Variations in rice cultivation practices in the Senegal River Valley between 2003 and 2014: an analysis based on MODIS time series

As a result of government policies irrigated rice cultivation in the Senegal River Valley has shifted to from the wet to the dry season. Moreover, private and government investments have led to doubling of the total rice acreage since 2007 and cropping intensity has increased from1 to 2 crops per year. However, temperature extremes (both hot and cold) during the rice growing season frequently lead to yield losses or failure. Better knowledge is required to where and when rice is cultivated and which areas are mostly affected. The use of multi-temporal satellite imagery can provide useful information to track the dynamics of rice production, including extent of cultivated area per season, agricultural practices such as cropping intensity, and crop establishment and harvest periods. This information will ultimately be linked to temperature data bases to assess zones that are frequently exposed to climatic risks. Improved spatial knowledge on climatic risks will be exploited by extension services to target recommendations on the choice of rice varieties and optimal planting dates. In this context, inter-annual variation of agricultural practices in the Senegal River Valley were analyzed by applying the PhenoRice algorithm (Boschetti et al., 2014) to time series of 16-days composite vegetation indexes 250m resolution MODIS images (Products MOD13Q1 and MYD13Q1) acquired between 2003 and 2014. The algorithm allows to identify the main rice cultivated zones and to estimate the dates of occurrence of the main phenological stages (crop establishment, crop emergence, flowering) on the basis of analysis of spectral indices (SIs) time series. Estimated acreages of rice cultivated in wet and dry seasons areas were compared with official statistics, while rice planting and harvest estimates were analyzed with respect to a database of farmer surveys that includes more than 1000 reports of seeding or transplanting and harvest dates for a period of 10 years. Interannual changes of cultivated rice areas statistics showed a clear correlation with those estimated from PhenoRice (Hot season R2 =0.89; Cold season R2 =0.50; all R2 =0.76 ), and clearly highlighted a strong increase of cultivation of rice in the Senegal River Valley during the hot and dry season in the last years (Sowing dates between February and April). This result testifies the suitability of the PhenoRice algorithm to track spatio-temporal changes in rice cultivation practices, and its potential usefulness for rice monitoring in regions lacking high quality information on cultivation practices.