Derived Optimal Linear Combination Evapotranspiration - DOLCE

DOLCE V3.0 is an observationally constrained hybrid Evapotranspiration (ET) dataset with uncertainty estimates. It is derived by merging four global ET datasets:

- ERA5-land (Muñoz S. J., 2019)
- FLUXCOM METEO+RS (Jung et al., 2019)
- GLEAM v3.5a and GLEAM v3.5b (Martens B. et al., 2017).

The combination of these parent datasets is based on their performance against flux tower ET, and accounts for their error dependency. The weights assigned to the parent datasets vary by seasons (Dec-May / Jun-Nov) and climate regimes based on a simplified climate regime map. DOLCE V3.0 provides time-variant estimates of its uncertainty errors. 
Similar to DOLCE V2.1, DOLCE V3.0 is monthly, has 0.25-degree spatial resolution, available over 1980 - 2018, and covers the global land.
DOLCE V3.0 is suitable for time-series and trends analysis. The reduced number of participating datasets compared to previous versions is to ensure temporal consistency throughout the covered time period.
In comparison, DOLCE V2.1, which was derived from the combination of 11 global datasets, is suitable for climatological analysis, and is less biased against flux tower observations than DOLCE V3.0. However, temporal inconsistencies around the years 2001 and 2016 were detected in DOLCE V2.1 over the tropics. These inconsistencies result from the different lengths of records of the parent datasets. We, therefore, advise the user to select an appropriate dataset for their research based on the application.

Acknowledgements
We thank Franklin (Pete) Robertson (NASA Marshall Space Flight Center) for his valuable contribution to DOLCE V3.0. 
This dataset was created by Dr Sanaa Hobeichi as part of the Centre of Excellence for Climate Extremes (CLEX) Drought research program.