Support to the Evaluation of the Short-term Exploitation Opportunities for Ers Data within the Agribusiness Market

The first objective of this study was to investigate the capabilities of ERS-2 SAR data (supplemented by one optical image) for producing fully validated regional area estimates of important crops. This study included the development of a sampling strategy in order to determine whether subsets of the region could be extrapolated to obtain similar area estimates to those obtained for the entire region. The second objective was to identify short-term opportunities and to make recommendations for increasing the exploitation of SAR data within the Agribusiness market sector. This entailed the identification of the information requirements of the agribusiness sector, and an assessment of the capability of current radar sensors. ERS-2 data were acquired over the majority of East Anglia (UK) during the 1997/1998 growing season. All radar data were radiometrically and geometrically corrected using NRSC’s TSAR software. An IRS LISS-III optical image was also acquired (May 1998) and geocorrected. The ERS-2 data were classified for the winter and summer period using NRSC’s SAR CLASSIFIER and the optical image was classified using ERDAS IMAGINE software to produce regional classifications. The purpose of the optical image was to improve the area estimates for the summer period. To monitor large areas on a monthly basis is currently not cost effective so a sampling strategy was devised and the regional classifications were used to extract data for 30 mini-sites. Area estimates were recalculated from the mini-sites using a defined extrapolation methodology. The crop area estimates extrapolated from the mini-sites were compared to estimates obtained for the whole region. It was assumed that if results compared well at a regional level, the sampling methodology could also be applied at a national level. The classifications have been validated and the region and mini-site area estimates have been compared with area estimates derived from [27]. The area estimates obtained for the mini-sites were considerably different to those obtained for the entire region. Further research into sampling strategies and extrapolation methodologies will therefore be required. The area estimates derived from the optical image were superior to those derived from the SAR imagery. This project indicates that current radar sensors cannot satisfy the user requirements of the agribusiness sector. Future SAR sensors, including ENVISAT (multi-polarisation data) and TERRASAR (higher spatial resolution, multi-band, multi-polarisation data), may be able to provide improved results. AGRIBUSINESS REQUIREMENTS The information requirements of the agribusiness market were identified using findings of previous user requirement studies and NRSC in-house knowledge of the market. The majority of agribusiness contributors in the study were from Spain, France and the UK. The sectors involved included fertiliser and agrochemical manufacturers/distributors, agronomy service provision companies, seed companies, machinery manufacturers, food processors, grain millers and merchants, importers/exporters and commodity brokers, insurance companies and co-operatives. Nutrient status emerged as the most common information requirement, whilst crop identification and growth stage are the least in demand for the respondents represented. Certain expected national differences emerge, such as the greater demand for water stress information in Spain, and the greater demand for disease information in the UK. This work is funded by the European Space Agency, ESRIN, Via Galileo ,00044 FRASCATI, Italy. The most desirable products in order of priority include: Nitrogen status, Crop Condition/vigour map, Soil Maps, Lodging risk map, Final yield prediction, Crop establishment map, Confirm final yield map, Crop identification map. The fact that many individual information requirements were of a similar level of importance suggests a demand for an ERS service to meet multiple information needs with its products. The preference for a suite of products has been expressed by farmer/agronomist focus groups in the UK and by individuals in agricultural institutions and agribusiness. Furthermore, given the high temporal frequency and urgency of information requirements for crop management, product dissemination is a major issue and expense. The more information requirements are met by a package of products, the more acceptably the costs of service provision and data dissemination will be absorbed. Few applications (except for example disease detection) require turn-around of less than seven days to be of value. The remaining agribusiness requirements are satisfied by data delivered days or even weeks after acquisition, which places less pressure on the structure for product dissemination, even if the medium of delivery is hard copy or CD-Rom. The overall frequency requirements are predominantly for weekly or fortnightly information during the main growing period. Many of the delivery demands made by survey respondents are ideals. In fact, a monthly (ERS-based) service may well be acceptable where the current sources are weak or non-existent. In addition, there are also significant demands for data on a monthly or annual basis, which should not present a problem to image acquisition. Recent projects to determine the validity of an operational service to farmers and growers have been based on multispectral optical instruments. The products were deemed to be of great potential value to the agricultural community, but an operational service could not be provided due to cloud cover issues. It is likely that in order to offer a robust service, optical and radar data will need to be combined. It has been shown that current radar systems can provide a method of determining “crop group” as designated by the European Union’s subsidy scheme, but are not able to discriminate at the species or sub-species level. The ability of ERS SAR to identify crop groups by means of a classification technique, based on known ground truth, is tested within this study. SATELLITE DATA AND PROCESSING