Developing A Transboundary Aquifer Analytic Framework To Support Localised Groundwater Decision-Making Using Big Data

Big data analytics (BDA) is a modern and innovative platform of applications that include advanced analytical techniques such as data mining, statistical analysis, artificial intelligence, machine learning, and natural language processing. Regional data are generated through groundwater monitoring, remote sensing applications or global circulation models (GCM), however this is often too course for a local understanding. Groundwater managers rely on locally relevant information for effective operational decision making, however this is often missing. A Transboundary Aquifer (TBA) Analytic Framework was developed to match, integrate and model local hydrogeological data with regional earth-observation data using BDA. Drawing on the literature on BDA, a reference architecture for the TBA analytical framework was identified for application to various groundwater management scenarios in the Ramotswa Dolomitic Aquifer (Botswana - South Africa) and Shire Valley Alluvial Aquifer (Malawi - Mozambique). The TBA analytical framework allows for local clouds to store the local and regional structured and unstructured datasets and interconnecting these local clouds through a federated cloud infrastructure. In this regard, tools that are incorporated in the TBA analytical framework include data ingestion operators, data transformation operators, and feature extractors. Various machine learning algorithms and statistical techniques are incorporated in the TBA analytical framework to downscale the regional datasets. The downscaling involves selection of potential predictors and predictants variables based on data needs to address local groundwater management scenarios such as regulating groundwater abstraction to prevent groundwater depletion. Using the downscaled data the TBA analytical framework can be utilised to uncover patterns and statistical relationships in the datasets in order to model local groundwater processes such as cone of depression, groundwater levels forecasting, well protection zoning, amongst others.