Estimation for sparse vegetation information in desertification region based on Tiangong-1 hyperspectral image

Area of Science:

• Remote Sensing
• Vegetation Monitoring
• Desertification Studies

Background:

• Accurate estimation of sparse vegetation is crucial for understanding desertification dynamics.
• Hyperspectral imagery offers detailed spectral information for vegetation analysis.
• Traditional vegetation indices like NDVI may have limitations in arid and semi-arid environments.

Purpose of the Study:

• To compare the effectiveness of NDVI and SAVI in retrieving sparse vegetation information.
• To identify optimal spectral bands for vegetation parameter estimation using Tiangong-1 data.
• To develop and validate models for estimating vegetation coverage and biomass in a desertification region.

Main Methods:

• Utilized Tiangong-1 hyperspectral imagery from Sunite Right Banner, Inner Mongolia.
• Calculated NDVI and SAVI, correlating them with field-observed vegetation coverage and biomass.
• Determined optimal band combinations for vegetation indices and established linear regression models.

Main Results:

• SAVI demonstrated a higher correlation coefficient (approx. 0.8) with vegetation parameters than NDVI (approx. 0.7).
• Optimal band combinations were identified for both NDVI (630nm red, 910nm NIR) and SAVI (620nm red, 920nm NIR).
• SAVI-based regression models yielded higher R-squared values (up to 0.59 for coverage) and lower RMSE than NDVI-based models.

Conclusions:

• Tiangong-1 hyperspectral data effectively captures sparse vegetation conditions.
• SAVI is a more accurate index than NDVI for estimating vegetation coverage and biomass in desertification areas.
• The study provides a validated approach for remote sensing-based vegetation monitoring in arid environments.