3. Results and discussion3.1. Canopy coverThere was a good correlation between the observed and predicted CC for the calibrated data (0.90 R2 0.97). This meansthat the AquaCrop model accounted for 90 to 97% of the variability in annual CC (Fig. 3). The RMSE 7.3% andNRMSE 11.8% were low and considered good for the calibration of CC. The 0.90 EF 1.00 showed that the model simulated CC optimally. The d-index (0.981.00) clearly showed thatthe AquaCrop model simulated CC very well under soil fertility management practices (Table 4). Similarly the validateddata in 2016 had a high R2 (0.940.98) and an average lowerRMSE (2.064.60%) and NRMSE 16.7% compared with thecalibrated data. The EF 0.93 and d-index 0.99 were veryhigh for crop modelling. Generally, with b 1.00, the AquaCrop model did not overestimate CC under soil fertlity management except in V1N75, V1N100 and V2N50 for thecalibratded data and V1N25–V2N75 for the validated data. Theslight overestimations could be attributed to the response ofthe model to simulate canopy expansion under added soil fertility. Adeboye et al. [46] reported that the AquaCrop modelsimulated CC with R2 0.95, NRMSE 14.3%, d-index 0.97,and EF 0.84. The AquaCrop model simulated CC withR2 > 0.57 under soil fertility management [61]. Under rainfedconditions, AquaCrop simulated CC with R2 = 0.83 andRMSE = 10.5% [58]. The assessment of the model showed thatCC of soybeans under soil fertility management were wellsimulated.3.2. Soil water storageThe goodness of fit R2 was 0.75 for the calibrated data. Thisimplies that most of the variances in the SWS were explained