Data-Based Mechanistic (DBM) modelling is a Transfer Function (TF) modelling approach, whereby the data defines the model. The DBM approach, unlike physics-based distributed and conceptual models that fit existing laws to data-series, uses the data to identify the model structure in an objective statistical manner. The approach is parsimonious, in that it requires few spatially-distributed data and is, therefore, suitable for data limited regions like West Africa. Multiple Input Single Output (MISO) rainfall to riverflow modelling approach is the utilization of multiple rainfall time-series as separate input in parallel into a model to simulate a single riverflow time-series in a large scale. The approach is capable of simulating the effects of each rain gauge on a lumped riverflow response.
Within this paper we present the application of DBM-MISO modelling approach to 20778 km2 humid tropical rain forest basin in Ghana. The approach makes use of the Bedford Ouse modelling technique to evaluate the non-linear behaviour of the catchment with the input of the model integrated in different ways including into new single-input time-series for subsequent Single Input Single Output (SISO) modelling. The identified MISO models were able to improve the efficiency and understanding of the rainfall-riverflow behaviour within the study catchment. The paper illustrates the potential benefits of the methodology in modelling large catchments with sparse network of rainfall stations.
