An improved NDVI-based method to predict actual evapotranspiration of irrigated grasses and crops

Recent investigations have shown the potential of a water balance method based on Normalized Difference Vegetation Index (NDVI) data to estimate actual evapotranspiration (ETa) of Mediterranean grasslands and croplands. The method uses the Fractional Vegetation Cover (FVC) derived from NDVI to separately estimate vegetation transpiration and bare soil evaporation, which are limited by short-term meteorological water stress scalars. The current paper proposes a modification of these scalars to address the occurrence of water supply additional to rainfall, i.e. irrigated condition. The modified method utilises the temporal evolution of FVC since the beginning of the water stress period to infer the occurrence of irrigation and consequently modify the water scalars. The new method is tested in two study areas in Tuscany (Central Italy), where meadows and annual crops are grown in both rainfed and irrigated conditions. The high spatial fragmentation of these areas is addressed by the use of imagery taken by the recently launched Sentinel-2 MultiSpectral Instrument (MSI), which allows the identification of all main seasonal NDVI evolutions. The tests firstly demonstrate the null or marginal impact of the modification in rainfed condition. Secondly, the effect of the modified water stress scalars is evaluated for irrigated grasses and crops. The latter case includes two experiments in which the original and modified ETa estimates are assessed against relevant ground observations obtained by the processing of meteorological and soil water content datasets. The experimental results indicate that the new method is capable of correctly identifying the occurrence of irrigation and consequently improves the ETa estimation accuracy for irrigated grasses and crops. In particular, the modified method accounts for about 90 % of the seasonal ETa variance observed in the irrigated meadow and tomato field.