Detecting Forest Changes Using Dense Landsat 8 and Sentinel-1 Time Series Data in Tropical Seasonal Forests

摘要: The accurate and timely detection of forest disturbances can provide valuable information for effective management. Combining dense time series observations from optical synthetic aperture radar satellites has the potential to improve large-area monitoring. For various disturbances, machine learning algorithms might accurately characterize changes. However, there is limited knowledge especially on use detect through hybrid approaches that combine different data sources. This study investigated Landsat 8 Sentinel-1 detecting in tropical seasonal forests based a algorithm. random algorithm was used predict disturbance probability each observation using variables derived harmonic regression model, which characterized seasonality disturbance-related probabilities both sensors were then combined pixel. results showed combination achieved an overall accuracy 83.6% detection, higher than only (78.3%) or (75.5%). Additionally, more by combining Sentinel-1. Small-scale caused logging led large omissions disturbances; however, other detected with relatively high accuracy. Although had low this study, improved indicating value detection.