The system, which will predict yields three months before the harvest, is being developed in collaboration withthe British National Space Centre and the software company Logica, with a pounds 1.2m grant from the Department of Trade and Industry.
There are about 170,000 hectares of sugar beet in the UK, grown by 10,000 farmers. British Sugar has a contract with each one. The first step in forecasting yields involves keying Ordnance Survey map references from farmers of where sugar beet is growing into a geographical information system. This is matched up to satellite pictures of the UK to pinpoint the sugar beet fields on satellite images taken once a month in the May to August growing season. The images are then analysed to determine the amount of vegetation and its rate of growth.
This vegetation index is compared against a growth model of sugar beet, developed by the University of Nottingham and the Broom's Barn Experimental Station in Bury St Edmunds, which relates the amount of vegetation to the size of the root.
Weather data fromthe Meteorological Office and information on soil types from the Ordnance Survey are also factored into the yield-forecasting system.
'People have talked about using satellite observation in agriculture for 10 to 15 years,' says Ray Harris, who heads the project for Logica. 'But it isn't enough to just look at satellite images of crop fields. That only tells you what is growing there; it gives no indication of yield.'
A feasibility study has already been completed and the full-scale system is being developed using historical yield, satellite and weather data.
British Sugar spends pounds 190,000 per year forecasting yields by taking samples and applying the sugar beet growth model to aerial photographs.
'Yield forecasting is important because the whole of the business operates on the total amount of sugar beet produced,' says Richard Turner, agricultural services manager at British Sugar. The weather is the main factorinfluencing yields, which can vary from six to eight and a half tonnes of refined sugar per hectare.
All the UK crop is processed by British Sugar at its 10 factories between September and February. 'The yield-forecasting system will give us much better control over our business, and enable us to cut costs in a number of areas,' says Mr Turner. 'We rely on seasonal contract labour, and knowing when the first beet will be harvested and the yield means we can hire people at the right time.
'We can also fix better contracts with suppliers of our otherraw materials - lime, which is used in the purification process, and fuel, which powers the factory generators,' he says.
British Sugar's existing forecasting methods give it a reasonable idea of total yield, but Mr Turner says the satellite system will allow the company to forecast yield down to individual factory level. 'At the moment we only monitor 1 per cent of the crop, and can only do sampling from August onwards. This means we don't know in advance if we should divert beet from one factory to another if yields are uneven in different parts of the country.'
European sugar beet production is controlled by the Common Agricultural Policy, with British Sugar having a quota of 1.444 million tonnes per year. Any excess may be sold on the world market. Prices are much lower than protected EC prices, but Mr Turner says knowing yields in advance will allow the company to strike bargains earlier in the year when prices are higher. It will also allow it to get a better price for animal feed, a by-product made from sugar beet pulp.
British Sugar expects to use thesatellite system for real next year, although Mr Turner says the company will continue to use its existing methods until the new system is proved.
Logica and British Sugar also hope to sell the system to other European sugar beet processors, which face the same constraints. 'We believe this system will be one of the first commercial applications of remote sensing data in agriculture,' says Dr Harris.
'A great part of its success hangs on the fact that British Sugar and its partners at Nottingham University and Broom's Barn Experimental Station have spent a lot of time developing the sugar beet growth model. The same principles could now be applied to other crops.'
The next two candidates are peas and potatoes, which have market structures similar to sugar beet. Dr Harris thinks the techniques could also be applied to large-scale forecasting of yields of crops like coffee and cocoa, where such predictions could have a dramatic effect on commodities markets.
(Photograph omitted)Reuse content