Geospatial analysis of specific degradation in South Korea

Date
2019
Authors
Kang, Woochul, author
Julien, Pierre Y., advisor
Grigg, Neil S., committee member
Morrison, Ryan, committee member
Kampf, Stephanie, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
South Korea experienced many local and concentrated sediment problems such as landslides, upland erosion, rills and valleys, aggradation/degradation, and flood plain sediment deposition. These problems vary in space and time, therefore a reliable and consistent approach to model sediment processes is desirable. In contrast to sediment yield at the basin scale, Specific Degradation (SD) is defined as the ratio of the sediment yield divided by the watershed area. Field measurements of discharge and sediment concentration are analyzed at 70 stations in South Korea. Half of the sampled river basins (35 stations) represent streams in mountain regions and the other half represent rivers. The Modified Einstein Procedure (MEP) was used to determine the total sediment load at all stations. The Flow Duration – Sediment Rating Curve (FD-SRC) method was used to determine the sediment yield and specific degradation for all gauging stations. The annual sediment yield of 70 rivers and streams in South Korea ranged from 10 to 1,000 tons/km2▪yr. The application of three existing models from the literature showed Root Mean Square Errors (RMSE) in excess of 1,400 tons/km2▪yr and gave negative values of the Nash-Sutcliffe Efficiency coefficient (NSE) for existing models, which indicates that the observed mean is a better predictor than the model. The main characteristics of each watershed were analyzed using GIS tools such as ArcGIS version 10.3.1. The data used for the analysis included: (1) daily precipitation data at 60 stations from the Korea Meteorological Administration (KMA); (2) a detailed soil map from the National Institute of Agriculture Sciences; (3) a 5m by 5m resolution Digital Elevation Model (DEM); and (4) land cover raster data at a 10 m resolution from the Ministry of Environment (ME). Seven regression models based on these watershed characteristics are proposed to estimate the mean annual sediment yield and specific degradation. In decreasing order of importance, the meaningful parameters are: (1) drainage area; (2) mean annual precipitation; (3) percentage of urbanized area; (4) percentage of sand of the surface soil (upper 50cm); (5) percentage of wetland and water; and (6) morphometric parameters such as watershed average slope and two parameters of the hypsometric curve. The RMSE for the newly developed models decreased to 90 tons/km2▪yr and the NSE increased from -50 to 0.5, which shows good agreement between the model and the measured sediment yield on these watersheds. The calculated specific degradation and mean annual soil loss of mountain streams were larger than alluvial rivers. Erosion loss mapping at 5m, 30m and 90m was also developed from the Revised Universal Soil Loss Equation (RUSLE). Satellite images and aerial photos were used to better represent geospatial features affecting erosion and sedimentation. Long-term reservoir sedimentation measurements were available to determine the Sediment Delivery Ratio (SDR). An important finding from this analysis is that the percentage of the area covered with wetland and water is well-correlated with the estimated sediment delivery ratios. It suggests that the transfer of sediment to the rivers is affected by wetlands located near alluvial rivers. The erosion maps at 5m resolution could clearly show unique erosion features (i.e. hill slopes, croplands, and construction sites) and locate areas for sediment deposition (i.e. wetlands and agricultural reservoirs). In comparison, the gross erosion rates at 90 m resolution were highly distorted and could not delineate the areas with high upland erosion rates. Sustainable sediment management with these methodologies could be helpful to solve various erosion and sedimentation problems.
Description
2019 Summer.
Includes bibliographical references.
Rights Access
Subject
multiple regression model
specific degradation
sediment in South Korea
geospatial analysis
Citation
Associated Publications