Ensemble-characterisation of satellite rainfall uncertainty and its impacts on the hydrological modelling of a sparsely gauged basin in Western Africa

Many areas of the planet lack the infrastructure required to make accurate and timely estimations of rainfall. This problem is especially acute in sub-Saharan Africa, where a paucity of rain recording radar and sufficiently dense raingauge networks combine with highly variable rainfall, a reliance on agriculture that is predominantly rain fed and systems that are prone to flooding and drought. Satellite Rainfall Estimates (SRFE) are useful as they can provide additional spatial and temporal information to drive various downstream environmental models and early warning systems (EWS). However, when operating at higher spatial and temporal resolutions SRFE contain large uncertainties which propagate through the downstream models.This thesis uses the TAMSAT1 SRFE algorithm developed by Teo (2006) to estimate the rainfall over a large, data sparse and heterogenous catchment in the Senegal Basin. The uncertainty within the TAMSAT1 SRFE is represented using a set of ensemble estimates, each unique but equiprobable based on the full conditional distribution of the recorded rainfall, produced using the TAMSIM algorithm, also developed by Teo (2006). The ensemble rainfall estimates were then used in turn to drive a Pitman Rainfall-Runoff model of the catchment hydrology.The use of ensemble rainfall estimates was shown to be incompatible with the pre-calibrated parameter values for the hydrological model. A novel approach, the EnsAll method, was developed to calibrate the hydrological model which incorporated each individual ensemble member. The EnsAll calibrated model showed the greatest skill when driven by the ensemble rainfall estimates and little bias. A comparison of the hydrographs produced from TAMSIM ensemble rainfall estimates and that from an ensemble of perturbed TAMSAT1 estimates showed that the full spatio-temporally distributed method used by TAMSIM is superior to a simpler perturbation method for characterizing SRFE uncertainty.Overall, the SRFE used were shown to outperform the rainfall estimates produced from the sparse raingauge network as a hydrological model driver. However, they did demonstrate a lack of ability to represent the large interseasonal variations in rainfall resulting in large systematic biases. These biases were observed propagating directly to the modelled hydrological ouput.

Name dc:type.qualificationname

PhD

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Hull

Year dc:date.issued

2013

Author dc:creator

• Skinner, Christopher J.

Advisor dc:contributor.advisor

• Bellerby, T. J. (Timothy James)

• Geography

Language dc:language

en

Identifier

oai:hull-repository.worktribe.com:4214727

OAI identifier oai:identifier

oai:hull-repository.worktribe.com:4214727