Abstract. Soil erosion is one of the most significant environmental problems in China.
From 2010 to 2012, the fourth national census for soil erosion sampled
32 364 PSUs (Primary Sampling Units, small watersheds) with the areas of
0.2–3 km2. Land use and soil erosion controlling factors including rainfall
erosivity, soil erodibility, slope length, slope steepness, biological
practice, engineering practice, and tillage practice for the PSUs were
surveyed, and the soil loss rate for each land use in the PSUs was estimated
using an empirical model, the Chinese Soil Loss Equation (CSLE). Though the
information collected from the sample units can be aggregated to estimate
soil erosion conditions on a large scale; the problem of estimating soil
erosion condition on a regional scale has not been addressed well. The aim of
this study is to introduce a new model-based regional soil erosion assessment
method combining a sample survey and geostatistics. We compared seven spatial
interpolation models based on the bivariate penalized spline over triangulation (BPST)
method to generate a regional soil erosion assessment from the PSUs.
Shaanxi Province (3116 PSUs) in China was selected for the comparison and
assessment as it is one of the areas with the most serious erosion problem.
Ten-fold cross-validation based on the PSU data showed the model assisted by
the land use, rainfall erosivity factor (R), soil erodibility factor (K),
slope steepness factor (S), and slope length factor (L) derived from a 1 : 10 000
topography map is the best one, with the model efficiency coefficient (ME)
being 0.75 and the MSE being 55.8 % of that for the model assisted by the
land use alone. Among four erosion factors as the covariates, the S factor
contributed the most information, followed by K and L factors, and R factor
made almost no contribution to the spatial estimation of soil loss. The LS factor
derived from 30 or 90 m Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM) data worsened the estimation when used as the covariates for the interpolation of soil loss. Due to the
unavailability of a 1 : 10 000 topography map for the entire area in this study,
the model assisted by the land use, R, and K factors, with a resolution of
250 m,
was used to generate the regional assessment of the soil erosion for Shaanxi
Province. It demonstrated that 54.3 % of total land in Shaanxi Province
had annual soil loss equal to or greater than 5 t ha−1 yr−1. High
(20–40 t ha−1 yr−1), severe (40–80 t ha−1 yr−1), and
extreme (> 80 t ha−1 yr−1) erosion occupied 14.0 %
of the total land. The dry land and irrigated land, forest, shrubland, and
grassland in Shaanxi Province had mean soil loss rates of 21.77, 3.51, 10.00,
and 7.27 t ha−1 yr−1, respectively. Annual soil loss was about
207.3 Mt in Shaanxi Province, with 68.9 % of soil loss originating from the
farmlands and grasslands in Yan'an and Yulin districts in the northern Loess
Plateau region and Ankang and Hanzhong districts in the southern Qingba
mountainous region. This methodology provides a more accurate regional soil
erosion assessment and can help policymakers to take effective measures to
mediate soil erosion risks.
Assessment of Soil Erosion by RUSLE Model using Remote Sensing and GIS of Indla-Ghatkhed, District-Amravati, Maharashtra
