Stream Flow Prediction by Remote Sensing and Genetic Programming Technologies

Successful water quantity and quality management depends on accurate monitoring and prediction of stream flow. Conventional approaches in stochastic modeling of stream flow do not always reflect the non-linear relationship between rainfall/runoff and resultant stream flow. Geographical, topographical, meteorological, and hydrological features of the river basin are not explicitly addressed in conventional stochastic methods. It is widely accepted that stream flow is a response to these influential features, but that their control over stream flow is irregular in both space and time, producing a highly non-linear relationship. Recent advances in remote sensing and genetic programming technologies have shown potential to improve the prediction accuracy of stream flow rate in a river system by better capturing the non-linearity of the system. An integrated database of physical basin features is developed and used to support a semi-structure modeling approach to perform stream flow predictions. This approach exhibits, elicits, and summarizes the non-linear behavior between the rainfall/run-off patterns and the resultant stream flow rate in river system given the influence of dynamic basin features such as soil moisture, soil texture, vegetative cover, air temperature, and precipitation rate. The model will be applied and validated in the Lower Leon Creek in the San Antonia River Basin, Texas; results will provide immediate and valuable input to the Texas Water Development Board and the Texas Natural Resources Commission, both of which are concerned with water resources management in this state. Ultimately, the model developed here will constitute an adaptable non-linear functional form for performing essential stream flow forecasts in any basin. This will be a valuable tool for all federal, state, local, tribal, and even commercial groups concerned with providing a foundation for achieving regional-scale sustainability of water resources.