Phenology-Based Vegetation Index Differencing for Mapping of Rubber Plantations Using Landsat OLI Data

Hui Fan,Xiaohua Fu,Qiong Wu,Zheng Zhang

doi:10.3390/rs70506041

Abstract

Accurate and up-to-date mapping and monitoring of rubber plantations is challenging. In this study, we presented a simple method for rapidly and accurately mapping rubber plantations in the Xishuangbanna region of southwest China using phenology-based vegetation index differencing. Temporal profiles of the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Atmospherically Resistant Vegetation Index (ARVI), Normalized Difference Moisture Index (NDMI), and Tasselled Cap Greenness (TCG) for rubber trees, natural forests and croplands were constructed using 11 Landsat 8 OLI images acquired within one year. These vegetation index time series accurately demonstrated the unique seasonal phenological dynamics of rubber trees. Two distinct phenological phases (i.e., defoliation and foliation) of rubber trees were clearly distinguishable from natural forests and croplands. Rubber trees undergo a brief defoliation-foliation process between late December and mid-March. Therefore, vegetation index differencing between the nearly complete defoliation (leaf-off) and full foliation (leaf flushing) phases was used to delineate rubber plantations within fragmented tropical mountainous landscapes. The method presented herein greatly improved rubber plantation classification accuracy. Overall classification accuracies derived from the differences of the five vegetation indices varied from 92% to 96% with corresponding kappa coefficients of 0.84–0.92. These results demonstrate the promising potential of phenology-based vegetation index differencing for mapping and monitoring rubber expansion at the regional scale.

Highlights

Over the past several decades, massive rubber plantations have become major drivers of deforestation and other types of habitat degradation in Southeast Asia and southern Yunnan of China [1,2,3]
The lowest vegetation index values of the rubber plantations occurred in early February
This finding is inconsistent with Senf et al [10], who failed to differentiate effectively between rubber plantations and natural forests based on the MODIS Enhanced Vegetation Index (EVI)-based phenology information and explained that the phenology of natural forests partly resembled that of rubber trees in Xishuangbanna

Summary

Introduction

Over the past several decades, massive rubber plantations have become major drivers of deforestation and other types of habitat degradation in Southeast Asia and southern Yunnan of China [1,2,3]. Understanding and quantifying the large-scale impacts of rubber plantations requires accurate mapping of rubber plantations and timely measurements of the conversion of other land-cover types to rubber plantations [3]. Numerous studies have used satellite imagery for delineating rubber plantations [3,10,11,12,13,14,15,16]. These studies can be categorized into two broad groups: monotemporal image classification [11,13,14,15,16] and phenology-based classification [3,10,12]

Objectives

Results

Discussion

Conclusion