Detection of subtle deforestation due to logging using satellite remote sensing in wet and dry savanna woodlands of Southern Africa uri icon

abstract

  • Tropical deforestation through logging activities poses a direct threat to biodiversity. However, the detection of logging has remained a challenge. Based on study sites in Zimbabwe and Zambia, we tested whether the Normalized Difference Vegetation Index (NDVI) and the Coefficient of Variation in NDVI (CVNDVI) derived from high and medium spatial resolution satellite data could be used to detect logging in dry and wet miombo woodlands. Separately, we integrated NDVI and CVNDVI in logistic regression to test whether each can be used to successfully predict logging in the study sites. We tested whether the spatial resolution of satellite data has an effect in detection of logging using NDVI and CVNDVI derived from Landsat 8 and Worldview-2. Based on the ROC curves, we concluded that remotely sensed data could provide an effective predictive tool for detecting logging. However, in wet miombo woodlands the predictive power of remotely sensed data is weak
  • Tropical deforestation through logging activities poses a direct threat to biodiversity. However, the detection of logging has remained a challenge. Based on study sites in Zimbabwe and Zambia, we tested whether the Normalized Difference Vegetation Index (NDVI) and the Coefficient of Variation in NDVI (CVNDVI) derived from high and medium spatial resolution satellite data could be used to detect logging in dry and wet miombo woodlands. Separately, we integrated NDVI and CVNDVI in logistic regression to test whether each can be used to successfully predict logging in the study sites. We tested whether the spatial resolution of satellite data has an effect in detection of logging using NDVI and CVNDVI derived from Landsat 8 and Worldview-2. Based on the ROC curves, we concluded that remotely sensed data could provide an effective predictive tool for detecting logging. However, in wet miombo woodlands the predictive power of remotely sensed data is weak.

publication date

  • 2017
  • 2017
  • 2017