Australia-wide Machine Learning Evapotranspiration for Trees (AMLETT) model for the Lower Limestone Coast

This dataset contains the monthly evapotranspiration (ET) data at 30 meters spatial resolution from Feb 2000 to Dec 2020 for Lower Limstone Coast, South Australia, specific to plantation forests (Pinus radiata and Eucalyptus globulous) and the extent of mapped forests in the region. The data outputs employ extensive plantation forest in-situ ET data collected by CSIRO (Benyon et al., 2006) and machine learning to create a model that estimates forest ET across the proposed timeseries. The approach used to undertake this has been developed recently in the Murray-Darling Basin by Doody et al. (2023), referred to as the AMLETT model (Australia-wide Machine Learning ET for Trees model) which is applicable to use in any location where field tree ET data is available to help train the machine learning processes. \n\n\nLineage: Field-measured evapotranspiration (ET) data from 21 sites from 2000 to 2008 were used to evaluate a remotely sensed ET dataset (CMRSET). In order to improve the estimations of CMRSET (https://portal.tern.org.au/metadata/21915), a scale factor was calculated to evaluate the CMRSET performance with field data. Landsat 7/8 surface reflectance, CMRSET and climatic data were collected from 2000 to 2020 for the Lower Limestone plantation area. Those data were coupled with a machine learning method to predict variations of the scale factor. Finally, a new ET dataset was generated, specific to Pinus radiata and Eucalyptus globulous, by dividing the predicted scale factor from CMRSET.