Estimating daily air temperature across the Southeastern United States using high-resolution satellite data: A...

Area of Science:

• Environmental Science
• Epidemiology
• Remote Sensing

Background:

• Accurate near-surface air temperature (Ta) is vital for environmental epidemiology but limited by sparse weather station coverage.
• Satellite surface temperature (Ts) offers potential for broad, local exposure estimates.
• The southeastern USA presents unique modeling challenges due to climate, water bodies, and humidity.

Purpose of the Study:

• To develop and validate a high-resolution (1km×1km) daily Ta estimation model using satellite Ts data for the southeastern USA.
• To assess the model's performance in a region with complex climatic and geographic features.
• To provide a reliable method for estimating spatio-temporal Ta for epidemiological studies.

Main Methods:

• Incorporated satellite surface temperature (Ts) to estimate daily near-surface air temperature (Ta).
• Calibrated Ts-Ta relationships using mixed linear models, land use data, and daily varying slopes.
• Employed linear regression and spatio-temporal smoothing for data gaps when satellite Ts were unavailable.

Main Results:

• Achieved a high cross-validated R² of 0.952, indicating excellent model performance.
• Demonstrated good agreement with NASA's MERRA model (R² = 0.969, RMSPE = 1.376°C).
• Successfully predicted Ta reliably across a large, diverse geographical area.

Conclusions:

• Satellite Ts data can be reliably used to predict high-resolution daily Ta.
• The developed model is effective even in regions with varied topography and weather patterns.
• This approach enhances the availability of crucial air temperature data for environmental health research.