Dr. Ralph Cordey
Marconi Research Centre
GEC Research Limited
West Hanningfield Road
Great Baddow
Essex, CM2 8HN
United Kingdom
Co-Investigators:
G. E. Keye, DRA, Farnborough
J. R. Baker, British National Space Centre
S. Quegan, University of Sheffield
G. M. Foody, University of Swansea
N. J. Veck, National Remote Sensing Centre
A. Wielogorska, Hunting Technical Services
A Study of the Potential of Multicomponent SAR Imagery for Agricultural and Forestry Studies


OBJECTIVES
Develop the methods to fully exploit multicomponent SAR data sets of agricultural and forestry areas.

Determine the radiometric information content of multicomponent SAR imagery of agricultural and forested areas by developing backscatter models.

Develop techniques for the derivation of specific target information from a multicomponent data set, and consequently determine the optimum set of measurement parameters for use in specifying future SAR missions.

Develop improved image processing techniques for the extraction of specific image attributes from these images.


PROGRESS
To date, the analysis team have received more than 50 SIR-C or X-SAR image products. They cover three sites:


Generally, the data have been of a very high quality. However, some problems have been noted with SIR-C antenna pattern corrections over Tapajos, and other artifacts (e.g., nadir ambiguities and interference in certain data). Feedback on these has been given to JPL.

The English data takes encompass agricultural and forestry research sites, while the Brazilian sites cover areas of tropical forest and regeneration.

Analysis is being conducted by GEC-Marconi Research Centre, primarily on the agricultural site at Feltwell, by RSADU on Thetford Forest, Tapajos and Manaus, by the University of Sheffield on Tapajos and Feltwell, and by the University of Swansea on Manaus.

At GEC-Marconi Research Centre, imagery of the agricultural test site from both missions has been ingested into a database. Field-averaged mean backscatter levels have been investigated to date, with particular attention given to examining differences between SIR-C/X-SAR observations early and late in the growing season, with previous results from AIRSAR campaigns in 1989 and 1991 in the middle of the growing season. Essentially the aim here has been to provide a clearer indication of the radar channels, whether polarimetric or multifrequency, which are necessary for effective crop identification. A preliminary investigation has also been undertaken of the accuracy of soil moisture retrieval using the algorithm of Pascale Dubois.

At RSADU, Thetford Forest imagery from both missions are being entered into a long- standing database which also contains various AIRSAR, ERS-1, and JERS-1 images of the forest. Comparisons are being made of the relationships between backscatter cross sections as a function of forest biomass between the different sensors. Over Tapajos, only SIR-C imagery from the October mission is currently being analyzed, because of the calibration problems with the April acquisition. The October data have been added to a Tapajos database containing JERS-1, ERS-1 satellite images, and images from the Canadian SAREX experiment. Biomass change is, again, the focus of analysis, with field data having been gathered in 1994 as part of a joint Brazil/UK ground experiment. The Tapajos analysis is also being related to work on the Manaus CSAP site. Along with ground measurements for sample plots in the Manaus region made in 1993 and 1995, an April scene is being used to validate the relationships between biomass and microwave backscatter derived at Tapajos. The Sheffield team have started examining multipolarized image segmentation, using the agricultural Feltwell area as a test site. They also have an interest in sites at Tapajos, although these were not ideally imaged by the selected Tapajos swath.

Field work at Manaus has continued. This year, the team from Swansea undertook a further field campaign and collected data from 22 forests of different ages. A preliminary analysis of Manaus imagery has been undertaken.


SIGNIFICANT RESULTS
Feltwell Agriculture

The agricultural experiments at somewhat unconventional times of the growing season have yielded some very clear and fascinating results. With relatively little ground cover by crops, we found the L-band channels of limited effectiveness in discriminating crop types. This compares markedly with observations at the height of the growing season made with the AIRSAR. Rather, we see polarimetric C-band observations as the main source of discrimination. Information contained in the HH-to-VV correlation was of most apparent value. The L- and C-band results have been backed up by simulations using a second-order radiative transfer backscatter model. It was also observed that, under these conditions, C-VV and X-VV backscatter measurements were very highly correlated.

Initial attempts at using Dubois' algorithm for soil moisture retrieval have not proved successful. However, there is probably a need for more careful consideration of the calibration status of the data before more firm conclusions are drawn.

No open-literature publications have yet been made. An account of the agricultural discrimination work has been prepared for the British National Space Centre (GEC-Marconi Research Centre Report MTR 95/51 by P. Wright et al.), and of the soil moisture retrievals for the European Space Agency ("SAR Retrieval Algorithms for Land Applications" MTR 95/36, by R. Cordey et al).


Thetford Forest

The SIR-C L-HH data over Thetford Forest support the conclusions already reached from JERS-1 and AIRSAR data and show a monotonic rise in backscatter with biomass density up to 50-80 tonnes per hectare, with saturation being reached thereafter. The behavior of SIR-C L-HV is currently being investigated to determine its dynamic range and saturation point.

Manaus

Early results from the Swansea team are currently being written up for a letter in the International Journal of Remote Sensing.


FUTURE PLANS
Across the various test sites, there remains a considerable program of work to be undertaken. The priority for the agricultural analysis is to find an appropriate route to publication. Presentation at PIERS (Innsbruck) is anticipated. Elsewhere, analysis programs are to be pursued. Joint analysis of tropical forest biomass estimation between the Tapajos and Manaus sites is planned at RSADU, while the Swansea team are to begin bringing optical data (Landsat) into their analyses. The Sheffield group plan to continue their studies with more detailed statistical analyses of the agricultural imagery over Feltwell.

Generally, SIR-C/X-SAR data have attracted much attention as a multiparameter satellite data set and it is anticipated that the interest in the UK will continue to broaden beyond the confines of the specific agricultural and forestry experiments reported here.

Table of Contents

Converted to HTML by Alvin Wong, al.wong@jpl.nasa.gov

The Jet Propulsion Laboratory
4800 Oak Grove Drive
Pasadena, Cailfornia 91109