Skip to main content

Exploitation of AnGular effects in Land surfacE observations from satellites (EAGLE)

Exploitable results

With the advent of a new generation of global imaging spectroradiometers capable of acquiring simultaneous multi-angle observations, both qualitatively new approaches as well as quantitative improvements in accuracy are achievable that exploit the multi-angle signals as unique and rich sources of diagnostic information. EAGLE intended to create strong interfaces with the user community and with operational production services to be implemented in Europe in the GMES framework and this will contribute to the establishment of a European capacity for global monitoring of environment. The list below shows a brief description of the work performed by the EAGLE consortium: 1. A website has been constructed and used as a tool for information exchanging. The website has been actualised continuously. All the partners have contributed with their suggestions, data and results and recommend consulting this web as key point of the EAGLE project. 2. A software package to compute biophysical parameters from PAL in IDL as well as another software package in C++ to retrieve land surface emissivity (LSE) and land surface temperature (LST) from Seviri data have been developed. Moreover, a physics-based diurnal temperature cycle model (written in C++) with only five determined parameters has been developed to remove the effects of orbit drift and inherent issues of polar-orbit satellite on the retrieved LST over a large region and for long-time period. All software packages work well with PC and we recommend using it for processing of NOAA, AATSR, Seviri and MODIS images. 3. Maps of normalised difference vegetation index (NDVI), surface temperature, surface emissivity and albedo have been calculated from PAL data. NDVI dynamics for the years 1982-2001 have been determined and expressed as Fourier amplitude and phase data per pixel. Land surface temperature product has been produced with MODIS data from March 2000 to 2006 over Europe. Project partners recommend the use of these maps to study land surface phenology in Europe. 4. This report includes the database related to the EAGLE field experiments. The database is composed by field measurements and biophysical maps (temperature, emissivity, fractional vegetation cover, etc.) obtained from AHS and ASTER data. Project partners recommend the use of this high quality database in the future. 5. Project partners recommend using their procedure in order to generate albedo maps. 6. They also recommend the method developed to estimate the LAI and fraction cover from multi-spectral data. To this end a database has been constructed by means of simulations of Modtran-Prospect-Sailh model and a selection of optimal channels and angles for LAI retrieval has been undertaken. 7. Several algorithms have been developed to perform atmospheric corrections in both the visible/near infrared and the infrared channels of AATSR. Finally, a method has been recommended to estimate the LAI and fraction cover from AATSR data. 8. A new atmospheric correction scheme has been developed to get the brightness temperatures at ground level in mid-infrared and thermal infrared channels of Seviri. Based on the concept of the temperature independent thermal spectral indices, a new physics-based algorithm has been developed to retrieve land surface emissivity (LSE) from Seviri data. This algorithm has been applied to about 30 days Seviri/MSG images and got about 30 maps of land surface emissivity in the Seviri/MSG channels 4 (3.7 ?m), 7 (8.7 ?m), 9 (11 ?m) and 10 (12 ?m) in 2004, 2005 and 2006, covering the land area going from 20W to 60E and from 0 to 60N. During the development of this LSE retrieving method, a semi-empirical kernel-driven BRDF called Ross thick-Li sparse-R model used in MODIS visible and near-infrared channels has been extended and a modified Minnaert's BRDF model has been developed simultaneously for the estimation of directional emissivity from MSG-SEVIRI retrieved bi-directional reflectivities in mid-infrared region. The intercomparison of the results obtained by these two BRDF models showed that the semi-empirical BRDF called Ross thick-Li sparse-R model is superior in accuracy to the modified Minnaert's model in the modelling of the bi-directional reflectivities in mid-infrared region and leads to an accurate directional emissivity. 9. Projects partners have developed a method to provide accurate emissivity maps from AATSR data by the combination of VIS-NIR channels. The NDVI thresholds method (NDVI) has been applied to the remote sensing data obtained in the field campaigns carried out in July 2004. 10. A practical and operative split-window algorithm has been developed to obtain LST specifically from Seviri data. Also, a set of split-window, and dual-angle algorithms have been developed for AATSR. The radiative transfer code (Modtran 4) and the ECMWF reanalysis atmospheric profiles have been proposed to perform atmospheric corrections in thermal and mid infrared channels. Since The ECMWF reanalysis (ERA) data are given at the gridded spatial resolution of 0.5 latitude/longitude for four main UTC time: 00:00, 06:00, 12:00 and 18:00, a bilinear interpolation was used to produce the spatial resolution corresponding to the images, and LST diurnal cycle model was used to generate the series of brightness temperature at ground level. 11. A theoretical method and a practical algorithm have been developed to estimate the atmospheric total column water vapour from Seviri and AATSR split-window channels. Moreover, an algorithm has been developed to estimate the aerosol optical depth (AOD) in the atmosphere from AATSR bi-directional measurements in visible and near-infrared channels. 12. A comparison of LST products obtained from AATSR, MODIS and Seviri instruments over Barrax site in Spain in coincidence of a field campaign in July 2004 has been carried out. The results showed that the LST retrieved by Seviri is in general higher (rms 4.5 K in the daytime) than that derived by MODIS and AATSR, this can be only explained by the inaccuracy of Seviri's calibration, a cross-calibration of SEVIRI infrared channels with the Terra-MODIS channels has been undertaken. A new comparison of LST has been redone with the new calibrated BTs (brightness temperatures) of Seviri over the Iberian Peninsula area and the Egypt & Middle East area in July of 2004 and July of 2005 according to the quality assurance (QCs) defined in the MODIS/Terra LST products. The results revealed that the Seviri LSTs are consistent with the LSTs extracted from the MODIS/Terra LST products and the AATSR LST with accuracy of 1-3 K depending on the different types of surface. 13. A Monte-Carlo method has been developed to simulate the directional thermal infrared exitance and a thermal infrared database was constructed with both CUPID and Monte-Carlo models. This database was used to (1) establish a simplified analytical directional thermal model theoretically relating the directional measurement to the component temperatures; (2) calibrate the developed model and (3) make the accuracy assessment. Several leaf angle distribution (LAD) functions were compared and their effects on extinction coefficients were analysed with the field measurement of LAD conducted in FIFE and two sites in Australia. A combined thermal and visible/NIR radiative transfer models has been developed to simulated the satellite measurements, the latter has been used to separate component temperatures with a multi-perceptron neural network. Moreover, an analytical parameterization of directional emission was proposed to account for the hot spot effect in thermal IR. The evaluation of the component temperatures retrieved from the simulated AATSR data using respectively the neural network based algorithm and the simplified analytical directional thermal model both developed during the period covered by this project has been performed. The results showed that both models can give a reasonable accuracy to retrieve component temperatures, but the neural network based algorithm is superior to the simplified analytical directional thermal model in the accuracy of component temperature retrievals. 14. For estimation of sensible and latent heat fluxes using satellite earth observation data, the SEBS (surface energy balance system) was implemented as an open source code and implemented in the ESA open source software BEAM as a plug-in. This version has been tested in the ESA Advanced Training Course in land remote sensing in August 2007. 15. The operation algorithm to estimate the roughness length for heat transfer is implemented in SEBS. 16. The work made with the DAIS data allowed us to analyse the feasibility of using hyperespectral airborne data for environmental studies (for example in the context of agricultural precision). 17. The GLC2000 land classification map and a digital elevation model product could initially be taken as a basis for deriving a map of fixed biophysical model parameters. It was expected that vegetation dynamics during the growing season can mainly be expressed by only two temporarily varying parameters, the total leaf area index (LAI) and fraction brown leaf area (fB). Both are parameters of the integrated soil-leaf-canopy radiative transfer model SLC, which was to be used to generate synthetic images that can be compared to actual images. HANTS was also used to reconstruct a synthesis from the harmonic components at four frequencies, namely the zero frequency (which gives the yearly mean), the yearly cycle, the half-yearly, and the four-month period. The HANTS results obtained from the SPOT-VGT data of the year 2002 have been processed in order to produce a so-called 'map of minimum vegetation'. By analysing the dynamic behaviour of the SPOT reflectance channels in the red, the near infrared (NIR) and the mid infrared (MIR), it was possible to find the moment of minimum vegetation under snow-free conditions. A crop growth model Wofost, which is implemented in the crop growth monitoring system (CGMS) with a database set up over Europe and developed for the services of the European Commission, was used to simulate, on a daily basis, land surface biophysical variables, i.e. LAI and the weight of dead leaves which can be transformed to fraction of brown leaf area.