An integrated SVR and crop model to estimate the impacts of irrigation on daily groundwater levels. (January 2018)

• Record Type: Journal Article
• Authors: Guzmán, Sandra M.; Paz, Joel O.; Tagert, Mary Love M.; Mercer, Andrew E.; Pote, Jonathan W.
• Is Part Of: Agricultural systems. Volume 159(2018)
• Page Start: 248
• Page End: 259
• Publication Date: 2018-01
• Subjects: DSSAT -- CROPGRO -- Support vector machine (SVM) -- Irrigation -- Soybean -- Groundwater; Agricultural systems -- Periodicals; Agriculture -- Environmental aspects -- Periodicals; 338.16
• DOI: 10.1016/j.agsy.2017.01.017
• Languages: English
• ISSNs: 0308-521X

Abstract

Abstract: As groundwater resources are used more intensively, the need to define appropriate strategies to plan and manage irrigation systems under diverse climatic conditions becomes increasingly important. To promote more efficient irrigation practices, accurate and optimal information regarding the interaction between crop water use and groundwater sustainability is needed. In this study, we outlined a modeling approach that combines the features of a crop growth model and a support vector regression (SVR) model for the comprehensive assessment of groundwater variability under different soybean ( Glycine max [L.] Merr) irrigation thresholds throughout the growing season. The 20%, 40%, 50% and 60% thresholds of available water were calibrated using the CROPGRO-Soybean model to simulate daily irrigation requirements of soybeans grown in the Mississippi Delta Region (MDR). The daily crop water requirements along with precipitation and previous daily groundwater levels were used as inputs in the SVR to evaluate the predicted response of daily groundwater levels to different irrigation demands. We examined the performance of the SVR model based on the Mean Squared Error (MSE) and its ability to capture the seasonal variability in groundwater levels under different scenarios. Results demonstrate that higher groundwater irrigation volumes significantly affect the daily availability of groundwater. However, more volume does not represent significantly higher soybean yields. We conclude that the hybrid crop-SVR model is able to assess the subsurface water response to multiple scenarios of groundwater available for irrigation and provide useful information for the decision making. Highlights: We linked the CROPGRO-Soybean and SVR models for irrigation management. This study evaluates the impacts in groundwater from crop production. CROPGRO predicted soybean yields and irrigation requirements in Mississippi. SVR provides baseline information for groundwater irrigation management. A lower irrigation threshold has a short-time impact in the groundwater level variability.