Designing agrometeorological information services in Ghana by combining local and scientific forecasting knowledge

Summary

Agriculture is a key source of food and income security for farmers in Ghana. The farming sector, however, is severely affected by climate variability and change, with subsistence farmers impacted worst of all. Providing reliable, accessible and actionable agrometeorological forecast information can enable smallholder farmers to increase their adaptive capacity in the face of climate variability and change. This study investigated weather and climate information services for smallholder farmers, particularly, how the quality, accessibility and use of such services can be improved. Several methods and tools were developed drawing on and/or integrating scientific and local forecasting knowledge systems. The study first identified and investigated trends in the variability and predictability of key local agrometeorological indicators, namely wet season onset dates, dry spell occurrence and seasonal rainfall. Forecast performance was assessed using both dynamical (i.e., ECMWF System 4) and statistical (i.e., influence of sea surface temperatures) models, which were compared to weather station observations across Ghana. Results show high interannual variability of the agrometeorological indicators, especially across the coastal zone and in northern Ghana. Performance of the scientific, model-based seasonal forecasts in reproducing the observed interannual variability was a function of location, lead time, categories and the agrometeorological indicators considered. Categorical (probabilistic) seasonal predictions provided smallholder farmers valuable information for coping with climate variability. Forecast performance was greater for the coastal savanna than for the other agroecological zones.

Next, the potential for improving forecast accuracy by combining local and scientific, model-based forecasts was explored in the Ada East district case study location. There, I found that combining a specific set of local forecast indicators could improve local forecast performance. Particularly, one-day rainfall forecasts could be improved by use of local and scientific forecasts side-by-side, depending on the set of local forecast indicators observed by farmers. The range of identified local forecast indicators offers the potential for development of other approaches to integrate local and modern forecasting systems, to improve and enrich each.

I also analysed a citizen science coproduction experiment to produce and implement an ICT-based climate information service tailored for smallholder farmers in the Ada East district of Ghana. In this case, the co-design of user-friendly digital tools (smartphone apps) and coproduction of local and scientific forecast information with and for smallholder farmers facilitated access, understanding and the usefulness of the tools and information for decision-making. Implementing such a service requires intensive collaboration between researchers and a dedicated group of farmers and extension agents to build a basis for information production and dissemination in the area of interest. In this case, the collaboration included a capacity building component, as well as monitoring and technical assistance, especially in the development phase.

Overall, the current research indicates that access to and use of reliable, locally actionable agrometeorological forecast information is possible in Ghana and elsewhere. Three key findings of this research warrant particular mention: (i) scientific forecasts need to be tailored to smallholder farmers’ needs; (ii) integration of local forecasting knowledge can add value to scientific, model-based climate information services; and (iii) a tailor-made ICT-based weather and climate information platform can effectively deliver useful and actionable information to smallholder farmers while providing a vehicle for feedback.