Visit to CGIAR centres: CIMMYT and CIAT

Andrew Challinor
Feb. 2002

In accordance with the PROMISE objectives regarding links to end users and the exploration of the potential of seasonal forecasting in the agricultural context, visits to the major CGIAR (Consultative Group on International Agricultural Research) centres are being carried out. This report summarises the visits to two such centres. CIMMYT is the centre responsible for maize and wheat improvement, and CIAT specialises in tropical agriculture.

An introductory seminar entitled ÒSeasonal forecasting for agricultureÓ was presented at both institutions, and meetings with individual scientists formed the basis for exploring research issues and potential projects and collaborations. A questionnaire was also distributed in order to obtain feedback on some of the opportunities provided by seasonal forecasting, and any foreseeable problems which this research area presents. An important issue is the matching of spatial the spatial scale of the forecast to the spatial scale of the application. Rainfall is the forecast variable of most interest, although this does vary with application. Another important research area is the exploitation of the probabilistic nature of seasonal forecasts; in some cases it was only after discussion that this was seen by agricultural specialists as an opportunity rather than a barrier. Much progress was made in such discussions as they opened up new possibilities for collaborative work. The following paragraphs summarise these ideas and opportunities, by giving examples of each type of application discussed.

Yield forecasting: CIAT scientists expressed an interest in using seasonal forecasting in conjunction with their crop modelling expertise to develop a risk analysis tool for beans in Central America.Depending on the density of sites available, a downscaling and/or regional modelling study could be performed in order to obtain the input data required for the crop model. The probabilistic seasonal forecast output could be used to assess the risk associated with the growing of different varieties. Whilst drought tolerant varieties are ideal in low-rainfall years, they produce reduced yields (when compared to standard varieties) if rainfall is nearer to climatology. A forecast lead time of a few months would be required in order to assess the risk associated with each of these varieties.

Crop scheduling: A knowledge of when dry spells are likely to occur would allow growers to plan planting for a suitably dry spell just prior to rain. Planting could also be scheduled such that the risk of water stress at critical phenological stages such as flowering is minimised. Discussions at CIMMYT revealed that dry spells of three weeks or more are the most significant for this type of planning.

Disease control: If the potential spread of crop diseases could be more accurately predicted, spraying could be undertaken earlier, and damage to the crop limited. Specific example for the case of cassava were discussed at CIAT. The principle diseases for cassava are correlated with rainfall and humidity, and lead times of three weeks could be useful for this application. The need for high spatial resolution presents the greatest challenge in this case. There was also an interest expressed in assessing the impacts of climate change on the pattern and spread of crop diseases.
Another example, discussed at CIMMYT, is the use of seasonal forecasting to determine years for which maize streak virus is likely to present a serious problem. This virus can reduce yield to near-zero, and in one third of seasons it has an important impact on productivity.

Participatory research: An idea that arose from more than one source was to use local growers knowledge in conjunction with forecast information in order to maximise the value and accuracy of the forecasts. The integration of scientific agricultural knowledge and local knowledge is being explored by a group of scientists at CIAT. A day was spent at a participatory research meeting in a school house in rural Colombia (in the Cali area), where a scale model of the local area was used to facilitate this integration of knowledge. A collaboration where this idea is applied to seasonal forecasting is being explored.

Visit to IDEAM

A visit was also undertaken to Colombian national institute for hydrology, meteorology and the environment (IDEAM). Although the institute does not currently undertake crop modelling studies, there is an interest in the work of PROMISE in this respect. The seasonal forecasting carried out at IDEAM uses empirical models, although there is a group looking at using the ECMWF forecasts.