Abstract
Simulating soil erosion variation with a temporal land use database reveals long-term fluctuations in landscape patterns, as well as priority needs for soil erosion conservation. The application of a multi-year land use database in support of a Soil Water Assessment Tool (SWAT) led to an accurate assessment, from 1977 to 2006, of erosion in the upper watershed of the Yellow River. At same time, the impacts of land use and landscape service features on soil erosion load were assessed. A series of supervised land use classifications of Landsat images characterized variations in land use and landscape patterns over three decades. The SWAT database was constructed with soil properties, climate and elevation data. Using water flow and sand density data as parameters, regional soil erosion load was simulated. A numerical statistical model was used to relate soil erosion to land use and landscape. The results indicated that decadal decrease of grassland areas did not pose a significant threat to soil erosion, while the continual increase of bare land, water area and farmland increased soil erosion. Regional landscape variation also had a strong relationship with erosion. Patch level landscape analyses demonstrated that larger water area led to more soil erosion. The patch correlation indicated that contagious grassland patches reduced soil erosion yield. The increased grassland patches led to more patch edges, in turn increasing the sediment transportation from the patch edges. The findings increase understanding of the temporal variation in soil erosion processes, which is the basis for preventing local pollution.
Original language | English |
---|---|
Pages (from-to) | 1358-1366 |
Number of pages | 9 |
Journal | Science of the Total Environment |
Volume | 408 |
Issue number | 6 |
DOIs | |
Publication status | Published - 15 Feb 2010 |
Externally published | Yes |
Keywords
- soil erosion modeling
- landscape pattern
- SWAT
- Yellow River
- tempo-spatial dynamics