Satellite Derived Evapotranspiration Is Advantageous for Drought Detection

Analysing Drought Impact by Combining Evapotranspiration and Vegetation Signal

A combination of satellite derived evapotranspiration and vegetation signal improves locating and analysing drought affected areas. In the scope of drought monitoring, satellite vegetation signal, e.g., fraction of vegetation cover is a very reliable indicator of vegetation health. However, for assessing drought stress (i.e. the current conditions vegetation is exposed to), evapotranspiration is also a very important variable. In this sense, the difference between reference and actual evapotranspiration can correlate with the drought stress as experienced by vegetation.

Next figures display satellite derived evapotranspiration (top) and vegetation (bottom) from LSA SAF; all products are prepared on a daily basis from Meteosat satellite. Actual evapotranspiration (MET, LSA-302), difference between reference (ET0, LSA-303) and actual evapotranspiration, fraction of vegetation cover (FVC, LSA-421) and FVC anomaly (difference between current vegetation and its long-term average in 2004-2016) are shown for August 2017 and 2018. ET0 and ET are calculated as monthly total in the same August periods. ET approaches ET0 when there is ample water. Therefore, a large difference between the two indicates a shortage of water, which could lead into drought when dry conditions are persistent.



Evapotranspiration difference (ET – ET0) and FVC anomaly both highlight that mostly areas of S and SE Europe experienced strongest drought in August 2017. The local details might differ but the general picture is quite consistent for both evapotranspiration difference and FVC anomaly maps. Only negative anomaly is shown in FVC anomaly figures, i.e., positive anomaly is masked. Conclusions are similar for August 2018, but this time the vegetation and evapotranspiration drought signal is present in N, W and Central Europe. 




Temporal evolution of evapotranspiration difference and FVC anomaly was analysed at two European locations (Metlika, SE Slovenia and Herentals, N Belgium) in 2017 and 2018 (next Fiigure). Both locations have homogeneous land cover where the dominant land type is fields (grasslands and crops). Droughts typically occur only if accumulated evapotranspiration difference passes the threshold value. The figures illustrate that evapotranspiration difference can give an earlier warning for the developing drought as evapotranspiration drought signal precedes the drought signal from vegetation. However, the temporal lag between the drought signals from evapotranspiration and vegetation is highly variable from location to location and depends on land cover fraction, vegetation type and pixel homogeneity.



Combined evapotranspiration and vegetation satellite imagery can provide a fairly accurate overview of drought conditions on a larger geographic scale. For analyses on a local scale, caution is needed: local characteristics, e.g., land cover and type, vegetation type as well as irrigation need to be taken into account. However, any interpretations should be done having in mind that different thresholds as a function of vegetation types need to be considered. Therefore, interpretations are simpler over satellite pixels that have homogeneous land cover and we should analyse drought signals for every vegetation type separately. Certain vegetation types, e.g., Mediterranean vegetation have a higher tolerance for vegetation stress and are able to withstand higher shortages of water than vegetation that grows in more humid and less hot climates.