Northern hemisphere ice sheet expansion intensified Asian aridification and the winter monsoon across the mid-Pleistocene transition", "url": "https://www.nature.com/articles/s43247-023-00686-9" }, { "title": "Magnetostratigraphical dating of loess deposits in China", "url": "https://www.nature.com/articles/300431a0" }, { "title": "A comprehensive dataset of luminescence chronologies and environmental proxy indices of loess-paleosol deposits across Asia", "url": "https://www.nature.com/articles/s41612-023-00555-4" }, { "title": "800-kyr land temperature variations modulated by vegetation changes on Chinese Loess Plateau", "url": "https://www.nature.com/articles/s41467-019-09978-1.pdf" }, { "title": "Northern hemisphere ice sheet expansion intensified Asian aridification and the winter monsoon across the mid-Pleistocene transition", "url": "https://eprints.gla.ac.uk/292178/" } ], "license": "https://creativecommons.org/licenses/by/4.0/" }