1.南宁师范大学;2.广西交通职业技术学院;3.广西林业科学研究院;4.广西遥感空间信息科技有限公司;5.广西大学
国家自然科学基金(42164001)；广西2022年本科教育教学重点项目建设经费(经费号:602030389173301)
- LIAO Chaoming 1
- LI Youru 1
- XU Binbin 2
- Qin Yuanli 3
- ZengShunqi 4
- YUN Ziheng 1
- GU Guanhai 5
- WEI Yuanyuan 1
- SHI Chunrou 1
- 摘要 | |
- 访问统计 |
- 参考文献 [27] | | | |
- 文章评论
钦州市作为世纪工程“平陆运河”承建主阵地，依托现代遥感技术产出高精度土地利用/覆被变化（LUCC，Land Use/Cover Change）数据集，对其国土空间规划及生态保护建设具有重要意义。本文面向北部湾滨海地区，以钦州市为例，基于谷歌地球引擎(Google Earth Engine，GEE)及Landsat遥感影像，在随机森林模型中融合光谱特征、纹理特征、指数特征以及地形特征，完成2012—2022年LUCC数据集制作及时空格局演变分析，引入最优参数地理探测器实现驱动机制探讨。结果表明:(1)经参数优化的随机森林模型可实现遥感信息有效提取，各期LUCC产品总体精度(OA，Overall Accuracy)均在0.88至0.92之间，Kappa系数介于0.86至0.90之间，结合若干4km×4km解译图斑与Google Earth同期同位高分辨率影像进行整体准确性评价，表明地物的分类结果与实际地物具有较好的一致性。(2)2012—2022年钦州市林地面积增加91.93km2，耕地面积减少284.73km2，建设用地面积提高180.05km2，综合土地利用动态度呈上升趋势。(3)研究期内，钦州市用地演变主要由经济动力（GDP）和地形特征（DEM和坡度）驱动，在2012—2017、2017—2022、2012—2022年三个阶段中，两者对时空格局演变呈现双因子增强作用，生态要素（NPP、降水）与GDP的交互效应在不同阶段呈现不同程度的双因子增强作用。本研究可为钦州市土地资源管理提供数据支撑，所产出的LUCC数据集对指导北部湾滨海区域发展有借鉴价值。
The city of Qinzhou, serving as the primary construction site for the century project "Pinglu Yunhe", exploits modern remote sensing technology to generate high-precision land use/cover change (LUCC) datasets, which exert substantial implications for its national spatial planning and ecological protection and construction. This research, concentrating on the Beibu Gulf coastal area and taking Qinzhou as an exemplar, employs Google Earth Engine (GEE) and Landsat remote sensing images to construct a LUCC dataset during the period from 2012 to 2022 and analyze its temporal and spatial pattern evolution by means of a random forest model integrating spectral, texture, index, and terrain features. The study also introduces the optimal parameter geographic detector to explore the driving mechanisms. The results demonstrate that: (1) The optimized random forest model can effectively extract remote sensing information, with overall accuracies (OA) ranging from 0.88 to 0.92 and Kappa coefficients ranging from 0.86 to 0.90. The overall accuracy and Kappa coefficient of each period"s LUCC product are in line with the classification results of several 4km×4km interpretation patches and Google Earth"s high-resolution images of the same period and location, indicating that the classification results of land features are consistent with the actual land features. (2) From 2012 to 2022, the area of forest land in Qinzhou increased by 91.93km2, the area of cultivated land decreased by 284.73km2, and the area of built-up land rose by 180.05km2. The overall trend of land use and cover change is ascending. (3) During the study period, the land use and cover change in Qinzhou was mainly driven by economic factors (GDP) and terrain features (DEM and slope), with the two factors presenting a dual-factor enhancement effect on the temporal and spatial pattern evolution in the three phases of 2012-2017, 2017-2022, and 2012-2022 respectively. The interactive effect between ecological factors (NPP and precipitation) and GDP shows different degrees of dual-factor enhancement at different stages. This study can provide data support for land resource management in Qinzhou and the produced LUCC datasets can serve as a reference for guidance.
