Validation and testing of the FAO AquaCrop model under different levels of nitrogen fertilizer on rainfed maize in Nigeria, West Africa

- We evaluate AquaCrop ability to simulate the yield of rainfed maize in Nigeria.
- The results showed that the model simulates the final grain yield reasonably well.
- The difference between observed and simulated yields ranges between +19% to −30%.
- The evaluation results for the grain yield show a NRMSE of around 8%-17%.
- Overall, the agreement between simulated and observed yields is considered satisfactorily.

The objectives of this study were to evaluate AquaCrop's ability to simulate the cumulative grain yield of rainfed maize for different soil fertility levels in the northern Guinea Savanna zone of Nigeria. Seven years (2007-2013) of field experimental data on maize grown under rainfed condition at the Institute for Agricultural Research were used to calibrate (2007 data set) and validate (2008-2013 data set) AquaCrop. We assessed the agreement between model simulated and actual maize yields using correlation coefficient, R2, and the index of agreement, d, as well as the NRMSE. R2 values ranged from 0.82 to 0.99 while values of d ranged from 0.6 to 0.88, indicating a moderate poor agreement to very good agreement. The NRMSE varied between 8% (indicating “excellent” agreement) and 17% (good agreement). On the other hand, in percentage terms the differences between actual and simulated yield range from +19% to −30%. Of the 19 treatments evaluated, 13(68%) are within 10% of each other, generally considered very good, three (16%) are within 20%, considered acceptable; and 3 (16%) are >20%, considered poor. Furthermore, simulated yields systematically over-estimate observed yields a not uncommon result that suggests the need for additional calibration. The grain and biomass yields evaluation results were consistent with other validation studies of the model.