Remote-Sensing based Global Evapotranspiration estimates at high spatial resolution through Sentinel-2 satellite imagery and meteorological data. The CWRweb tool.

This work introduces a new webGIS tool to estimate the Crop Water Requirement (CWR), using time series of satellite images and meteorological data, at high spatial resolution and a global scale. This CWRweb tool provides users with information on the temporal evolution of the CWR, as a first approach of the crop evapotranspiration, as well as other parameters of interest. This process is implemented via web and requires no proficiency in remote sensing.

The implemented calculation of the evapotranspiration under standard conditions (ETc) stands on the robust FAO-56 methodology, based on the relationship between the Crop Coefficient and the Reference Evapotranspiration (Kc-ETo). The CWRweb tool adopts the single crop coefficient approach, combining the effects of both, crop transpiration and soil evaporation into a single coefficient (Kc). These Kc values derive from the NDVI time series of Sentinel-2 multispectral satellite images, for a broad range of crops (horticulture, woody crops, and other crops) and natural vegetation, assuming a general component for the soil evaporation.

The CWRweb tool benefits from the potential of the Sentinel-2A & B satellite constellation to provide users with free time series of images with a spatial resolution of 10m × 10m and a revisit frequency of 2-3 days. The high frequency of Sentinel-2 imagery allows to obtain daily Kc values through interpolation of NDVI data from cloud-free images at high spatial resolution. Online access to massive databases of satellite images, such as those of the Copernicus Data Space Ecosystem program (https://dataspace.copernicus.eu/), together with recent advances on meteorological numerical models to provide global ETo layers at different gridding size, are boosting the operational use of the CWRweb tool.

The CWRweb tool runs and graphically displays daily ETc, as well as NDVI, Kc, and ETo values used in its calculation, for a selected time interval. Results can be provided at both, field and pixel scales. An assessment of the CWRtool was conducted by comparison against the OpenET tool on a selection of crops-sites in California, USA. An average uncertainty of RMSE=0.9 mm·d^-1, with a negligible bias, was obtained in a performance analysis using the OpenET ensemble outputs as a reference, using 15 different locations, and data for the period 2016-2024. These results are promising and reinforce the potential of the CWRweb tool for the operational estimation of global evapotranspiration at a high spatial resolution.