Groundwater Sodium Level Assessments in the Dadaab Subcounty in Northern Kenya

Abstract

Abstract— Groundwater sodium content was studied in the Dertu area following the 2008 complaints that pregnant mothers were having problems relating to hypertension, still-births and stunted growth for the babies born there, alongside weak bones for these children. The local leadership opined that the water source being used could play a role in these calamities. Sodium content in the Merti aquifer waters close to and far from the study area were analyzed with a view to helping make informed predictions regarding potential groundwater development in the area. To that effect, data table frames of historical data were assembled and analyzed using ML algorithms. The study reveals that the anomalous levels of Na in the Waters is purely geo-genic, derived from the geological interactions between the flowing groundwater and the parent rocks in the study area. Samples of groundwater wells were analyzed and, were also examined using WHO and KEBS standards for water quality, which indicate that groundwater in a selected sampling sites, were unsuitable for domestic purpose and irrigation. The study also reveals that the sodic mineralization varies from one place to another and as such, there would be some specific portion of the study area that one can pinpoint as being free of Na mineralization: there were as many places with anomalous sodic levels, as were those with acceptable Na+ levels. The prediction model so developed is deemed to be helpful in predicting the sodic content of an un-drilled proposed borehole site , given geospatial parameters and selected groundwater hydraulic variables. The model thus proves useful in making decisions under uncertainty, as far predicting sodium content is concerned. Keywords— Water quality, Groundwater, Chloride, Sodium, Neural Networks, Backpropagation.