Increasing cropping options in seasonal floodplain wetlands of sub-Saharan Africa: A remote-sensing approach for assessing available green water for cultivation

Producing more food for a growing population requires sustainable crop intensification and diversification, particularly in high-potential areas such as the seasonal floodplain wetlands of sub-Saharan Africa (SSA). With emerging water shortages and concerns for conserving these multi-functional wetlands, a further expansion of the cropland area must be avoided as it would entail increased use of blue water (surface and groundwater) for irrigation and infringe on valuable protected areas. We thus advocate an efficient use of the prevailing green water (plant-available water stored in the soil) on the existing cropland areas in seasonal floodplain wetlands, where small-scale farmers grow a single crop of rainfed lowland rice during the wet season. However, soil moisture at the onset of the rains (pre-rice niche) and residual soil moisture after rice harvest (post-rice niche) may suffice to cultivate short-cycled crops. We developed a methodological approach to analyze the potential for green water cultivation in the pre- and post-rice niches in the Kilombero Floodplain in Tanzania, as a representative case for seasonal floodplain wetlands in SSA. The three-step approach used open-access remote sensing datasets to: (i) extract cropland areas following a cross-comparison of multiple land cover products; (ii) analyze soil moisture conditions using evaporative stress indices to identify the pre- and post-rice niches (using MOD16A2GF potential (PET) and actual (AET) evapotranspiration products), and (iii) quantify the green water availability in the identified niches (using an ensemble mean of SSEBop and WaPOR to calculate AET).

Results showed that the WaPOR land cover product reliably identified cropland areas in Kilombero, followed by CGLS-LC, while ESA-CCI largely miss-captured the cropland extent and MCD12Q1 did not capture almost any cropland areas. Estimates of the AET ensemble mean product of 2.6 mm/day were comparable with previously reported values in Kilombero cropland (2.05–2.74 mm/day) and were correlated with NDVI (MOD13Q1) on the monthly basis (R² = 0.58; p <0.05), demonstrating the good performance of the AET ensemble mean product. We further identified distinct patterns of green water being available both before and after the rice-growing period. Based on the analyses of evaporative stress indices, the pre-rice niche tended to be longer (~70 days with AET of 20–40 mm/10-day) but also more variable (inter-annual variability >30%) than the post-rice niche (~65 days with AET of 10–30 mm/10-day, inter-annual variability <15%). These findings confirm the large potential for cultivating short-cycled crops beyond the rice-growing period on at least 53% of the total cropland area. A wider application of the developed approach in this study can help identifying opportunities and guiding interventions towards establishing cropping intensification and diversification practices in floodplain wetlands in SSA. The uncertainties, limitations, and implications of the proposed approach are discussed.