Context and objectives
Good farm management is only possible as far as raw material availability and quality are known. In animal husbandry, these raw materials are mainly the forage produced in the exploitation. Therefore, the estimate of forage stocks and quality is of prime importance.
To do so, dynamic models, describing the relationship between crop growth (yield), crop quality and environmental factors (solar irradiance, temperature, nutrient availability) has been set up. However these models often fail when growing conditions are not optimal. One solution is to calibrate the modelling with some information on the actual status of the crop during the season, such as the leaf area index (LAI), that could be provided by remote sensing.
Another solution is to define an empirical relation between a parameter such as the fraction of the photosynthetically active radiation that is absorbed (FAPAR) at a fixed phenological period, through remote sensing, and the yield of the crop at the end of the season.
The aim of the present project is to use this double approach in order to predict forage yield and quality at field scale. Once calibrated this tool will be developed for the farmer.
Expected scientific results
Among the satellite products currently available on the market, we have researched these susceptible to present a recurrence and an operational character drawn to the scale of the project. It appears that only images with a low spatial resolution typically AVHRR or VEGETATION reply to these criteria. The use of this type of images is conditioned by the existence of prairie zones sufficiently vast and the detection on these zones of seasonal change. We have put in evidence the existence in the Southeast of Belgium sufficiently zones of prairies that form a homogeneous structure drawn at the scale of the square km to justify the well funded utilisation of images of low spatial resolution. We have equally shown that it will be necessary to use daily synthesis images of the VEGETATION or AVHRR type. These images have to be calibrated, and undergo atmospheric corrections and a masking of clouds. Even if results of NDVI measures during the course of a season undertaken in this work are not very convincing due to the quality of the image that were used, it appears possible to follow the evolution of vegetation of the prairies. We have equally seen that when data are not contaminated by clouds, it appears possible to describe of photosynthetic activity differences between regions and sub regions.
The agro meteorological model developed by OGER (1994), even if its development is not ended, offer already good results. The necessary elements for the constitution of an decision assistance system are present. A pilot action developed with partners of the agricultural world could help concretise the development of a such system.