During last two decades the increasing availability of remotely sensed acquisitions in the thermal infrared part of the
spectrum has encouraged hydrologist community to develop models and methodologies based on these kind of data. The
aim of this paper is to compare three methods developed to assess the actual evapotranspiration spatial distribution by
means of remote sensing data. The comparison was focused on the differences between the "single" (SEBAL) and "two"
source (TSEB) surface energy balance approaches and the S-SEBI semi-empirical method. The first assumes a semiempirical
internal calibration for the sensible heat flux assessment; the second uses a physically based approach in order
to assess separately the soil and vegetation fluxes. Finally, the last one is based on the correlation between albedo and
surface temperature for evaporative fraction estimations. The models were applied using 7 high resolution images,
collected by an airborne platform between June and October 2008, approximately every 3 weeks. The acquired data
include multi-spectral images (red, green and near infrared) and thermal infrared images for surface temperature
estimation. The study area, located in the south-west cost of Sicily (Italy), is characterised by the presence of typical
Mediterranean cultivations: olive, vineyard and citrus. Due to irrigation supplies and rainfall events, the water
availability for the crops varies in time and this allowed to perform the comparison in a wide range of the modelled
variables. Additionally, the availability of high spatial resolution images allowed the testing of the models performances
at field scale despite the high vegetation fragmentation of the study area. The comparison of models performance
highlights a good agreements of model estimations, analyzed by means of MAD (Mean Absolute Differences) and
MAPD (Mean Absolute Percent Differences) indices, especially in terms of study area averaged fluxes. The analysis in
correspondence of various crop fields highlights higher differences for low vegetation coverage and for scarce water
availability.
|