AI Model for Estimating Crop Water Demand: An Artificial Intelligence (AI) Model to Improve Agricultural Water Use Efficiency Using Field, Plant, and Weather Data
Lead Researcher: Pasha Fayzul
Industry Partners: WiseConn, PowWow, AgH2O, Dynamax,
Jain Irrigation, and Irrometer

The need: Relying on evapotranspiration to determine the amount of irrigation to apply is a common practice but can result in over or under irrigating. Identifying which other parameters are most important for contribute to estimating crop water requirement will be useful in efforts to better predict crop water requirements in real time more accurately, information that can be used to improve irrigation timing and amounts.  

 The goal:  Multiple numerical techniques were applied to analyze and estimate crop water requirement, providing the basis to develop an artificial neural network (ANN) model. 

 The impact:  Out of seventeen (17) variables and their combinations, ET (evapotranspiration), SR (solar radiation), AT (air temperature), ST (soil temperature), and RH (relative humidity) were most important in terms of predicting the crop water requirement correctly. The ANN model developed can successfully capture the general pattern and dynamics of the crop water requirement. Compared to a baseline condition, the average water and energy savings potential anticipated over a growing season using this model is about 15%.