英文摘要: |
A novel phenology-based algorithm, namely the Landsat-derived renormalized modified normalized burn ratio (RMNBR) index, was developed with Xishuangbanna in southern China, also the second-largest national base of natural rubber production. The RMNBR algorithm included two key steps, namely the modification of NBR (or MNBR) by normalizing near infrared (NIR) and two short-wave infrared (SWIR 1/2) bands of Landsat-8 Operational Land Imager (OLI) and, the re-normalization of MNBR between two critical phenological stages of rubber plantations, i.e. defoliation and foliation. It highlighted the temporal differences of surface reflectance of three non-visible bands (i.e. NIR, SWIR 1 and SWIR 2) in canopy density and moisture content of rubber plantations from defoliation to formation during the dry season. Then, the RMNBR was used to identify mature rubber plantations (pixels) in 2018, by combining two auxiliary masks of Landsat-based forest of the same year and topographically-suitable area of rubber trees cultivation. This makes the discrimination of mature rubber plantation straightforward as it only needs to distinguish mature rubber plantations (RMNBR < 0) from other land cover types (RMNBR > 0) within the two critical periods. The phenology-based RMNBR algorithm greatly enriches the remotely-sensed approaches for rubber plantations mapping at the regional scale. |