A comparison of weather data in a mechanistic model of organism body temperature

Background/Question/Methods

Mechanistic models of organism body temperature are an important tool for predicting the effect of global climate change on species biogeographic ranges. These models typically need to be supplied with weather data including solar radiation, air temperature, wind speed, air pressure, relative humidity, and precipitation rates. In situ weather stations provide the best weather data for estimates of organism body temperature. However, weather stations are only accurate for a small geographic area, and it is time-consuming and expensive to deploy the number of weather stations that would be necessary to cover biogeographic ranges of most species. Remotely sensed data and reanalyzed data, which cover large geographic areas, are an alternative source of weather data for mechanistic models of organism body temperature. Different permutations of station, reanalysis, and satellite weather data were supplied to a mechanistic model of intertidal mussel body temperature. Model estimates were compared to seven years of Mytlius californanus biomimetic logger data from the West Coast of North America using forecast verification methods. Results/Conclusions Models supplied with all types of input weather data tended to overhindcast the number of high temperature days that M. californianus along the West Coast of North America actually experienced. As expected, the weather station data provided the best estimates of daily maximum body temperatures. However, the combination of North American Regional Reanalysis data (an 8x daily 32 km grid scale product) with surface downward shortwave flux from the GOES satellites also provided good estimates of Mytilus californianus daily maximum body temperatures. This suggests that reanalysis and remote sensing data are promising sources of weather data for mechanistic models.