Spatial and temporal maize soil water extraction (Depletion) dynamics: Part I. Development and evaluation of a soil water extraction model

A stepwise soil water extraction (depletion) model was developed and evaluated against actual evapotranspiration (ET_a) as determined by soil water balance for center-pivot irrigated maize (Zea mays L.) at the University of Nebraska-Lincoln South Central Agricultural Laboratory (SCAL) near Clay Center, Nebraska. The model performance was evaluated using three different soil physical properties data sets: (1) variable soil data collected for each root zone depth every 0.30 m to 1.50 m (site-specific data) from 39 sites in the field, (2) field-Averaged root zone depth properties every 0.30 m to 1.50 m (field-Averaged data), and (3) NRCS Web Soil Survey data. All analyses were conducted for three irrigation levels [fully irrigated treatment (FIT), 75% FIT, and rainfed] and five nitrogen levels (0, 84, 140, 196, and 252 kg N ha^-1). On average, the extraction model underestimated water use early in the growing season, when soil water content was at or near field capacity. For instance, between the V6 and V7 growth stages (41 to 49 days after planting), soil water extraction using the site-specific, field-Averaged, and NRCS data deviated from soil water balance-determined ET_a by -5 mm, -8 mm, and -23 mm, respectively. However, on average, minimal differences existed during the reproductive growth stages (0.4, 0.4, and -7 mm week ^-1 for the site-specific, field-Averaged, and NRCS data analyses, respectively), which corresponded to the critical irrigation period in the research area. Total seasonal water extraction versus seasonal ET_a had root mean square differences (RMSD) and coefficients of determination (R²) of 14.9 mm and 0.85, 16.1 mm and 0.83, and 88.7 mm and 0.25 for the site-specific, field-Averaged, and NRCS analyses, respectively. The performance of the model varied due to the scale of soil properties available, with the site-specific soil properties performing best due to the model's ability to more effectively differentiate soil water extraction amounts between treatments, since each individual soil layer's field capacity is accounted for in the extraction calculation. For the site-specific data, the daily R ² and RMSD values for all data points within an irrigation regime were 0.97 and 0.4 mm, 0.96 and 0.4 mm, and 0.96 and 0.5 mm for rainfed, 75% FIT, and FIT, respectively. Within each irrigation regime, the 0 kg N ha ^-1 treatment had the highest R² and lowest RMSD values compared to the other N treatments. Daily RMSD values for 0 kg N ha^-1 were 0.2, 0.3, and 0.2 mm for rain-fed, 75% FIT, and FIT, respectively. Closer agreements (lower RMSD values) between the extraction comparisons were achieved for rainfed conditions than for the irrigated settings, most likely as a result of smaller error in the partitioning of water throughout the soil profile, as the soil water distribution or redistribution had more stable conditions for rainfed conditions than for the irrigated settings. This research provides invaluable data and information for understanding some of the fundamentals of soil water extraction for application in variable-rate irrigation and nitrogen management.