Assessing the impact of national food security policies on irrigated rice cultivation in Senegal using advanced remote sensing and modelling technologies

In the aftermath of the rice crisis in 2008 the Government of Senegal launched the program The Great Offensive for Food and Abundance running from 2008 to 2015. The goal was to achieve self-sufficiency for the major crops that are imported by making large investments in the agricultural sector aiming self-sufficiency by 2018 through intensification and area expansion incentives. The goal of this study was to analyze policy-induced trends in rice area, cultivation patterns and choice of varieties in the Senegal River Valley using advanced remote sensing and modelling techniques. We deployed the PhenoRice algorithm and time-series of 250 by 250 meter resolution MODIS satellite imagery between 2003 and 2016 to analyse cultivated areas in the dry and wet seasons, start and end of the rice season and length of the growing season. The results were validated with long term observations of farmer cultivation practices in two regions and with official statistics of rice cultivated areas. PhenoRice rice areas are reasonably correlated with the reported statistics (r2 = 0.71) and it was concluded that the algorithm is able to detected trends. However, the actual detected rice area is consistently lower than the reported areas. This was attributed to the mismatch between field size and pixel size as well as the fact that the schemes are relatively small and located in dry, desert-like environments. Before 2008 cropped areas during the dry season were minor (<1,000 ha). However from 2008 onwards dry-season rice cultivation was promoted and by 2015 the dry season area (28,000 ha) had become more important than the wet season (17,000 ha). This occurred mainly in the delta region where also rice double-cropping has become a common practice (approximately 6,000 ha). Yield potentials in the dry season are higher while in the current state river flows are sufficient to meet the demands. Finally we analyzed the start, end and length of the rice seasons. We found a significant reduction in average length of the wet season by 0.53 days y-1; this is explained from the choice for short-duration variation to cater for rice double-cropping within the current climate limits.