TAMSAT-ALERT v1: a new framework for agricultural decision support

Abstract

Abstract. Early warning of weather-related hazards enables farmers, policy makers and aid agencies to mitigate their exposure to risk. We present a new operational framework, Tropical Applications of Meteorology using SATellite data and ground based measurements-AgricuLtural EaRly warning sysTem (TAMSAT-ALERT), which aims to provide early warning for meteorological risk to agriculture. TAMSAT-ALERT combines information on land-surface properties, seasonal forecasts and historical weather to quantitatively assess the likelihood of adverse weather-related outcomes, such as low yield. This article describes the modular TAMSAT-ALERT framework and demonstrates its application to risk assessment for low maize yield in northern Ghana (Tamale). The modular design of TAMSAT-ALERT enables it to accommodate any impact or land-surface model driven with meteorological data. The implementation described here uses the well-established General Large Area Model (GLAM) for annual crops to provide probabilistic assessments of the meteorological hazard for maize yield in northern Ghana (Tamale) throughout the growing season. The results show that climatic risk to yield is poorly constrained in the beginning of the season, but as the season progresses, the uncertainty is rapidly reduced. Based on the assessment for the period 2002–2011, we show that TAMSAT-ALERT can estimate the meteorological risk on maize yield 6 to 8 weeks in advance of harvest. The TAMSAT-ALERT methodology implicitly weights forecast and observational inputs according to their relevance to the metric being assessed. A secondary application of TAMSAT-ALERT is thus an evaluation of the usefulness of meteorological forecast products for impact assessment. Here, we show that in northern Ghana (Tamale), the tercile seasonal forecasts of seasonal cumulative rainfall and mean temperature, which are routinely issued to farmers, are of limited value because regional and seasonal temperature and rainfall are poorly correlated with yield. This finding speaks to the pressing need for meteorological forecast products that are tailored for individual user applications.