AbstractsEarth & Environmental Science

Temperature variability is an important driver of many important global and regional processes, which has inspired researchers to understand and predict the spatial variability of surface air temperature. This importance has increased demand for quality, high resolution gridded climatological datasets that deliver detailed information on the variability of temperature at regional scales. Several interpolation and extrapolation techniques have been introduced that use point data sources (land-based data from weather stations). However, the scarcity of weather stations with long-term records and good spatial coverage and the impacts of a non-stationary climate limits these traditional methods. Through the analysis of existing ground-based temperature sensors we have shown that there are inadequate ground-based measurements to estimate the spatial variability of daily min/max temperature. Furthermore, we have shown that existing interpolation methods are insufficiently accurate to estimate the local temperature at ungauged locations because they cannot capture anthropogenic (e.g. urban heat island) or microclimatological (e.g. cold air pooling) effects. This result implies that, in general, ground-based temperature measurements are too sparse to capture the spatial variability of temperature. Together with satellite observations, gridded meteorological variables can provide important information of the complex interactions of these features in order to accurately map temperature across broad regions. Satellite remote-sensing is another way for acquisition of land surface temperature (LST) data. However, due to technical constraints, satellite thermal sensors are incapable to supply both spatially and temporally dense LST image data. The reason for this is that the spatial and temporal resolutions of a satellite thermal sensor are anti-correlated, meaning that a high spatial resolution is related with low temporal resolution and vice versa. The trade-off between spatial and temporal resolution of satellite data, encouraged us to apply the Moderate Resolution Imaging Spectroradiometer (MODIS) as a source of remote-sensed land surface temperature data to capture many of the rapid biological and meteorological changes that MODIS (Spatial Resolution [bands 20-23]: 1km, 5km) observes in every 1 to 2 days. This work develops a new method for integrating remote sensed and ground-based observations of temperature to account for anthropogenic and microclimatological impacts on the surface air temperature. This method is based on a mathematical function that relates the temperature at each point in space as a summation of the remote-sensed measurement at that location and a spatially dependent bias term, which is calculated using the ground based measurements. This model combines the spatial patterns captured within the remote-sensed measurements with the high accuracy of the land-based embedded sensors to construct continuous maps of daily min/max temperature over broad regions. Thus, this model is able to capture the underlying…