Grid order prediction of ephemeral gully head cut position: Regional scale application

Abstract

Abstract Predicting ephemeral gully (EG) location is essential for many land management decisions and for accurate application of widely used soil erosion models. Topographic indexes are widely used for identifying EG head locations, even though the calibration of the topographic index models limits their use for broad scales. The objective of this research was to test the accuracy and usability of grid order in predicting EG head locations at the regional scale compared to other topographic indexes commonly used for this purpose. Three hundred fifty-three EGs in eight watersheds located within different landforms (Major Land Resource Areas) in Iowa, USA, were visually digitized, georeferenced, and analyzed using 3-meter LiDAR-derived elevation datasets and orthoimages. The location prediction accuracy, spatial location prediction error stability, and the variation of grid order critical threshold (CT) values were compared to predictions from four commonly used topographic indexes, specific contributing drainage area (As), Compound Topographic Index (CTI), Stream Power Index (SPI), and a variation of Stream Power Index (AS2). Results indicate that using As, CTI, SPI, and AS2 to identify flow length upstream from predicted EG head locations would require careful calibration for each watershed or substantial errors in prediction accuracy would occur. In contrast, the accuracy of using grid order was acceptable at the regional scale even without model calibration with Nash Sutcliff Efficiency estimates ranging from 0.29 to 0.59; the distance between surveyed and predicted EG head locations was within 20\xa0m for all studied EGs. The grid order value of 4 was the best CT value when the slope gradient was less than 10%, and 3 for steeper slopes. These results could be easily applied for different landforms across wider areas. The result of this research supports the prediction of EGs using automated systems without calibration at a broad scale where quality DEMs are available.