Related Experiment Videos
Downscaling Land Surface Temperature in Complex Regions by Using Multiple Scale Factors with Adaptive Thresholds.
Yingbao Yang1, Xiaolong Li2, Xin Pan3
1School of Earth Science and Engineering, Hohai University, 8 Buddha City West Road, Nanjing 210098, China. yyb@hhu.edu.cn.
Sensors (Basel, Switzerland)
|April 4, 2017
Summary
This study presents an improved method for downscaling land surface temperature (LST) in urban areas. The approach adaptively selects scale factors for more accurate LST mapping, proving effective across seasons.
Area of Science:
- Earth Observation
- Remote Sensing
- Urban Climatology
Background:
- Coarse-resolution land surface temperature (LST) from satellites poses challenges for urban studies.
- Existing LST downscaling methods often lack focus on complex urban landscapes with mixed surface types.
Purpose of the Study:
- To develop and evaluate an adaptive LST downscaling approach for urban areas with mixed surface types.
- To improve the accuracy and reliability of LST retrieval in complex urban environments.
Main Methods:
- Utilized a multiple linear regression model linking LST with multiple scale factors.
- Employed correlation coefficients (CCs) and CC thresholds for adaptive selection of scale factors within a moving window.
- Validated the approach using Landsat 8 thermal imagery of Nanjing City across different seasons.
Main Results:
- Achieved satisfactory downscaling results with R² of 0.87 and RMSE of 1.13 °C on August 11th.
- Demonstrated consistent accuracy and seasonal availability comparable to other methods.
- Showcased best performance in vegetated areas and acceptable results in water bodies.
Conclusions:
- The proposed adaptive LST downscaling method is efficient and reliable for urban areas.
- The approach offers consistent performance across various seasons and surface types.
- Future work includes refining downscaling in challenging regions and applying the model to lower spatial resolutions.