Spatial Variability of Sea Surface Temperature Effect in Utah and Long-term Streamflow Forecasting Using Relevance Vector Machine

Presenter Information

Niroj Shrestha

Location

Eccles Conference Center

Event Website

http://water.usu.edu/

Start Date

3-30-2011 11:40 AM

End Date

3-30-2011 12:00 PM

Description

Long-term prediction of streamflow plays an important role on planning and decision making in the river basin scale. This information provides how much water will be available in the next season so that the farmers can plan accordingly how much land to irrigate, and how much livestock to purchase. Financial commitment made early in the season can result in substantial economic losses if the resulting seasonal flow do not subsequently supply enough irrigation water. So, predicting the future water availability is a key step to successful water resource management in arid regions: The data driven model derived from statistical learning theory was made a choice. It relates input/output without trying to understand the underlying physical process. They are characterized by their ability to quickly capture the underlying physics and provide predictions of system behavior using historical data. The model is developed in the form of Multivariate Relevance Vector Machine (MVRVM). Prediction is made for both monthly mean discharge and volume of water passing the streamflow site for next six months based on past streamflow data, snowpack in the mountain and local and regional meteorological conditions. Sea surface temperature (SST) is chosen to represent regional meteorological condition. The work then consists of identifying the best locations of sea surface temperature for the given location' of streamflow sites in the state of Utah. It is noticed that influence of Pacific Ocean is dominant in Utah than that of Atlantic Ocean. For the prediction of monthly mean discharge, sea surface temperature (SST) of North Pacific usually developed the best model for Northern and Central Utah while SST of Tropical Pacific developed the best model in Southern Utah. For the volumetric prediction, the sea surface temperature of North Pacific developed the best model prediction in most of the streamflow sites in Utah.

This document is currently not available here.

Share

COinS
 
Mar 30th, 11:40 AM Mar 30th, 12:00 PM

Spatial Variability of Sea Surface Temperature Effect in Utah and Long-term Streamflow Forecasting Using Relevance Vector Machine

Eccles Conference Center

Long-term prediction of streamflow plays an important role on planning and decision making in the river basin scale. This information provides how much water will be available in the next season so that the farmers can plan accordingly how much land to irrigate, and how much livestock to purchase. Financial commitment made early in the season can result in substantial economic losses if the resulting seasonal flow do not subsequently supply enough irrigation water. So, predicting the future water availability is a key step to successful water resource management in arid regions: The data driven model derived from statistical learning theory was made a choice. It relates input/output without trying to understand the underlying physical process. They are characterized by their ability to quickly capture the underlying physics and provide predictions of system behavior using historical data. The model is developed in the form of Multivariate Relevance Vector Machine (MVRVM). Prediction is made for both monthly mean discharge and volume of water passing the streamflow site for next six months based on past streamflow data, snowpack in the mountain and local and regional meteorological conditions. Sea surface temperature (SST) is chosen to represent regional meteorological condition. The work then consists of identifying the best locations of sea surface temperature for the given location' of streamflow sites in the state of Utah. It is noticed that influence of Pacific Ocean is dominant in Utah than that of Atlantic Ocean. For the prediction of monthly mean discharge, sea surface temperature (SST) of North Pacific usually developed the best model for Northern and Central Utah while SST of Tropical Pacific developed the best model in Southern Utah. For the volumetric prediction, the sea surface temperature of North Pacific developed the best model prediction in most of the streamflow sites in Utah.

https://digitalcommons.usu.edu/runoff/2011/AllAbstracts/14