Effect of Watershed Delineation and Climate Datasets Density on Runoff Predictions for the Upper Mississippi River Basin Using SWAT within HAWQS

Abstract

The quality of input data and the process of watershed delineation can affect the accuracy of runoff predictions in watershed modeling. The Upper Mississippi River Basin was selected to evaluate the effects of subbasin and/or hydrologic response unit (HRU) delineations and the density of climate dataset on the simulated streamflow and water balance components using the Hydrologic and Water Quality System (HAWQS) platform. Five scenarios were examined with the same parameter set, including 8- and 12-digit hydrologic unit codes, two levels of HRU thresholds and two climate data densities. Results showed that statistic evaluations of monthly streamflow from 1983 to 2005 were satisfactory at some gauge sites but were relatively worse at others when shifting from 8-digit to 12-digit subbasins, revealing that the hydrologic response to delineation schemes can vary across a large basin. Average channel slope and drainage density increased significantly from 8-digit to 12-digit subbasins. This resulted in higher lateral flow and groundwater flow estimates, especially for the lateral flow. Moreover, a finer HRU delineation tends to generate more runoff because it captures a refined level of watershed spatial variability. The analysis of climate datasets revealed that denser climate data produced higher predicted runoff, especially for summer months.

Quantifying the effects of watershed subdivision scale and spatial density of weather inputs on hydrological simulations in a Norwegian Arctic watershed

“…The reason is because the capacities to capture the important features of the watersheds and variation of rainfall regime of the coarse watershed delineation are better than those of the fine one. This argument is somewhat against the argument in the study in the Upper Mississippi River Basin, USA (Chen et al 2021).…”

“…In this case, the trends of changes were not consistent for every water balance component, particularly in ET (Chaplot 2014). Another study in the 4,91,700-km 2 Upper Mississippi River Basin, USA, found that the simulation results of surface runoff, lateral flow, groundwater flow, and water yield (WYLD) increased approximately 0.98, 92, 2.73, and 2.07%, respectively, when the number of sub-watersheds and HRUs increased (Chen et al 2021).…”

“…In contrast, the accuracy of peak flow prediction was concluded to be decreased (approximately 20%) when numbers of subwatersheds increased in a study in the 384-km 2 Grote Nete River catchment, in Flanders, Belgium (Rouhani et al 2009). Recently, a study in the 491,700-km 2 Upper Mississippi River Basin, USA, found that the fine schemes of watershed delineation (12-digit sub-basin scenario) yielded higher values of streamflow simulation (approximately 1.79-7.17%) compared to the coarser schemes (8-digit sub-basin scenario), since the fine schemes are able to capture a sophisticated level of the spatial variation of watershed features including variation of rainfall regime (Chen et al 2021). This finding was agreed with the finding of a previous study in the 152.29-km 2 Little Washita watershed, near Chickasha, Oklahoma, USA, since they pointed out that the number of sub-watersheds strongly impacted the simulated streamflow hydrograph (Norris & Haan 1993).…”

“…Many previous studies around the world applied the SWAT model to investigate the impacts of watershed subdivision on the results of hydrological simulations, including runoff (Norris & Haan 1993;Bingner et al 1997;Jha et al 2004;Arabi et al 2006;Rouhani et al 2009;Chaplot 2014;Munoth & Goyal 2019a), water balance components (Tripathi et al 2006;Chaplot 2014;Chen et al 2021), and streamflow (Mamillapalli et al 1996;FitzHugh & Mackay 2000;Haverkamp et al 2002;Jha et al 2004;Muleta et al 2007;Rouhani et al 2009;Aouissi et al 2013Aouissi et al , 2018Chiang & Yuan 2015;Ozdemir et al 2017;Pignotti et al 2017;Chen et al 2021). Regarding the studies on runoff, there were inconsistent among previous findings.…”

“…The reason is because runoff from the upper sub-watersheds could reach to the outlet of the watershed only after runoff from the lower sub-watersheds has been already discharged (Rouhani et al 2009). Furthermore, increasing the number of sub-watersheds leads to increasing channel slope and drainage density that result in higher simulation results of some water balance components (Chen et al 2021). Also, the change in drainage density also influences the accuracy of runoff prediction (Goodrich 1992).…”