thumbnail

Improving groundwater predictions utilizing seasonal precipitation forecasts from general circulation models forced with sea surface temperature forecasts

Journal of Hydrologic Engineering

By:
, , and
DOI: 10.1061/(ASCE)HE.1943-5584.0000776

Links

Abstract

Recent studies have found a significant association between climatic variability and basin hydroclimatology, particularly groundwater levels, over the southeast United States. The research reported in this paper evaluates the potential in developing 6-month-ahead groundwater-level forecasts based on the precipitation forecasts from ECHAM 4.5 General Circulation Model Forced with Sea Surface Temperature forecasts. Ten groundwater wells and nine streamgauges from the USGS Groundwater Climate Response Network and Hydro-Climatic Data Network were selected to represent groundwater and surface water flows, respectively, having minimal anthropogenic influences within the Flint River Basin in Georgia, United States. The writers employ two low-dimensional models [principle component regression (PCR) and canonical correlation analysis (CCA)] for predicting groundwater and streamflow at both seasonal and monthly timescales. Three modeling schemes are considered at the beginning of January to predict winter (January, February, and March) and spring (April, May, and June) streamflow and groundwater for the selected sites within the Flint River Basin. The first scheme (model 1) is a null model and is developed using PCR for every streamflow and groundwater site using previous 3-month observations (October, November, and December) available at that particular site as predictors. Modeling schemes 2 and 3 are developed using PCR and CCA, respectively, to evaluate the role of precipitation forecasts in improving monthly and seasonal groundwater predictions. Modeling scheme 3, which employs a CCA approach, is developed for each site by considering observed groundwater levels from nearby sites as predictands. The performance of these three schemes is evaluated using two metrics (correlation coefficient and relative RMS error) by developing groundwater-level forecasts based on leave-five-out cross-validation. Results from the research reported in this paper show that using precipitation forecasts in climate models improves the ability to predict the interannual variability of winter and spring streamflow and groundwater levels over the basin. However, significant conditional bias exists in all the three modeling schemes, which indicates the need to consider improved modeling schemes as well as the availability of longer time-series of observed hydroclimatic information over the basin.

Geospatial Extents

Additional Publication Details

Publication type:
Article
Publication Subtype:
Journal Article
Title:
Improving groundwater predictions utilizing seasonal precipitation forecasts from general circulation models forced with sea surface temperature forecasts
Series title:
Journal of Hydrologic Engineering
DOI:
10.1061/(ASCE)HE.1943-5584.0000776
Volume
19
Issue:
1
Year Published:
2014
Language:
English
Publisher:
American Society of Civil Engineers
Contributing office(s):
Office of the AD Water
Description:
12 p.
Larger Work Type:
Article
Larger Work Subtype:
Journal Article
Larger Work Title:
Journal of Hydrologic Engineering
First page:
87
Last page:
98
Country:
United States
State:
Georgia
Other Geospatial:
Flint River Basin