Impacts of DEM resolution and area threshold value uncertainty on the drainage network derived using SWAT

Abstract

Many hydrological algorithms have been developed to automatically extract drainage networks from DEM, and the D8 algorithm is widely used worldwide to delineate drainage networks and catchments. The simulation accuracy of the SWAT model depends on characteristics of the watershed, and previous studies of DEM resolution and its impacts on drainage network extraction have not generally considered the effects of resolution and threshold value on uncertainty. In order to assess the influence of different DEM resolutions and drainage threshold values on drainage network extraction using the SWAT model, 10 basic watershed regions in China were chosen as case studies to analyse the relationship between extracted watershed parameters and the threshold value. SRTM DEM data at 3 different resolutions were used in this study, and regression analysis for DEM resolution, threshold value and extraction effects was done. The results show that DEM resolution influences the selected flow accumulation threshold value; the suitable flow accumulation threshold value increases as the DEM resolution increases, and shows greater variability for basins with lower drainage densities. The link between drainage area threshold value and stream network extraction results was also examined, and showed a variation trend of power function y = ax b between the sub-basin counts and threshold value, i.e., the maximum reach length increases while the threshold value increases, and the minimum reach length shows no relation with the threshold value. The stream network extraction resulting from a 250 m DEM resolution and a 50 000 ha threshold value was similar to the real stream network. The drainage network density and the threshold value also shows a trend of power function y = ax b ; the value of b is usually 0.5.

Effects of Input Data Content on the Uncertainty of Simulating Water Resources

“…The use of different DEM sources affects the watershed delineation, resulting in different terrain characteristics and physical structures of the stream network (Table 1), which is expected to influence model performance [32,40,49]. Before discretizing the watershed into eight sub-basins, we used the same upstream drainage threshold for the area required to form the beginning of a stream (4.74 km 2 ), because Wu et al [50] reported that different thresholds may affect the streamflow. Additionally, the watershed was classified into three land-slope classes: <7%, 7-15% and >15% that are considered moderate, medium and steep slopes, respectively, according to German guidance for soil surveying and mapping [51].…”

A rainwater control optimization design approach for airports based on a self-organizing feature map neural network model

“…According to different algorithm principles, the algorithms for extracting the flow data and the accumulation of the flow are divided into two types: a single flow algorithm and multi-flow algorithm [20]. Considering the size of the data and the quality requirements of the design model for the data analysis results, the maximum slope single flow algorithm [21][22] is used to extract the flow direction and the aggregate cumulative data in the innocent DEM.…”

Classifying Dams for Environmental Flow Implementation in China

“…The most popular hydrological models in China are the Xin'anjiang model, the Soil and Water Assessment Tool (SWAT), and VIC. As previously mentioned, VIC-derived flow data were applied in this study [32][33][34]. The input data for VIC were national rainfall, temperature, and wind data of 0.25 × 0.25, and the output flow data series was of 1952-2012.…”