Toward high resolution and daily thermal infrared measurements for agricultural water management

Sustainable water use in agriculture, while ensuring high yield returns, is key to tackling challenges imposed by climate change and population growth. Knowing the crop water status within the field allows for optimized water consumption by matching management practices to the actual crop water demand. Science and applications communities have made clear the needs and requirements for daily, field-scale (< 100 m) evapotranspiration (ET) data for agricultural applications. Current and planned space missions with thermal infrared (TIR) measurements either have high-spatial or high- temporal resolution, but not both, making it hard to capture the field-scale variability required for irrigation and crop growth modeling.

Hydrosat has innovated through the technical barriers to achieving field-scale, global TIR and VNIR measurements for ET every day, multiple times per day. With an upcoming launch en route to a 16+ smallsat constellation, Hydrosat data will be a game-changer and will significantly advance our ability to monitor and manage agricultural systems. An Early Adopter daily 20-m surface temperature product is already available now and can be used to accurately track crop water supply and demand within a specific field.

Here we show the potential of a new method that combines the spatio-temporal advantage of Hydrosat Early Adopter (along with freely available satellite data) with the predictive ability of crop model simulations to overcome the limitations of existing methods of irrigation management at the field and sub-field levels. The method was validated over multiple corn fields in the US Corn Belt, exploring a wide range of environmental conditions and management practices and across multiple growing seasons (2019-2021). ET, soil moisture, and yield data collected during the season were used for validation.

First, high spatio-temporal resolution thermal and multispectral satellite data were used to derive ET and leaf area index (LAI) during the crop growing season. Using these products, phenological development and soil-water components of the APSIM crop model were calibrated to accurately determine (and improve upon) farm-level predictions, both in terms of soil moisture content and end-of-season yield. Our method successfully estimated soil moisture with high accuracy (RMSE of 1.43 mm/mm and rRMSE of 7.47 %) and predicted yield reliably up to 14 weeks before harvest, with a strong correlation to independently collected measurements (RMSE of 1162 kg/ha and rRMSE of 7 %). The proposed approach has the potential for driving irrigation management decisions while quantifying end-of-season yield without the need for in situ data.