Improve the Production Forecast

Climatic change may engender several kinds of impact on agriculture (Delecolle et al., 2000):

• On production, in terms of quantity, and also of quality, by direct effect of the climate on crop productivity and also indirect effect on diseases, insects, and weeds.

• On various upstream modifications of the consumption of irrigation water, fertilisers, herbicides, and pesticides, and downstream in the quality of the products available and/or sold.

• On the environment, particularly if the frequency and intensity of rainfall, combined with an increased use of nitrogen, mineral elements and pesticides leads to a leaching or a run-off of these substances.

• On the rural environment, if climatic change forces the abandonment of certain species or the introduction of new ones, the modification of land-use, and the development of water management projects.

The integration of these various components represents the main challenge needing to be addressed and coordinated in the near future.


Various projects have already been carried out to evaluate the impact of climate change on agriculture. They generally incorporate the effect of the increase in CO2 on photosynthetic production, in addition to the direct effect of modification of climatic factors (particularly temperature, radiation, rainfall). The expected effects are highly variable depending on the cultivated species or forest ones studied, and the regions under consideration. They also depend on the climatic scenarios being used. A global increase in CO2 generally leads to an elevation in crop productivity. This latter increase depends on carbon metabolism : it is more marked for C3 plants (like wheat) which are more frequent in temperate latitudes, than for C4 plants (like maize), which are more current in tropical agriculture.

Thus, if we forecast an increase in rice productivity in Northern countries in relation to the broadening of the period favourable to the crop, we forecast a fall in the productivity of this cereal in numerous countries in South-East Asia, particularly linked with the negative effect of high temperature induced spikelet sterility (Matthews et al., 1997). The current assessment of this work has been carried out in the context of the work of the IPCC but it remains incomplete.

The empirical models based on statistics available in agrometeorology have been established with implicit combinations between climatic variables which will have to be discussed again because of climate change. Thus, we want to emphasize the fact that predictions must increasingly rely on crop simulation models that are likely to effectively combine the differentiated effects of CO2 and the various variables of the climate on the physiological processes. To address this issue, we propose sophisticated dynamic models which simulate and integrate specific mechanisms:

• Ecophysiology of the aerial parts of plants (development, aerial growth, elaboration of yield),

• Soil functions in interaction with the underground parts of plants (root growth, water uptake, nitrogen uptake, transfers),

• Management of interactions between cultivation techniques and the soil-crop system, whether they concern the contribution of water, fertiliser or climate.

Integrating the accumulated knowledge in terms of the influence of climate, the soil, and culture practices on production, these agronomical models can be provided with climatic data from the general circulation models. They should thus provide a production forecast per type of crop, the quantities of water or fertilisers consumed, and allow to test strategies for adapting to modifications in the environment. The results from this will need to be validated using numerous experimental results. Taking into account the indirect effects of disease or insects, as well as weeds, however, still remains to be achieved in numerous cases.

Numerous of these models exist around the world (Hoogenboom, 2000). Without wishing to be exhaustive, we can cite CERES models (for maize, wheat, millet, sorghum or rice) and the CROPGRO models (for soya, peanuts, etc.). The CERES model has been used to quantify the consequences of climatic change in France on the production of wheat and maize (Delecolle et al., 1995).

STICS software (Brisson N. et al., 1998 - Brisson N. et al., 2002), perfected in France by a multidisciplinary team at the INRA (National Institute for Agronomic Research) with the collaboration of other research organisations and the agricultural profession is a multi-crop model. Coupling it with parameters from the soil database and spatialised meteorological parameters (Perarnaud et al., 1997, Ruget et al., 2001), enables the estimation of grass production in France.

The international community should be able to take advantage of the results of the research carried out in the context of crop modelling. However, the use of these models must be approached with caution, ensuring that certain input parameters (types of soil, meteorological data, etc.) are adapted to the local or regional conditions of use. Therefore, retrospective evaluation of the models with series of observed data, together with their sensitivity and uncertainty analyses, are essential steps to build up confidence in model predictions.


Originally crop simulation models were designed to operate on a homogenous plot of land. The use of these models to evaluate the impact of climatic change on a regional scale (on production, water consumption or nutrient use) necessitates the use of specific techniques for spatial representation. It is essential to apply the model to a spatial unit defined according to use (simulation unit) and then aggregate the results (yield or quantity of water consumed by the crop) at the regional scale. This type of application necessitates a considerable volume of input data to understand the spatial and temporal variability of the studied parameter. Databases and geographical information systems (GIS) are essential tools in implementing such applications. However, it is appropriate to analyse the level of precision and therefore the possible level of simplification in terms of each layer of GIS information (soil, management practices or meteorological parameters). One of the greatest difficulties lies in taking account the spatial heterogeneity of the soil, which is not always available in digital form and might suffer from a lack of precision in terms of georeferencing. Moreover, the calculation of the different parameters of the soil water balance (like the available water), although estimated from the rules of pedotransfer, need improvement.

In the Sudano-Sahelian region, agricultural productivity is closely linked to the variations of the climate and, more precisely, to the intensity and duration of periods of drought. Thus, to forecast the yield of millet in the Sahelian area, the precision applied to water stress is essential regarding the forecast of the volume and the spatial and temporal rainfall distribution, and regarding the plant's use of water. However, other determinants of the crop production such as the cultivar, sowing date, and plant density make forecasting more complex. The development of models which account these factors enables the evaluation of risks and technological scenarios in real time or for the future (Samba et al., 2001). We can then for example visualise delayed sowing in certain areas or identify areas which may suffer a production deficit by using remote sensing and modelling outputs. Figure 6a presents the divergence between the beginning of the 1998 season and the average for the years 1961 to 1990 in countries in the Sahelian area, allowing us to visualise early or late sowings in certain areas in relation to a reference year. This model may be used to forecast the yields of the current year by adjusting the historical date and the provisional climatic data in real time. To counter the lack of rainfall data, rainfall estimation methods using satellite imaging (Meteosat) are used. Figure 6b illustrates the differences in yield estimated (at the end of the growing season) between the current year and the average of the last thirty years. This device, illustrated here for short term forecasting, will also allow us to evaluate the adaptation of Sahelian agriculture to climatic change by using GCM outputs.

The use of crop simulation models (Hansen and Jones, 2000) fed with spatialised information from diverse origins with varying degrees of uncertainty, leads to the propagation of errors which may distort the final results. It is therefore appropriate to carry out theoretical research to quantify these errors and try to minimise them.

Was this article helpful?

0 0

Post a comment