Improving groundwater recharge estimation in the Kano Plains Aquifer, Western Kenya through global sensitivity analysis and WetSpass-M model validation using groundwater level fluctuations

The Kano Plains aquifer system in western Kenya, to the east of the Winum Gulf of the Lake Victoria. WetSpass-M model was used to estimate recharge for the period between October 2016 to December 2017 during a period of groundwater level monitoring campaigns in the area. To account for model uncertainty, four global sensitivity analysis methods including Sobol, Fourier Amplitude Sensitivity Test, Random Balance Designs Fourier Amplitude Sensitivity Test, and Delta-Moment-Independent Measure were used. The model was validated using the water table fluctuation method. Model sensitivities ranging from 0 to 0.7 were calculated. Except Sobol, the sensitivity results are comparable. The most sensitive parameters are average rainfall intensity in the wet indicating that recharge is rainfall-driven when moisture is abundant; the alfa coefficient in the dry season possibly because it is associated with land cover indices, number of rainy days. Sobol analysis produces well-distributed sensitivity indices suggesting its strength to capture interactions and variances across multiple parameters. Recharge estimates from WetSpass-M aligned well with those calculated using the water table fluctuation, at a specific yield of 0.005(-). The calculated maximum and minimum monthly recharge values are 120 mm and 15 mm in April and August 2017 respectively. As would be expected, enhanced recharge is received in rainy seasons and regions with sandy soil. • Recharge for the Kano Plains aquifer has been estimated using the WetSpass-M1.2.1 model. • Global sensitivity analysis was used to assess model uncertainty. • Average intensity is the main factor in wet, and alfa coefficient in dry seasons. • Recharge estimates from WetSpass-M align with water table fluctuation method results. • Soil texture and weather are the main factors influencing recharge in Kano Plains. Effect of land cover is negligible.