Proximate Sensing and Computer Modeling to Enhance Food and Water Security by Improving Agricultural Water Management

Abstract

The global population continues to grow in both number and impact per capita on the environment, exerting more pressure on already degraded lands. Moreover, climate emergency impacts on the environment demand immediate actions to preserve our civilization. Agriculture accounts for 69% of total human water withdrawal, ranking first by consumptive water use among major economic sectors. Thus, any improvement to agricultural water management can significantly save water contributing to enhance and secure food production. The consolidation and the emergence of new technologies as computer modeling and remote sensing (RS) allow us to analyze and simulate different strategies, expediting improvements to agricultural water management. In this research, HYDRUS-2D computer model was used to analyze and develop improved water management practices for leaching salts on saline agricultural soils. HYDRUS-2D simulations represented the actual conditions properly, producing similar salinity patterns and flow lines to those of registered on the lab experiments. Those results were further validated by the good agreement when comparing the breakthrough curves, which is a more quantitative way to compare results. HYDRUS-2D models were developed to simulate and test an alternative method to leach salts. Compared to traditional leaching, both improved strategies reduce water consumption, off-site pollution, and save time. Also, aiming to improve agricultural irrigation management, a novel RS approach was proposed, and pilot tested. Using RS data, from UAS and Satellites, a new trapezoidal space was developed from the near-infrared transformed reflectance (NTR) and normalized difference vegetation index (NDVI) data to estimate and map of soil moisture (SM). The estimated SM was evaluated against in situ observations from TDR sensors and a traditional RS approach based on thermal data and NDVI. SM from the novel approach was further compared to SM estimates from the traditional surface temperature-NDVI approach. Also, HYDRUS-2D numerical simulations were conducted to evaluate SM distribution throughout the soil profile. Results indicate that the NTR-NDVI estimation accuracy varies with soil depth, with r between 0.18 and 0.93 and RMSE between 0.02 and 0.18 cm3 cm-3. The best estimations using UAS- and Satellite-based NTR-NDVI approach were obtained for the near surface soil layer. The NTR-NDVI SM estimates were comparable to LST-NDVI SM. Root-zone SM and fractional PEW maps’ spatial variability depended on the resolution of the source data used. The generated high spatial resolution root-zone SM and PEW maps enable crop producers to increase water productivity by applying water more precisely and reducing environmental impacts.