A Geographic Information-Assisted Temporal Mixture Analysis for Addressing the Issue of Endmember Class and Endmember...

Area of Science:

• Remote Sensing
• Geographic Information Systems
• Urban Environmental Analysis

Background:

• Spectral Mixture Analysis (SMA) is crucial for urban biophysical fraction parameterization.
• Endmember class and spectra selection are critical but challenging steps in SMA.
• Spatial heterogeneity in urban landscapes introduces significant errors in traditional SMA.

Purpose of the Study:

• To address the errors caused by endmember variability in SMA.
• To propose a novel method, Geographic Information-Assisted Temporal Mixture Analysis (GATMA), for improved urban land cover analysis.
• To enhance the accuracy of fractional land cover estimation in heterogeneous urban environments.

Main Methods:

• Logistic regression analysis to link land use/land cover with socio-economic factors.
• Classification tree method for identifying endmember class status.
• Ordinary kriging for generating spatially varying endmember spectra.
• Fully constrained temporal mixture analysis for fractional land cover examination.

Main Results:

• GATMA achieved high accuracy with RMSE of 6.81%, SE of 1.29%, and MAE of 2.6%.
• Significant accuracy improvements were observed across the entire study area.
• The method demonstrated effectiveness in both developed and less developed urban zones.

Conclusions:

• The variability of endmember class and spectra is essential for accurate unmixing analysis.
• GATMA effectively accounts for endmember variability, leading to superior results compared to traditional SMA.
• The proposed method offers a robust approach for detailed urban environmental monitoring and analysis.