Estimating changes and trends in ecosystem extent with dense time-series satellite remote sensing

Area of Science:

• Ecology and Remote Sensing
• Environmental Monitoring
• Geospatial Analysis

Background:

• Quantifying ecosystem extent is crucial for conservation and monitoring deforestation.
• Satellite remote sensing provides valuable data, but static snapshot analyses limit trend accuracy and uncertainty estimation.
• Existing methods often require costly, time-intensive reference data.

Purpose of the Study:

• To develop and apply a novel method for estimating ecosystem extent trends using dense time-series satellite imagery.
• To improve the accuracy and uncertainty assessment of ecosystem loss estimates.
• To analyze deforestation trends in the Hukaung Valley Wildlife Sanctuary, Myanmar.

Main Methods:

• Utilized all available Landsat satellite imagery, creating dense time-series classified maps.
• Incorporated covariates like time, cloud cover, and seasonality into trend estimation.
• Applied a generalized additive mixed model to Landsat image classifications (1999-2018).

Main Results:

• Significant forest extent decline of 0.274%/year (SE = 0.078) was detected in Hukaung Valley.
• Forest cover decreased from 91.70% (SE = 0.02) in 1999 to 86.52% (SE = 0.02) in 2018.
• The new method improved trend estimation, enabling significance testing and confidence intervals.

Conclusions:

• The dense time-series approach enhances the reliability of ecosystem loss trend analysis.
• This method reduces the need for extensive temporal reference data and quantifies uncertainty.
• Accurate trend quantification is vital for effective conservation and environmental management strategies.