A New Topographic Conditioning Method to Integrate Surface Depressions in Physically Based Hydrologic Modeling
Abstract
Digital elevation models (DEMs) are critically important inputs to physically based distributed hydrological modeling. Traditionally, small-scale topographic depressions in DEMs are treated as artifacts because of DEM error and its limited ability to represent real depressions due to the low accuracy and resolution. Hence, they are often removed by depression filling, but this can result in substantial modifications to the DEM and affect the modeled flow paths. With the increasing availability of high accuracy and resolution DEMs, the assumption that all topographic depressions are artifacts no longer holds. To properly account for the representation of surface depressions in DEMs, we devised an alternative approach that (1) retains topographic features likely to be true depressions while removing errors and artificial sinks caused by road embankments and bridges and (2) incorporates an adapted regression-based river smoothing algorithm to minimize DEM modification. The proposed approach was evaluated against the conventional depression filling method and a rainfall-recession test was performed on an impervious surface using a distributed hydrologic model in Goodwin Creek Experimental Watershed, Mississippi, U.S. where farm ponds are a common feature. Compared to the conventional method, the proposed method has a significantly lower impact on the DEM, with nearly half the number of modified cells and a four-fold reduction in elevation changes. Furthermore, the DEM processed using the proposed method was able to identify 86% of the farm ponds from aerial imagery and to more successfully simulate pooling. Additionally, we investigated the impact of DEM conditioning on surface water storage and its distribution between river and non-river storage. The proposed method showed a higher river storage and lower non-river storage than the conventional method. The former is due to the preservation of the original channel bottom elevation by the proposed method and the latter can be attributed to the increase in flat surfaces by the conventional method. Lastly, the streamflow response for the proposed method was slower with a lower peak. Our method provides an automated and reproducible framework and can further enhance physical realism in hydrologic modeling studies sensitive to surface depressions.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.H13A..08J