摘要:广西是我国最大的甘蔗种植和蔗糖产业基地，但长期受自然灾害影响甘蔗单产产量不高，及时获取其多年种植面积与长势的时空动态信息，可为区域甘蔗种植优化、灾害风险管理及蔗糖产业结构调整等提供重要科学支撑。首先，基于LANDSAT8 OLI遥感影像的6-5-2优化波段组合，引入NDVI、DEM等辅助识别特征变量，采用随机森林分类法进行多时相连续解译，并借助Google Earth高清遥感影像比对修正，获得了高精度的2014-2018年广西甘蔗种植面积分布；其次，基于MODIS-NDVI数据，构建长势差值监测模型，实现了近5年广西甘蔗茎伸期长势动态监测。结果表明：1）本文解译方法效果良好，广西甘蔗种植面积的总体分类精度高于92%，Kappa系数均大于0.8，面积相对误差5年均值为-10.7%。2）2014-2018年，广西甘蔗种植面积呈“前期急减，后期缓增”的变化趋势；主要种植区以崇左、南宁及来宾市为主，全区种植面积呈局部成片集聚、总体破碎分散的分布格局，并与研究区地形地貌、土壤类型、河流水系分布等下垫面环境要素密切相关。3）NDVI差值模型能清晰反映广西甘蔗茎伸期长势的年际和年内的时空变化特征，各年度内的甘蔗长势在好、正常、差等状态间交替转变频繁。上述成果可为揭示广西甘蔗对区域气候变化、旱涝交替及下垫面水土墒情动态的响应机制，开展区域甘蔗种植结构优化及其资源利用效率评估等奠定科学基础。
Abstract:Sugarcane planting in Guangxi has been affected by natural disasters, resulting in decreased yields. The information on spatio-temporal dynamics of the sugarcane planting area and growth can provide a reference for planting structure optimization and facilitate disaster control. This study incorporated 652 optimized band combinations of the LANDSAT 8 Operational Land Imager (OLI), normalized difference vegetation index (NDVI), digital elevation model (DEM), and other auxiliary identification characteristic variables into the random forest classification method to interpret continuously in multi-temporal aspects. Google Earth and high-resolution remote sensing image comparison and correction were used to obtain a high-precision sugarcane planting area distribution in Guangxi from 2014 to 2018. The MODIS-NDVI data was used to build a monitoring model of the growth potential difference for dynamic monitoring of sugarcane stem elongation in Guangxi in the last five years. The results showed that: 1) the interpretation method was effective, the overall classification accuracy of sugarcane planting area in Guangxi was >92%, the Kappa coefficient was >0.8, and the five-year mean area relative error was -10.7%. 2) In 2014-2018, the planting area of sugarcane in Guangxi had rapidly decreased in the early stage and slowly increased in the late stage. The main planting areas were in Chongzuo, Nanning, and Laibin. The whole planting area showed a distribution pattern of local agglomeration and overall fragmentation and dispersion, which was closely related to the underlying environmental elements, such as topography, soil type, and river system distribution. 3) The NDVI difference model reflected the interannual and intra-annual spatio-temporal changes in the elongation trend of sugarcane stems in Guangxi, and the yearly growth trend of sugarcane changes frequently between good, normal, and poor. These results revealed the response mechanism of sugarcane in Guangxi to regional climate change, alternation of drought and flood, and the dynamics of soil and water conservation on the underlying surface. Furthermore, this study provides a scientific foundation for optimizing the regional sugarcane planting structure and evaluating water resource utilization efficiency.
PDF全文下载地址:
http://www.ecoagri.ac.cn/article/exportPdf?id=174fb175-8d35-45f2-b6b7-8114d5ac05d9
