Detecting historical changes to vegetation in a Cambodian protected area using the Landsat TM and ETM+ sensors

- Landsat CDR and relatively normalized images compared for mapping change using ΔVI
- Agreement was  >94% with ΔVI threshold set at lower and upper 5% of values.
- Relative radiometric normalization suppressed differences in vegetation response.
- EVI was more resilient to residual atmospheric effects present in images.
- ΔEVI from normalized images was used to analyze change to a Cambodian protected area.

The Phnom Kulen National Park (PKNP), Cambodia is faced with increasing environmental pressures and disturbances to vegetation. Changes in satellite vegetation indices (VIs) over time can be used to understand historical changes to vegetation in the PKNP provided images are calibrated to surface reflectance. Relative radiometric normalization (RRN) is one method to achieve this while the Landsat Climate Data Record (CDR) surface reflectance product provides an alternative. The objectives of this study are to: (1) determine the magnitude of differences between VIs produced using the CDR product and RRN images; (2) determine whether any differences in the VIs produced using the RRN and CDR images will impact on the areal extent of change in vegetation detected; and (3) determine the spatial and temporal extent of historical changes to vegetation in the PKNP. The choice of RRN or CDR had subtle but important impacts on the areal extent of detected changes to vegetation. The RRN images reduced inter-annual variations in vegetation response. The EVI was more resilient to residual atmospheric effects in the RRN and CDR images. Two relatively large regions of the PKNP were found to be relatively undisturbed. Recent changes to vegetation were found in areas that were previously undisturbed, and the proportion of the PKNP where changes were detected was increasing. Future work will include images from other sensors to reduce spatial and temporal gaps, and improve the temporal range of the methodology.