Options
Model tree technique for streamflow forecasting: A case study in sub-catchment of Tapi River Basin, India
Journal
Advances in Streamflow Forecasting: From Traditional to Modern Approaches
Date Issued
2021-01-01
Author(s)
Sharma, Priyank J.
Patel, P. L.
Jothiprakash, V.
Abstract
Efficient streamflow forecasts are vital for water resources management in a river basin, especially for flood management. The process-based models are usually preferred for streamflow modeling, but they warrant intense data and computational requirements. The data-driven models (DDMs) rely upon computational intelligence and machine learning algorithms by assuming the presence of considerable amount of data. The DDMs attempt to establish a relationship between the input and output variables governing a physical process without explicit knowledge about the physical behavior of the system. Model tree (MT) is one such DDM, which is based on the principle of information theory. The MT involves splitting the multidimensional parameter space into subspaces and generating linear models for each of them according to the overall quality criterion. Thus, MT adopts piecewise linearization of nonlinear hydrologic processes to produce easily interpretable outputs. The application of MT in streamflow forecasting is demonstrated through a case study of the Purna River in Tapi Basin, India.
Subjects