Development and Uncertainty Analysis of an Empirical Recharge Prediction Model for Nevada’s Desert Basins

Brian J. Epstein, Greg M. Pohll, Justin Huntington, and Rosemary W.H. Carroll

ABSTRACT
Nevada is one of the driest and most urbanized state in the nation, where the mean annual precipitation is about 11 inches, and 88% of the state’s population lives in either the greater Las Vegas or Reno metropolitan areas. As a result, appraisals of available water resources for both in-basin uses and inter-basin transfers are primary objectives for many hydrology studies. Hydrologists have focused on estimating available water resources in Nevada since the 1940s. The challenge associated with measuring recharge directly has lead to the use of indirect methods with most basin water budgets dependent on empirically derived recharge prediction methods (e.g. Maxey-Eakin, Nichols). The bootstrap brute-force recharge model (BBRM) optimization algorithm is developed to recalibrate empirical models by incorporating measurement error in precipitation, recharge variability and regression model uncertainty. The 1998 PRISM precipitation map more accurately reproduces observed precipitation compared to the 1965 Hardman precipitation map and is used in the BBRM development. Results show that, on average, the Maxey-Eakin model predicts the lowest recharge volumes while the Nichols model predicts the largest recharge volumes. The BBRM predicts recharge volumes in between the Maxey-Eakin and Nichols approaches. The Maxey-Eakin produces the lowest errors for basins with low expected recharge, but the mean behavior of the BBRM is capable of explaining the highest percentage of recharge variability (adjusted r2=0.76). The resultant BBRM 95-percent confidence intervals show that uncertainty increases substantially at larger precipitation rates. Better quantification of basin discharge and precipitation will improve model prediction and reduce uncertainty in model behavior.

Keywords:
Citation:
DOI:

Water Quality Data Modeling To Minimize Eutrophication in the Truckee River, Nevada.

Anpalaki J. Ragavan

ABSTRACT
High total phosphorus concentrations (TP) are the major cause for eutrophication which leads to poor water quality in Truckee River, Nevada. Identifying the exact pattern of relationship among multiple independent variables related to high TP is important to implement remediation methods. In this study linear and non-linear relationship of multiple independent variables to TP in Truckee River were modeled. Independent variables included alkalinity(ALK), total soluble phosphorus concentrations(STP), stream flow(SF), water pH (pH), water temperature(TEMP), Dissolved oxygen(DO) and dissolved organic carbon(DOC) sampled monthly at the same time as TP (from January 1997 to December 2007) over six monitoring sites (McCarran Bridge(MB), Wardsworth Bridge(WB), Steamboat Creek(SC), Derby Dam(DD), Lockwood(LW), and North Truckee Drain(NTD)) along Truckee River, Nevada. Seasonal variations and man-made intervention in TP were included in the analysis. A non-linear multiple regression model built for the dependent variable (TP), as a function of the independent variables was optimized with respect to TP, subject to non-linear constraints and boundary values. Increase in SF, DO and TEMP resulted in decreased TP by non-linear relationships. Increases in STP, DOC and ALK increased TP linearly. Fitted non-linear regression model closely predicted TP in the river explaining 97% of total variation (R2 =0.908) in TP. Model forecasted TP agreed well with observed TP. All independent variables influenced TP significantly at 1% significance level (p<0.0001). All six sites contributed significantly towards overall TP at 5% significance level (p<0.0001). Site SC is currently the major contributor of high TP followed by NTD in Truckee River.

Keywords:
Citation:
DOI: