Developing seasonal rainfall scenarios for food se

Authors: Gregory J Husak Christopher C Funk Joel Michaelsen Tamuka Magadzire Kirk P Goldsberry

Publish Date: 2013/01/30

Volume: 114, Issue: 1-2, Pages: 291-302

PDF Link

Abstract

Rainfed agriculture in SubSaharan Africa accounts for 95  of the local cereal production impacting hundreds of millions of people Early identification of poor rainfall conditions is a critical indicator of food security As such monitoring accumulated seasonal rainfall gives an important midseason estimate of final accumulated totals However characterizing the remaining uncertainty in a season has largely been ignored by the food security community This paper presents a new technique describing rainfall conditions over the duration of a cropgrowing cycle by combining estimated rainfalltodate with potential scenarios for the remaining season based on available satellite rainfall estimates the common tool for rainfall analysis in Africa The limited historical record provided by satellite rainfall estimates using previous seasons provides only a coarse view of likely seasonal totals To combat this scenarios developed by bootstrapping dekadal data to create synthetic seasons allow for a finer understanding of potential seasonal accumulations Updating this throughout the season shows a narrowing envelope of seasonal totals converging on the final seasonal result The resulting scenarios inform the expectations for the final seasonal rainfall accumulation allowing analysts to quantify and visualize the uncertainty in seasonal totals Giving decision makers a tool for understanding the likelihood of specific rainfall amounts provides additional time to enact and mobilize efforts to reduce the impact of agricultural droughtMuch of the developing world relies heavily on rainfed agriculture In SubSaharan Africa rainfed farms account for 95  of the local cereal production Wani et al 2009 employing 70  of the people Gelb and World Bank 2000 Wani et al 2009 When widespread drought strikes crops millions of people mostly women and children face hunger Such was the case in 2009 when poor rainfall contributed to an increase of 53 million foodinsecure people in SubSaharan Africa according to the annual “State of Food Insecurity in the World” publications by FAO Food and Agriculture Organization of the United Nations 2008 2009 Fortunately effective early warning in 2009 helped to provoke effective food aid responses totaling 36 million metric tons1 Effective and early rainfall monitoring based on satellite rainfall estimates Funk et al 2003 Artan et al 2007 plays an important role in guiding humanitarian responses These estimates while imperfect Dinku et al 2008 capture the spatial patterns of rainfall and can be compared against previous years to identify anomalous rainfall Funk and Verdin 2009 As the satellite record increases with time there is a more robust archive of estimates to compare with current values Exploiting this historical information effectively can lead to more accurate and timely drought early warning saving lives and livelihoods and limiting the economic disruptions produced by extreme weather This paper attempts to combine crop phenology information and satellitebased rainfall estimates to quantify cropspecific rainfall accumulations in a probabilistic framework Monitoring rainfall over the course of the growing cycle of a crop provides a proxy for crop performance that can be compared to other seasons in the satellite record Generating endofseason results using rainfall from previous years to fill out a season helps to define both the expectation and variability in the remainder of the cropgrowing season This variability gives a sense for the uncertainty related to the estimated expectation an important element for decision makers to determine the appropriate responseThe following section describes the rainfall estimates and how the season is parameterized using this rainfall Then a brief evaluation of seasonal rainfall totals based on the satellite rainfall record is described for Southern Africa highlighting regimes prone to rainfallrelated crop yield reductions Section 3 details the scenario monitoring technique with an example from the 2009–2010 growing season for Southern Africa The results are displayed through a series of maps presenting different characterizations based on the scenario analysis Section 4 steps through the season for a single administrative zone of Zimbabwe capturing the narrowing cone of uncertainty includes a validation of the technique by comparing endofseason conditions to the forecast range developed in midseason and raises potential issues going forward The final section summarizes the major findings and potential applications of this researchCurrent estimates show that nearly 160 million people are undernourished in SubSaharan Africa Food and Agriculture Organization of the United Nations 2011 In Southern Africa more than one in three people face undernourishment Food aid to SubSaharan Africa in 2010 accounted for nearly twothirds of all global food aid with 72  of the received amount coming in the form of “Emergency Aid” World Food Programme 2011 Providing tools for analysis to inform decisions regarding the correct amount of aid and timeframe for delivery to prevent undernourishment can directly impact millions of livesThe analyses presented in this report are based on the National Oceanic and Atmospheric Administration NOAA satellite rainfall estimates RFE2 Xie and Arkin 1997 The RFE2 integrates daily rainfall observations geostationary infrared cold cloud duration rainfall fields and passive microwave rainfall retrievals to produce 01° daily grids of rainfall for Africa While satellite rainfall fields are associated with significant bias in areas of complex terrain Dinku et al 2008 this bias can be reduced using highresolution climatologic rainfall fields Funk et al 2007 to track agricultural crop water deficits Funk and Verdin 2009 Furthermore in areas without much complex topography such as Zimbabwe the RFE2 has been shown to be a preferred rainfall estimate Dinku et al 2008 The RFE2 drives a number of monitoring products used in the food security analysis such as the standardized precipitation index SPI and water requirement satisfaction index WRSI The ability to use RFE2 for quantitatively monitoring food security in Africa is affirmed in literature Tadross et al 2005 Verdin et al 2005 Sawunyama and Hughes 2008 Tadesse et al 2008Algorithms exist to utilize RFE2 data to establish some seasonal parameters for Africa Onset of rains establish the start of season SOS for each location Detection of SOS for Southern Africa begins in September which is the earliest possible planting date This research uses the technique developed in West Africa AGRHYMET 1996 and reinforced in WRSI literature Verdin and Klaver 2002 Senay and Verdin 2003 defining SOS as the first dekad receiving 25 mm of rainfall followed by two dekads summing to 20 mm While there may be local deviations from this as a viable SOS it is a widely supported approximation Tadross et al 2005 Tadesse et al 2008 Funk and Budde 2009 Crespo et al 2011 Harrison et al 2011The WRSI is a measure of the fraction of the water demand of a crop that is being met by soil moisture and rainfall Frere and Popov 1986 Senay and Verdin 2003 Verdin and Klaver 2002 The WRSI incorporates components of rainfall potential evapotranspiration and soil moisture Additionally a crop coefficient specific to the crop type determines the water demand of the crop for the phonological cycle from the SOS for the entire LGP The WRSI is a leading indicator of crop yield in monitoring African food production and is therefore a logical product to use in defining growing season characteristics for this study It should be noted that for many regions there is a good correspondence between seasonal rainfall totals what this study investigates and end of season WRSI values Verdin et al 2005

Keywords: