Satellite and eddy covariance analysis reveals short-lived evapotranspiration changes after fire in Mediterranean woodland

Ecosystem evapotranspiration (ET) varies through space and time in response to environmental gradients and disturbances like fire. Field-based techniques (e.g. eddy covariance) can directly measure how ET responds to fire; however, these measurements are localised and only represent small areas. Remotely sensed ET products can potentially evaluate fire impacts on ecosystem ET, but their performance for this task remains largely unassessed. This paper uses three widely available remote sensing products: the CSIRO MODIS ReScaled ET (CMRSET), MODIS ET (MOD16), and Penman–Monteith–Leuning (PML), to assess ET changes caused by a fuel reduction burn in a Mediterranean woodland in comparison to the changes detected using eddy covariance measurements. All remotely sensed ET products were first compared to ET measured at the Gingin OzFlux eddy covariance tower at annual and monthly timescales. Then, we tested the ability of these products to identify fire-induced changes by comparing ET between burned and unburned areas. At the annual timescale, CMRSET, MOD16, and PML agreed well with eddy covariance with R² ranging from 0.44 to 0.56 and RMSE equalling 3.0 to 4.0% of the annual eddy covariance estimate. However, the agreement was worse at the monthly timescale with R² ranging from 0.14 to 0.54 and RMSE equalling 10.0 to 18.0% of the monthly eddy covariance estimate. In addition, CMRSET exhibited a two-month phase lag with measured ET. Eddy covariance suggested that the fuel reduction burn decreased ET by 56 mm over the first year post-fire (i.e. 10% of the annual pre-fire ET); however, ET quickly returned to pre-fire conditions as vegetation regrew. All remote sensing products could detect the direction, timing, and duration of the fire disturbance, but PML showed the greatest potential in quantifying ET changes over the first year following the fire, with an error of 9%, or approximately 5 mm/year. Based on these results, we are confident in PML capability to assess fire impacts on ecosystem ET in Mediterranean drylands.