Uncertainty, calibration and validation of the Mississippi Irrigation Scheduling Tool model

Author(s): Prabhu, R.; Lee, N.; Wadsworth, M.; Sassenrath, G.; Schmidt, A.; Crumpton, J.; Rice, B.; van Riessen, H.; Thornton, R.

Implementation and use of a model requires an estimate of its accuracy. The Mississippi Irrigation Scheduling Tool (MIST) is an on-farm decision support tool to assist farmers in irrigating. The accuracy of the model is critical in designing good water management protocols. This research presents the results of the uncertainty analysis of the MIST model, showing the margin of error (uncertainty) of the irrigation advice. The basis for the verification and validation of the model is also given. The MIST calculates the daily soil water balance in a crop field from daily weather measurements, irrigation, and rainfall, accounting for crop type, planting date, soil type, tillage, and other field-specific information. The model output informs farmers of when irrigation is needed. The uncertainty analysis determines the margin of error in the irrigation decision and gives a range within which irrigation is feasible. The current uncertainty analysis also gives essential information on the influence of input parameters on the final irrigation recommendation calculated by the water balance.

The uncertainty calculations were based on Taylor's Series Method for the calculation of the total systematic uncertainty arising from measurement error of variables in the water balance calculation. The errors in measurement were one standard deviation in range, equivalent to an uncertainty with a confidence level of 68.2%. Because the current day's soil water balance depends on the previous day's water balance, the computations are iterative. As equations cascade to calculate the daily water balance, the uncertainties also propagate through the equations. Initially, uncertainty quantifications were performed for two sets of water balance calculations using local weather data. The final uncertainties for the water balance were of the order 3-6%, which is within the acceptable range for error.

The MIST water balance calculations were validated using local weather data consisting of rain days, and significant changes in the solar radiation, relative humidity and wind speed. The final water balance results showed values within acceptable ranges and comparable to in situ measurements of soil moisture. The final relative uncertainty in the water balance value was around 9%, which is in the normal range of margin of error. The current MIST web-based application and uncertainty quantification have been verified and validated for current parameters. The accuracy of the model was shown to be suitable for use by farmers in the Mississippi Delta area, and will help improve water management in crop production systems.

Go back


Past Conference Archive