基于高时空分辨率可见光遥感数据的桉树种植面积提取方法研究

Extraction of Eucalyptus Planting Area Based on High Spatio-Temporal Resolution Visible Remote Sensing Data

摘要: 为了缓解国内木材市场的供求矛盾，我国高度重视速生林基地建设，而桉树作为世界三大速生树种之一，因其易于种植、适应性强且轮伐周期短的特点而具有较大的经济价值。大量、单一的速生桉代替了原有的自然林和耕地，造成生物多样性损失、局部水资源短缺等生态环境问题。对该区域桉树林的种植面积和种植结构变化的精确监测是客观评价该地区桉树林种植与生态环境变化关系的关键。本文与澜沧地区为例，分别对Sentinel-2和MOD09GA进行植被指数提取和缨帽变换，并基于时空数据融合技术，分别融合植被指数和缨帽变换指数，获取高时空分辨率NDVI数据和TCA数据，提取桉树林的植被变化特征，实现高精度的桉树林分布制图。结果表明：1) 基于时空融合数据提取的橡胶林物候变化特征能够实现桉树林的识别，识别精度可以达到80.12%，Kappa系数达到0.71；2) 用NDVI指数数据分类时，能够获取比TCA指数数据更高精度的分类结果，表明植被指数数据在高时空数据融合及植被遥感应用中有较好的应用前景。

Abstract: In order to alleviate the contradiction between supply and demand in the domestic timber market, China attaches great importance to the construction of fast-growing forest bases. As one of the world’s three fastest-growing tree species, eucalyptus has great economic value because of its easy planting, adaptability and short rotation cycle. A large number of single fast-growing rafts have replaced the original natural forests and cultivated land, causing ecological and envi-ronmental problems such as biodiversity loss and local water shortages. Accurate monitoring of the planting area and planting structure changes of the eucalyptus forest in this area is the key to objectively evaluate the relationship between the plantation of the eucalyptus forest and the ecological environment in the region. In this paper, we take sentinel-2 and MOD09GA in Lancnag region as examples to extract vegetation index and transform tassel cap, and based on the spatio-temporal data fusion technology, the vegetation index and the cap transformation index are respectively combined to obtain the high spatial-temporal resolution NDVI data and TCA data, to extract the characteristics of vegetation changes in the eucalyptus forest and to achieve high-precision eucalyptus distribution mapping. The results show that: 1) The phenological change characteristics of rubber forest based on spatio-temporal fusion data extraction can realize the identification of eucalyptus forest; the recognition accuracy can reach 80.12%, and the Kappa coefficient reaches 0.71. 2) When using NDVI index data classification, it can obtain higher precision classification results than TCA index data, indicating that vegetation index data has a good application prospect in high-temporal data fusion and vegetation remote sensing ap-plications.