SWAT Model Application to Estimate Runoff for Ungauged Arid Catchments Experiencing Rapid Urbanisation: Riyadh case study

Abstract

The built-up area of Riyadh city increased from approximately 4.5 km² in 1950 to reach approximately 1,600 km² by 2022 spreading over vast areas of the Wadi Hanifah and Wadi As Silayy catchments. The rapid growth of the city has led to repeated urban flooding. There is an urgent need to study surface runoff and how it is affected by land-use/land-cover (LULC) change in the ungauged catchments of the city. This study addressed that knowledge gap and was the first attempt to calibrate, validate, and run a semi-distributed model to simulate runoff depths and discharge rates for Riyadh's main catchments and sub-basins using five historical and five future scenarios. The Soil and Water Assessment Tool (SWAT) was used for the modelling. TerraClimate evapotranspiration (ET) data was used to calibrate the SWAT model owing to a dearth of observed runoff data across Riyadh city. The literature review revealed that the use of Terraclimate ET to calibrate SWAT models is still very limited so far. The only previous study found is Herman et al. (2020). Therefore, this study is fairly unique in that it uses Terraclimate ET to successfully calibrate and validate a SWAT model. A one-by-one sensitivity analysis was performed to evaluate the impact of changing parameter values on the runoff simulations. The results indicated that simulated runoff sensitivity to selected parameter values in the calibrated SWAT models was minimal in the study area, where the relationships between simulated annual runoff and max and min runoff resulted in a very strong R2 (0.9998). The calibrated and validated SWAT models were run monthly and daily to simulate runoff and to assess the impact of several LULC change scenarios on surface runoff for both historical and future periods. The results of SWAT models of the main catchments and sub-basins located within the built-up areas demonstrated the positive effect of Riyadh’s development on runoff and discharge values for historical LULC scenarios and LULC 2030 probabilities scenarios. But the increasing rates of simulated runoff were not the same for all sub-basins due to the different proportions of urbanisation in each sub-basin. On the contrary, simulation results showed that runoff depths and discharge rates in sub-basins outside the boundaries of the built-up areas of Riyadh did not have significant changes when using historical LULC scenarios or LULC 2030 probabilities scenarios. The increase in runoff depths and discharge rates in the sub-basins reflected the direct influence of the urbanisation process on surface runoff. The increase in simulated surface runoff and discharge can be attributed mainly to the potential decrease of relatively permeable barren lands and the increase of impervious urban surfaces. Limitations faced during the SWAT model development suggest further research should aim to get detailed and accurate runoff estimates in Riyadh city to sufficiently assist decision-makers and city officials to adopt runoff and flood hazard management schemes in the city.