You are here:
A linear regression model for predicting PNW estuarine temperatures in a changing climate
Citation:
PAYNE, M. C., C. A. BROWN, D. REUSSER, H. LEE, II, AND M. R. FRAZIER. A linear regression model for predicting PNW estuarine temperatures in a changing climate. Presented at A linear regression model for predicting PNW estuarine temperatures in a changing climate, Seatle, WA, September 13 - 14, 2011.
Impact/Purpose:
Pacific Northwest coastal regions, estuaries, and associated ecosystems are vulnerable to the potential effects of climate change, especially to changes in nearshore water temperature.
Description:
Pacific Northwest coastal regions, estuaries, and associated ecosystems are vulnerable to the potential effects of climate change, especially to changes in nearshore water temperature. While predictive climate models simulate future air temperatures, no such projections exist for estuarine water temperatures. Therefore, forecasting water temperature is of critical importance to research concerning the ecological condition and response of nearshore habitats. Here we present a multiple linear regression model that is capable of reasonably forecasting estuarine water temperature using readily available data and that may be applicable to a range of coastal watersheds. Analysis of variance (ANOVA) and Akaike information criterion (AIC) model comparison statistics suggest that the most feasible model relies principally on sea-surface temperature (SST) and in situ air temperature. We use a nearshore, satellite-derived (AVHRR) SST product, along with weather station air temperature measurements, climatic and upwelling indices to build a localized model for Yaquina Bay Estuary, Newport, Oregon. In situ water temperature measurements are typically collected by moored buoys and tidal gauges. In order to mitigate the problems associated with the irregular (in space, time and quality) nature of those measurements, we combined NOAA tide gauge and Oregon State University Dock-moored YSICTD readings from 1991-present to develop and validate the model. In general, the model performs with a significant (p-value < 2.2e-16) R-squared value close to 0.5. Utilizing publically¬available model-input variables, such as the satellite SST product and weather station data should allow broad application of the model to other Pacific Northwest estuarine systems.