Impacts of DEM resolution, source, and resampling technique on SWAT-simulated streamflow. (September 2015)

• Record Type: Journal Article
• Authors: Tan, Mou Leong; Ficklin, Darren L.; Dixon, Barnali; Ibrahim, Ab Latif; Yusop, Zulkifli; Chaplot, Vincent
• Is Part Of: Applied geography. Volume 63(2015:Sep.)
• Page Start: 357
• Page End: 368
• Publication Date: 2015-09
• Subjects: DEM -- SWAT model -- ASTER -- SRTM -- Hydrology -- Sensitivity analysis; Geography -- Periodicals; Human geography -- Periodicals; Human ecology -- Periodicals; 910
• DOI: 10.1016/j.apgeog.2015.07.014
• Languages: English
• ISSNs: 0143-6228

Abstract

Abstract: The sensitivity of streamflow simulated with the Soil and Water Assessment Tool (SWAT) model to Digital Elevation Model (DEM) resolution, DEM source and DEM resampling technique is still poorly understood. The objective of this study is to compare SWAT model streamflow estimates in the Johor River Basin (JRB), Malaysia for DEMs differing in resolution (from 20 to 1500 m), sources (Shuttle Radar Topography Mission: SRTM v4.1, Advanced Space-borne Thermal Emission and Reflection Radiometer: ASTER GDEM2, EarthEnv-DEM90 and Global Multi-resolution Terrain Elevation Data 2010: GMTED2010) and resampling technique (nearest neighbour, bilinear interpolation, cubic convolution and majority). The key findings were as follows: (1) SRTM v4.1 (Root Mean Square Error (RMSE) = 11.16 m) and EarthEnv-DEM90 (RMSE = 12.4 m) had better vertical accuracy over the JRB compared to the ASTER GDEM2 (RMSE = 16.95 m); (2) Accurate annual streamflow simulations were obtained by using nearly all of the DEM resolutions, as pointed out by a relative error (RE) lower than 7% from 20 to 50 m and from 100 to 800 m DEMs; (3) Prediction errors were the lowest for ASTER GDEM2 (RE = 3.9%), followed by SRTM v4.1 (RE = 5.4%), EarthEnv-DEM90 (RE = 6.3%), and GMTED2010 (RE = 7.3%); (4) the majority and nearest neighbour resampling techniques performed the best (RE of 6.0%), followed by bilinear interpolation (RE of 7.2%) and cubic convolution (7.5%). The study indicates that DEM resolution is the most sensitive SWAT model DEM parameter compared to DEM source and DEM resampling technique for streamflow simulation within SWAT. Graphical abstract: Highlights: An approach to assess the sensitivity of SWAT streamflow to DEM input is developed. Uncertainties in DEM resolution, source and resampling method are investigated. SRTM v4.1 has better vertical accuracy than the ASTER GDEM in the JRB. The statement of "lower resolution is less accurate" is not necessarily correct. The majority resampling technique performed the best in SWAT streamflow simulation.