Year of Award

2013

Document Type

Thesis - Campus Access Only

Degree Type

Master of Science (MS)

Degree Name

Geography

Department or School/College

Department of Geography

Committee Chair

Anna E. Klene

Commitee Members

Brian Steele, Christiane von Reichert

Keywords

GIS, Sleeping Child Creek Watershed, Stream Temperature Model

Abstract

Prior to this research no stream temperature model existed for the Sleeping Child Creek Watershed, Montana, nor for any other watershed within the Bitterroot Subbasin, let alone ones which could be used operationally by non-specialists completely within a single GIS software package. Sleeping Child Creek Watershed is the site of the only seasonal PACFISH/INFISH Biological Opinion water-temperature monitoring location within the Subbasin and had an established population of native trout species known to be sensitive to stream temperatures. These factors combined to make SCCW an ideal location for this study to evaluate the potential for such an operational workflow. Fourteen water temperature sampling sites were monitored in the summer (July, August, and September) of 2010. Hydrologic and environmental variables were collected within a Geographic Information System (GIS) for the purpose of comparing two methods for predicting monthly stream-temperature distributions within a watershed using ArcGIS software. Each variable was assessed for its relationship to the observed stream temperatures. Both an ordinary least squares (OLS) and an exploratory regression module were analyzed within the GIS to evaluate the predictive strength of seven variables on stream temperature. Results of the OLS analysis indicated strong R² values of 0.87, 0.91, and 0.93 for July, August, and September, respectively. The variables elevation and slope passed all criteria for a “properly specified” regression model using ArcGIS©’s exploratory method and yielded R² values of 0.82, 0.86, and 0.89 for the study months. Assessment of the predicted stream temperatures revealed reasonable accuracy between the observed mean-monthly values and those estimated by both approaches. Cross-validation using the full OLS method fell within a range ±1.10°C for RMSE, MAE, and MAB for all three months. Comparison to an independent monitoring site showed that both approaches mean-monthly water temperatures predictions were within ±1.0°C. Both models showed the tendency to under-predict stream temperatures within low-order high-elevation tributaries to Sleeping Child Creek, outside of acceptable stream temperature ranges. Estimated higher-order stream temperatures fell just barely within the warmest acceptable thermal ranges for the local salmonoid species. Both of the techniques analyzed in this research were suitable for operational use to routinely predict stream temperature variability within in a small mountainous watershed using limited water temperatures and GIS-derived datasets for the input variables. From a statistical standpoint, the OLS technique was more accurate (as would be expected) and the same methodology could be implemented in any watershed. However, the exploratory approach may be suitable when initially trying to determine which variables to use from a long list of potential explanatory variables or when working in a new location.

This record is only available
to users affiliated with
the University of Montana.

Request Access

Share

COinS
 

© Copyright 2013 Gabriel J. Frazier