Optimizing Planting Dates for Agricultural Decision-Making under Climate Change over Burkina Faso/West Africa

Rainfed agriculture is the main source of income for population and the main driver of the economy in Africa, particularly in West Africa (WA). The agricultural system is characterized by smallholder and subsistence farming in a context of farmers' low capacity. In water-limited regions of WA, most of the crop management decisions are made based on the perceived risk of climate and the socio-economic conditions of the farmers. Therefore, technologies and approaches in the field of agricultural water management are likely to make a difference for agricultural development and thus food security. However, only those strategies which require little resources in terms of labor and money have a chance to engage a large number of farmers. As a farming strategic decision, the planting time has the potential to sustain crop production as well as to be adopted by farmers. In the context of high rainfall variability and little irrigation options in WA, the crop planting date in WA is a crucial tactical decision for farmers and therefore their major concern. With regards to the high intra-seasonal rainfall variability in WA, early planting dates can lead to crop failure due to long dry spells which occur shortly after planting. In contrast, late planting dates have the chance to avoid crop failure but they correspond to short growing seasons which can potentially reduce crop production. In this thesis, an approach to derive an optimal planting time has been developed. Based on the crop water requirements throughout the crop growing cycle, this planting date approach uses a process-based crop model in conjunction with a fuzzy rule-based planting date definition to derive optimized planting dates (OPDs). First, by taking into account the inherent uncertainties of rainfall measurements and computations issues, three fuzzy logic memberships, which are fully determined by two fuzzy parameters each, have been developed to represent the three main criterions used to define planting date. Then, the General Large-Area Model for annual crops (GLAM) and the fuzzy rule-based planting date have been coupled with a genetic algorithm optimization technique. Finally, this has been applied to calibrate GLAM for maize cropping and subsequently to derive OPDs for maize cropping. To allow a time window for crop planting, an ensemble member principle has been applied to derive a 10-member ensemble of optimized fuzzy parameters. Burkina Faso (BF) has been selected as a case study area to derive OPDs. The performance of the OPDs approach have been evaluated by comparing maize yield derived from the OPDs method and two state-of-the-art methods which are currently in use in WA. The analysis comprises both present climate and future climate projections. Present climate data encompassed observed data and European Centre for Medium-Range Weather Forecasts (ECMWF) Interim Re-Analysis (ERA-Interim) data over BF for the period 1961-2010 and 1980-2010, respectively. Future climate encompassed eight regional climate models outputs based on the greenhouse gas emission scenarios RCP4.5 and RCP8.5 covering the period 2011-2050. Beside the climate data, soil and observed maize yield data have been involved in this study. The results show that, on average, OPDs ranged from 1 May (South-West) to 11 July (North) across the country under present climate. In comparison to selected state-of-the art methods, the results suggest earlier planting dates across BF, ranging from 10-20 days for the northern and central BF, and less than 10 days for the southern BF. With respect to the potential yields, the OPD approach indicates that an increase of maize potential mean yield of around 20% could be achieved in water limited regions in BF. However, the potential yield surpluses strongly decrease from the North to the South. For future climate projections, the OPDs approach achieves approximately +15% higher potential maize yield regardless of the Regional Climate Model (RCM) and the time horizons. When the OPD approach is used as adaptation strategy, the change in maize mean yield varies between -23% and 34% from the baseline period (1989-2008) for the majority of locations. The regional mean yield deviations strongly depend on the location and RCM, particularly for the RCP8.5 scenario. On average, negative changes of mean yield is observed. Considering the period 2011-2050, RCMs ensemble mean of yield change is -3.4% for RCP4.5 and -8.3% for RCP8.5. Mean yield decreases are more pronounced for RCP8.5 during the period 2031-2050. These findings highlight the potential of OPDs as a crop management strategy. The implementation of the presented approach in agricultural decision support is expected to improve agricultural water-related risk management in WA. The OPD approach can be used in combination with seasonal climate forecasts to provide planting date information to farmers. As farmers concern is how to limit the risk of crop failure and to sustain their crop production, the prediction of the OPDs might be of high relevance for farmers and a guide in the choice of crops and the planning of labor. However, the predictability of OPDs has to be investigated and further in-field validation of OPDs is required before being implemented as short-term tactical decision by farmers. It is apparent however, that farmers need to combine OPDs with others suited farming practices to adequately respond to climate change. Moreover, in order to efficiently support agricultural long-term strategic decision-making in WA, it is worth to perform further multi-model ensemble simulations by using additional multiple RCMs driven by multiple Global Circulation Models (GCMs) and emissions scenarios. Such investigations might contribute to better capture the magnitude of climate change impacts on crop production, thereby enhancing the development of climate chance adaptation strategies