Early season irrigation detection and evapotranspiration modeling of winter vegetables based on Planet satellite using water and energy balance algorithm in lower Colorado basin

Authors: Dhungel, Ramesh; Anderson, Raymond - Ray; French, Andrew; Skaggs, Todd; SABER, MAZIN - University Of Arizona; SANCHEZ, CHARLES - University Of Arizona; SCUDIERO, ELIA - University Of California, Riverside

Submitted to: Irrigation Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/20/2023
Publication Date: 6/30/2023
Citation: Dhungel, R., Anderson, R.G., French, A.N., Skaggs, T.H., Saber, M., Sanchez, C., Scudiero, E. 2023. Early season irrigation detection and evapotranspiration modeling of winter vegetables based on Planet satellite using water and energy balance algorithm in lower Colorado basin. Irrigation Science. 42(1):15-27. https://doi.org/10.1007/s00271-023-00874-7.
DOI: https://doi.org/10.1007/s00271-023-00874-7

Interpretive Summary: Satellite-based remote sensing models have long been used to estimate crop water use and schedule irrigation. However, monitoring irrigation using previous generations of satellites was challenging as many of them had either low temporal resolution (frequency) or spatial resolution (pixel size) or both. In this study, we used a recently launched cubesat satellite constellation (Planet) with high spatial and temporal resolution (daily 3 meter spatial resolution) data to estimate crop water use and evapotranspiration (ET) using the water and energy balance model BAITSSS. Water balance models including BAITSSS need irrigation dates to initiate the model and for conducting water balance thereafter during the crop season. These irrigation events can be inferred by evaluating at surface reflectance of individual spectral bands as well as the combination of these bands such as the normalized differential vegetation index (NDVI) and moisture index. Our results showed that daily Planet data can detect the onset of irrigation, including the critical first irrigation event, and some successive irrigation events thereafter during the partial cover period. This irrigation detection improves model performance without the need for farmer or user supplied irrigation data. The results also indicate that Planet data can help to improve ET but also can be beneficial to overall crop water management through detection of irrigation onset. These results are of interest for irrigation managers and farmers who need a robust and easy to use crop water use model for irrigation scheduling and water deliveries.

Technical Abstract: Water shortages in the Western United States will continue to be one of the foremost American agricultural challenges in the coming years. As agriculture is the largest consumer of water in the western US, improvements in irrigation scheduling and modeling are needed to maximize production under limited water. Various satellite-based remote sensing models have been developed to estimate crop water use. However, water balance-based evapotranspiration (ET) models need field-scale irrigation information for initiating the seasonal soil water balance. This initialization has been challenging due to the lack of remotely sensed irrigation event data. In this study, we utilized a recently launched satellite constellation (Planet) with high temporal and spatial resolution data (daily,'~'3 m) to evaluate if Planet data can facilitate early season irrigation detection. We utilized normalized difference vegetation index (NDVI), moisture index, and individual spectral bands to detect moisture and ultimately infer irrigation. As part of this comparison, a hybrid two-source energy and water balance model BAITSSS (Backward-Averaged Iterative Two-Source Surface temperature and energy balance Solution) was used to estimate ET with Planet-based vegetation indices and irrigation information. We also compared the results to eddy covariance (EC) located at lettuce fields in Yuma, Arizona in the lower Colorado River basin between 2016 and 2020. Overall, the results indicated that Planet’s data helped to establish the field-scale onset of irrigation, which assisted to initiate soil water balance in the BAITSSS model, thus ultimately improving ET. Further, these results should support the development of near-real-time landscape-scale ET and should be highly beneficial to agricultural communities for sub-field-scale effective water management.