Modelling evapotranspiration in agricultural crops from data-fusion of high-resolution airborne and satellite imagery

Organised by Laboratory of Geo-information Science and Remote Sensing

Wed 15 March 2017 10:20 to 10:50

Venue Lumen, gebouwnummer 100
Room 1

By Ramin Heidarian Dehkordi (Iran)


Through the growing human population, freshwater availability is reduced especially in water limited areas which can be exacerbated with climate change. Monitoring evapotranspiration can provide useful information on water availability for water management applications in precision agriculture. Therefore, provision of detailed evapotranspiration information is gaining increasing attention.

Remote sensing technology can meet the requirements of precision agriculture in provision of evapotranspiration information, although there are several challenges in the use of remote sensing data. One of the limiting problems, especially in European countries and The Netherlands, is existence of the clouds which highly affect the availability of satellite imagery. In addition, generating remote sensing data which meet both spatial and temporal requirements of precision agriculture is not possible from an individual remote sensing sensor.

This research investigated the use of high-resolution airborne remote sensing data-sources (including Unmanned Aerial Vehicles (UAV)) in evapotranspiration monitoring by adopting remote sensing data fusion techniques and including in-situ sensor datasets (e.g., meteorological datasets). Acquired images by the MicaSense sensor (4.7cm spatial resolution) mounted on UAV platform was fused with collected data from an airborne-manned sensor (0.5m spatial resolution) and used in the ETlook model to estimate evapotranspiration of a potato field in the south of The Netherlands over the 2016 growing season. Evapotranspiration was also modelled using space-borne sensors including Landsat-8 and Sentinel-2 and the results were compared with airborne ET estimates. The comparison of the scenarios shows that there
is a relationship between the spatial resolution and observed variation in ET estimates.

The usefulness of high-resolution airborne datasets was further analyzed regarding their ability in minimizing the effect of background in ET estimates over the entire growing season. Highresolution airborne imagery illustrated the robustness of ETlook model in separation of ET estimates to evaporation and transpiration concepts. However, a limitation of the ETlook model was identified over the bare soil areas. Derivatives from ETlook model were compared with field data and ground-based measurements of Reusel station. Estimated relative root
zone moisture from remote sensing sensors showed several overestimations and underestimations compared to the observed soil moisture content measurements within the experimental field.

The result of the current study indicates the added value of high-resolution airborne imagery in provision of the accurate information about the crop status which serves as a sustainable agriculture intensification method in water management applications. Development of proper procedures for both data processing and computation of airborne derivatives (e.g., airborne surface albedo) together with the improvement of the ETlook model over the bare soil areas, can improve remotely-sensed estimates of ET using high-resolution airborne images.