Short-term streamflow forecasting with global climate change implications - A comparative study between genetic programming and neural network models

    Authors

    A. Makkeasorn; N. B. Chang;X. Zhou

    Comments

    Authors: contact us about adding a copy of your work at STARS@ucf.edu

    Abbreviated Journal Title

    J. Hydrol.

    Keywords

    streamflow forecasting; genetic programming; neural network; global; climate change; NEXRAD; sea surface temperature; SYSTEME-HYDROLOGIQUE-EUROPEEN; INSTANTANEOUS UNIT-HYDROGRAPH; IDENTIFIABLE COMPONENT FLOWS; SMALL UPLAND CATCHMENTS; RAINFALL; ENSO; RUNOFF; RIVER; PREDICTION; STATES; Engineering, Civil; Geosciences, Multidisciplinary; Water Resources

    Abstract

    Sustainable water resources management is a critically important priority across the globe. While water scarcity limits the uses of water in many ways, floods may also result in property damages and the loss of life. To more efficiently use the limited amount of water under the changing world or to resourcefully provide adequate time for flood warning, the issues have led us to seek advanced techniques for improving streamflow forecasting on a short-term basis. This study emphasizes the inclusion of sea surface temperature (SST) in addition to the spatio-temporal rainfall distribution via the Next Generation Radar (NEXRAD), meteorological data via local weather stations, and historical stream data via USGS gage stations to collectively forecast discharges in a semi-arid watershed in south Texas. Two types of artificial intelligence models, including genetic programming (GP) and neural network (NN) models, were employed comparatively. Four numerical evaluators were used to evaluate the validity of a suite of forecasting models. Research findings indicate that GP-derived streamflow forecasting models were generally favored in the assessment in which both SST and meteorological data significantly improve the accuracy of forecasting. Among several scenarios, NEXRAD rainfall data were proven its most effectiveness for a 3-day forecast, and SST Gulf-to-Atlantic index shows larger impacts than the SST Gulf-to-Pacific index on the streamflow forecasts. The most forward looking GP-derived models can even perform a 30-day streamflow forecast ahead of time with an r-square of 0.84 and RMS error 5.4 in our study. (C) 2008 Elsevier B.V. All rights reserved.

    Journal Title

    Journal of Hydrology

    Volume

    352

    Issue/Number

    3-4

    Publication Date

    1-1-2008

    Document Type

    Article

    Language

    English

    First Page

    336

    Last Page

    354

    WOS Identifier

    WOS:000255293600008

    ISSN

    0022-1694

    Share

    COinS