Developing an Original Pedotransfer Function Framework for Predicting Soil Hydraulic Properties in Nigerian Agroecological Zones

This dissertation develops and validates an original Nigerian pedotransfer function (PTF) framework for predicting soil hydraulic properties from readily measured soil data across all six Nigerian agroecological zones. Soil hydraulic properties including water retention at field capacity and permanent wilting point, saturated hydraulic conductivity, and unsaturated conductivity functions are fundamental inputs for hydrological modelling, irrigation scheduling, and crop growth simulation, yet their direct measurement is expensive and rarely performed for Nigerian soils. Existing international PTFs developed in temperate contexts show poor performance when applied to tropical African soils due to differences in clay mineralogy, organic matter quality, and soil structure. Using a systematic dataset development approach, this study characterised 280 soil profiles across the Sahel, Sudan savanna, Guinea savanna, Derived savanna, Moist forest, and Mangrove zones of Nigeria, measuring full soil hydraulic characterisation including van Genuchten parameters alongside easily measured predictor variables (texture, organic carbon, bulk density, cation exchange capacity, and iron oxide content). Multiple regression, artificial neural network, and support vector machine approaches were compared for PTF development. An ensemble PTF combining all three methods was validated against an independent test dataset of 80 profiles. The dissertation proposes the Nigerian Agroecological Pedotransfer Function (NAPTF) system as an original contribution, reducing prediction error for field capacity by 68 percent compared to international PTF application. The NAPTF is implemented as a freely accessible web-based tool for Nigerian soil scientists and agronomists. Original contributions include the PTF methodology, the NAPTF system, and the 280-profile Nigerian soil hydraulics dataset. Keywords: pedotransfer functions, soil hydraulic properties, Nigerian agroecological zones, machine learning, soil physics.