ICRISAT Newsletter writes:
At a recent workshop participants were trained on how to reliably simulate crop yields over space, and use aggregated forecasts to improve food security on sub-national to national scales using the CCAFS Regional Agricultural Forecasting Toolbox (CRAFT). CRAFT is a computer based decision support system to forecast short- and long-term crop yield and to analyze agricultural risks.
During the workshop, participants were shown how to generate soil, weather and observed data and import the data into CRAFT. Expertise in Geographic Information System (GIS) is a pre-requisite to use CRAFT which is a challenge for many users. Therefore, to make it user friendly, the workshop discussed on adapting Quantum GIS (QGIS) or
R functionalities to automate data preparation. Participants reported bugs while using CRAFT and these inputs were utilized in fine tuning the software.
The workshop discussions also aimed at using CRAFT as a benchmark for yield prediction in West Africa. “CRAFT not only provides an attractive software platform to investigate model complexity tradeoffs in spatial yield forecasting, it also learns from, and builds upon existing modeling skills and priorities in the region. As such it will help foster partnerships that increase the reliability and granularity of yield forecasts, and will thus help strengthen food security at coarser scales, and the delivery of farm support services into the last mile,” said Dr Pierre Sibiry Traore, Head GIS, ICRISAT-Mali and Project Leader, Capacitating African Smallholders with Climate Advisories and Insurance Development (CASCAID).
Commencing on 28 November, the two-week long workshop was organized at University of Florida, USA. It brought together West African resource persons to fine tune CRAFT using historical reference data from Mali’s cotton growing belt. Nine participants representing the Agrhymet Regional Centre, University of Bonn, Germany, University of Florida and ICRISAT took part in the training.
Source: ICRISAT “Happenings” Newsletter