Modelling and managing critical zone relationships between soil, water and ecosystem processes across the Loess Plateau

Lead Research Organisation: University of Leeds
Department Name: Sch of Geography


The Loess Plateau of China covers an area 2.5x the size of the UK (some 640,000 square km) in the upper and middle reaches of China's Yellow River and is renowned for having the most severe soil erosion in the world; deforestation, over-grazing and poor agricultural practice have resulted in degenerated ecosystems, desertification and unproductive agriculture in the region. To control severe soil erosion on the Loess Plateau, the Chinese government imposed a series of policies for fragile ecosystems, such as the 1999 state-funded "Grain-for-Green" project, which has resulted in significant land use changes. Related programmes have produced beneficial effects on soil erosion and water cycles. However, the impact of these changes in soil and water processes on related ecosystem services is unknown and demands further study. The proposed research will focus on three spatial scales: slope, watershed, and region. It uses a combination of A) experiments to collect environmental, biological and agronomic data; B) remote sensing data and C) modelling approaches.

A) Data collection: Four experimental stations located in four main topographical regions of the Plateau are chosen as case studies: 1) Ansai Comprehensive Experimental Station of Soil and Water Conservation; 2) Changwu Agro-ecology Experiment Station; 3) Guyuan Ecological Station; 4) Shenmu Erosion and Environment Station. At each station, treatments of different vegetative covers, slopes, and the practices of soil and water conservation at the plot scale were set up in the 1980s and data collections include: soil water, canopy size, runoff, soil losses and meteorological records. Most of the Chinese members of this project have been involved in prior studies at the stations.
At the slope scale, additional environmental, biological and agronomic data will be monitored in a sub-set of the plots. At watershed scale, four watersheds where the stations are located will be monitored. The spatial distribution of the following variables will be measured: precipitation, soil properties, vegetative types, canopy size, runoff and soil loss.

B) Remote sensing data collection: At the regional scale, remote sensing combined with ground-truthing data will be used to investigate the spatial variability of vegetation type, land cover, productivity, the components of water balance, soil losses, soil type, etc.

C) Modelling approaches: a cascade approach will be used to build an improved model framework applied to different spatial scales. Mechanistic soil-water-plant models will be applied to the slope scale. Their outputs will then be used as inputs for models at watershed level. Spatial empirical/statistical models will be used at the regional level. Observed and collected data from A) and B) will be used to further develop, calibrate and validate our models.
Model simulations at the slope level will be used to reveal the dynamic mechanisms in soil and water in different regions and analyse the effects of vegetation type, soil type, slope degree, climatic factors and management practice. Watershed models will estimate soil and water carrying capacity for different vegetation types, predict the effect of land use/cover changes on soil losses, water cycle and ecosystem services and evaluate management scenarios in the practices of soil and water conservation, vegetative changes, and ecosystem services. The soil and water carrying capacity for different vegetation types and the optimal ecosystem services will be addressed at the regional scale. Outreach workshops and demonstrations will disseminate knowledge to farmers and policy makers.

The proposed research will elucidate the coupled relationships between soil and water processes and agro-ecosystem services at various scales, and evaluate the effects of vegetation cover and changes in land use on water cycle, soil erosion, and ecosystem services across the Loess Plateau.

Planned Impact

The primary impact will be the coupled benefits of environmental sustainability and the promotion of economic development and social welfare in the Loess region. This will be achieved through better understanding the relationships between the soil and water processes and agro-ecosystem services, including grain production, net primary production, carbon sequestration, water retention and soil erosion control. This research will provide guidance on practices and managements for sustainable ecosystem services in the fragile region, which will secure food production and save water resources in the arid Loess Plateau. Farmers in the area will receive guidance, enabling them to make informed decisions to adopt optimal agricultural and water management practises, thus increasing income by maintaining or increasing agricultural production. Likewise, environmental agencies and other organisations will have the tools to enable more accurate estimation of water balance, enabling stable management of ecosystems services.
Title Research methods for the use of Geographically weighted models for environmental research 
Description GWR in R workshop Lex Comber1 and Paul Harris2 1 School of Geography, University of Leeds, Leeds, LS2 9JT, UK Email: 2 Rothamsted Research, North Wyke, EX20 2SB, UK Email: This workshop will describe how to apply Geographically Weighted Regression (GWR) as described in Brunsdon et al (1996) and Fotheringham et al (2002) to spatial data, using R and RStudio, the free open source statistical software. It will use the sub-catchment soils data for the Liudaogou watershed, as described in Wang et al (2009). The workshop aims are to get you using R / RStudio and this document describes the preparations you need to make before coming to the workshop. For the workshop you will need: 1) A computer with R and RStudio installed: R can be downloaded from here: and then you should download RStudio from here RStudio has R as its engine and has a nicer interface than R 2) To have gone through and run the code in the Owen guide (up to page 28) 3) You will need to have an internet connection - this is really important! 4) If you can, please bring your own data to explore using GWR. This should be a spatial dataset of some kind: a CSV or EXCEL file with latitude and longitude or a shapefile from ArcGIS are probably the most common formats. We can provide data if you do not have any. References Brunsdon, C.F., Fotheringham, A.S. and Charlton M. (1996). Geographically Weighted Regression - A Method for Exploring Spatial Non-Stationarity, Geographic Analysis, 28: 281-298. Fotheringham, A. S., C. Brunsdon, and M. Charlton. (2002). Geographically Weighted Regression: The Analysis of Spatially Varying Relationships. Chichester: Wiley Wang, Y., Zhang, X. and Huang, C., 2009. Spatial variability of soil total nitrogen and soil total phosphorus under different land uses in a small watershed on the Loess Plateau, China. Geoderma, 150(1), pp.141-149. 
Type Of Material Improvements to research infrastructure 
Year Produced 2017 
Provided To Others? Yes  
Impact We trained ~50 workshop attendees, all post grads in the use of advanced geocomputational methods, making all the materials available online to them and others