Hydraulic-hydrologic model for the Zambezi River using satellite data and artificial intelligence techniques
The development of hydraulic-hydrologic models is a challenge in the case of large catchment areas with scarce or erroneous measurement data and observations. With his study Mr. Dr. José Pedro Matos made several original contributions in order to overcome this challenge. The scientific developments were applied at Zambezi River basin in Africa in the framework of the interdisciplinary African Dams research project (ADAPT). First of all, procedures and selection criteria for satellite data regarding topography, rainfall, land use, soil types and cover had to be developed. With the goal to extend the time scope of the analysis, Dr. Matos introduced a novel Pattern-Oriented Memory (POM) historical rainfall interpolation methodology. When POM interpolated rainfall is applied to hydrologic models it effectively opens up new possibilities related to extended calibration and the simulation of historical events, which would otherwise be difficult to exploit. A new scheme for rainfall aggregation was proposed, based on hydraulic considerations and easily implemented resorting to remote sensing data, which was able to enhance forecasting results. Dr. Matos used machine-learning models in an innovative way for discharge forecast. He compared the alternative models (e.g. Autoregressive Moving-Average (ARMA), Artificial Neural Networks (ANN) and Support-Vector Regression (SVR)). Dr. Matos made then significant contributions to the enhancement of rainfall aggregation techniques and the study of limitations inherent to SVR forecasting model. He proposed also a novel method for developing empirical forecast probability distributions. Finally Dr. Matos could successfully calibrate, probably for the first time, a daily hydrological model covering the whole Zambezi River basin (ZRB).
Keywords: hydraulic-hydrologic models ; large catchment areas ; Zambezi River basin ; African Dams research project ; ADAPT ; Pattern-Oriented Memory ; POM historical rainfall interpolation ; interpolated rainfall ; machine-learning models
Record created on 2015-01-05, modified on 2016-08-09