Integrating indigenous knowledge with scientific forecasts in Lushoto, Tanzania

Working Paper 103.pdf




Improving food security needs appropriate climate related risk management strategies. These include using climate information to guide farm level decision-making. Progress has been made in providing climate services in Tanzania but there are significant gaps with regard to downscaled location specific forecasts, as well as generating timely, reliable and user friendly information. Majority of the farmers have been using indigenous knowledge (IK) forecasts to predict weather through observing the behavior of large animals, birds, plants, insects, and the solar system. IK is not often documented and is mainly sustained from one generation to another through oral history and local expertise, creating a wide inter-generational gap between its custodians and the young people. This study identifies and documents existing IK in weather forecasting in Lushoto district, northern Tanzania, and aims at promoting the integration of IK and scientific weather forecasting for climate risk management. Historical rainfall data was used in combination with data collected through household surveys, focus group discussions and key informant interviews. Majority of the farmers (56%) indicated that weather forecasts using IK were more reliable and specific to their location compared to scientific forecasts. Comparison was made of the seasonal March-April-May (MAM) forecasts in 2012 from IK and Tanzania Meteorological Agency (TMA), with both approaches predicting a normal rainy season. The IK forecasts were, however, more reliable in the long rainy MAM season compared to the short rainy October-November- December season. To improve accuracy, systematic documentation of IK and establishment of a framework for integrating IK and TMA weather forecasting is needed. There is also a need to establish an information dissemination network and entrench weather forecasting within the District Agricultural Development Programmes
Download the working paper here