Modelling water utilization patterns in apple orchards with varying canopy sizes and different growth stages in semi-arid environments

Accurate estimates of orchard evapotranspiration (ET) and its components are important for precise irrigation scheduling, irrigation system designs, and optimal on-farm water allocation particularly in water-limited environments. Direct measurements of ET remain costly, laborious and sometimes difficult to apply over heterogeneous surfaces such as crop fields. Therefore, accurate crop water-use models are required for on-farm precise water resources management. In this study, we adopted and improved the Priestley-Taylor Jet Propulsion Laboratory (PT-JPL) model developed by Fisher et al 2008 to estimate crop water use across different apple plants. Specifically, the model was developed to quantify the partitioning of apple orchard water use into beneficial (tree transpiration) and non-beneficial water use (orchard floor evaporation) as influenced by tree canopy cover. Data were collected in twelve orchards spread across key apple producing regions in the Western Cape Province of South Africa over three growing seasons (2014/15, 2015/16, 2016/17). Model ET estimates were tested against ET data measured; using the eddy covariance method and transpiration measured based on sap flow monitoring techniques. The results showed that the original Fisher PT-JPL model performed poorly in ET estimation across all the orchards under study. The model yielded lower R-2, ranging from 0.02 to 0.64 and Nash-Sutcliffe Efficiency (NSE) from -10.93 to 0.20. Thus, we subsequently improved the model by incorporating soil moisture and vapour pressure deficit stress factors and by introducing a variable Priestley and Taylor coefficient (a). The modified PT-JPL model demonstrated an improvement in ET estimates. The root mean square of error (RMSE) of the estimated daily ET varied from +/- 0.60 mm/d to +/- 1.99 mm/d whereas the mean absolute error (MAE) varied from +/- 0.49 mm/d to +/- 1.91 mm/d, while the R-2 varied from 0.54 to 0.75 in orchards with varying canopy cover. The findings of this work underscore the utility of the modified PT-JPL model for estimating ET and its components in apple orchards from planting until the trees reach full-bearing age.