Dividends in flow prediction improvement using high-resolution soil database Academic Article uri icon

abstract

  • © 2019 Study region: Upper Blue Nile, Ethiopia Study focus: The recent availability of high-resolution soil data offers a better representation of spatially varying hydrologic parameters and could help in building accurate models. Despite the release of the AfSIS 250 m resolution soil data in 2015, its benefits in improving streamflow prediction accuracy is yet to be evaluated. The focus of this study is to evaluate the improvement in prediction accuracy of an uncalibrated hydrologic model built using the recently released AfSIS 250 m soil database in comparison to four other soil databases which are under use for the last decade. New hydrological insights for the region: Limited streamflow data availability has been a major impediment to assessing water resources potential of watersheds. Recently the Africa Soil Information Service (AfSIS) provided 250 m resolution datasets with soil properties for up to six soil depth layers. The objective of this study is to evaluate the flow prediction accuracy gains from the AfSIS 250 m if any, in comparison to previously availed databases using un-calibrated SWAT (Soil and Water Assessment Tool) hydrologic models. The performance of the un-calibrated models was evaluated using streamflow at two of the gauging stations. Evapotranspiration output from the respective models was also compared to remote sensing derived evapotranspiration estimates (MOD16 ET and ALEXI). Only marginal improvements in streamflow simulation were achieved by using the most detailed AfSIS soil. A satisfactory agreement was found in most of the area between SWAT simulated ET and MOD16 ET during the wet seasons whereas ALEXI ET products are seasonally more consistent and comparable to SWAT estimates over wide land cover types. Robust evaluation of these datasets in different landscapes and geographic regions is warranted before a specific application.

altmetric score

  • 1

author list (cited authors)

  • Ayana, E. K., Dile, Y. T., Narasimhan, B., & Srinivasan, R.

citation count

  • 4

publication date

  • February 2019