Accurate representations of root zone soil moisture are essential for agro-pastoral drought monitoring, as well as water and energy balance studies for regional climate change and land-use change analysis. Unfortunately, soil moisture observations are generally sparse and point measurements do not necessarily represent the spatial heterogeneity of soil properties. The goal of this research is to investigate the relationships between plant phenology and soil moisture using remotely sensed vegetation greenness and in-situ soil moisture, heat flux and rainfall measurements. A thorough understanding of these relationships can help us make predictions about hard-to-measure variables (e.g. soil moisture, evapotranspiration) from satellite derived vegetation (i.e. the Normalized Difference Vegetation Index (NDVI)) and rainfall observations. Specifically we ask, what are the relationships between NDVI, soil moisture, evaporation, and rainfall in south west Niger? And how do these relationships differ depending on vegetation type and inter-annual rainfall variability? We focus on millet and fallow sites in the Wankama catchment where in-situ measurements are available during 2005 and 2006 from the Africa Monsoon Multidisciplinary Assessment (AMMA) database.
Results/Conclusions
Results show that vegetation cover and inter-annual variability impact how rainfall is partitioned into soil moisture and AET. Consistent with other literature we find that NDVI tends to be best correlated with the total amount of rainfall accumulated over the previous 30 days (R2 =0.72-0.91). We also found that near surface soil moisture (10-50cm) tends to be best correlated with rainfall totals over the previous ~40 days (R2 =0.75-0.85) with noticeable difference between vegetation types. The millet site, with less vegetation cover and more bare soil has lower maximum actual evapotranspiration (AETmax ) and NDVImax. It is not surprising then that the millet site tends to have higher moisture content in the root zone (10-150cm) and for a longer duration compared to the fallow site. These data are the first steps in building a model for the prediction of root zone soil moisture with NDVI. Moreover, these results confirm that that accurate representation of vegetation (e.g. rooting depth, percent coverage) is essential for determining the partitioning of precipitation into soil moisture storage and actual evapotranspiration. These measurements will be used to parameterize water balance models that are used for agricultural drought monitoring in the Sahel.