Home / Science & Engineering / Hawaii may not be accurately predicting groundwater contamination

Hawaii may not be accurately predicting groundwater contamination

Map of pollution
Map of elevated nitrate detection probability in the Pearl Harbor-Honolulu aquifer developed using a multiple variable logistic regression model.

A recent audit of Hawai‘i’s Source Water Assessment Program (SWAP) drinking water source susceptibility analysis shows that the state’s current approach may not be accurately predicting groundwater contamination at local drinking water sources.

The research was performed by Alan Mair, a former postdoctoral fellow at UH Mānoa’s Department of Geology & Geophysics and Water Resources Research Center (WRRC), and Aly El-Kadi, a professor of Geology & Geophysics and an associate director of the WRRC.  The analysis is published in the WRRC Bulletin – July 2013 issue

First, researchers mapped detections of 11 target contaminants in the Pearl Harbor, Honolulu, and Lahaina aquifers over the last 12 years, including elevated nitrate, chlorinated solvents, agricultural fumigants, and pesticides.

Much of the contamination they saw was related to agricultural and military land use.  For example, elevated nitrate concentrations mainly corresponded with areas of active and former pineapple and sugarcane cultivation.  However, other potential sources of this pollutant include on-site sewage disposal systems.

Next, researchers examined the performance of existing risk models to see how well they predicted contamination.

“In the Pearl Harbor-Honolulu aquifer sector, the high detection frequencies of target contaminants combined with low PCA (potentially contaminating activities) scores indicate that the SWAP (source water assessment program) risk model does not adequately identify CZDs (capture zone delineations) with the highest actual groundwater vulnerability (i.e., actual detections of target contaminants),” wrote Mair and El-Kadi.

Similar performance issues were also noted for the PCA risk scores assigned to the Lahaina aquifer sector.

To address this gap, the WRRC researchers are proposing an additional, alternative risk assessment approach using a set of variable logistic regression models developed to take into account chemical monitoring data as well as explanatory information such as hydrogeology, land use, and drinking water well geometry and location.

“The significant improvements in predictive capability using a statistical modeling approach confirm that changes can be made to the existing SWAP risk model to make it more reflective of observed target contaminant detections,” the researchers wrote.  “The objective logistic regression modeling approach described in this study represents a suitable addition to the current subjective SWAP risk model.”


Read the full article:  “Source water assessment program in Hawai‘i:  Audit of the susceptibility analysis for groundwater sources,” by Alan Mair and Aly El-Kadi, in the Water Resources Research Center July 2013 Bulletin.

Contact:  Philip Moravcik, moravcik@hawaii.edu, (808) 956-3097