Relating site-specific weather data to regional monitoring networks in the Lake States

Abstract

Meteorological variables most frequently used in ecological studies include, or are derived from, daily precipitation and air temperature. In many studies, weather data from the nearest permanent weather station are used as surrogates for on-site measurements. This study discusses the problems with this approach and illustrates methods for developing regression equations for calculating site-specific daily minimum, average, and maximum temperatures, and precipitation amount from regional monitoring information. Meteorological data were collected at four sites along a 650 km climatic gradient from Houghton County (47°N, 89°W) to Oceana County (43°N, 86°W), Michigan. Data from several National Oceanic and Atmospheric Administration (NOAA) stations within 40 km of each study site were related to measurements made at each site. Predictive ability was improved by using information from more than one NOAA station to predict on-site temperature. No increase in predictive ability resulted from including information from more than a single station when predicting precipitation. For the majority of the sites and climate variables, the best relationships were not obtained by using the nearest NOAA station; variables such as the relative distance from large bodies of water or elevation appeared to be influential. Since distance between two locations is rarely the most important factor governing the relationship between their climates, the use of the nearest weather station is not the best method to describe weather conditions on specific sites of interest.