基于改进遥感生态指数的宁夏沿黄平原区生态环境质量评价

区域生态环境质量评价是国民经济建设与可持续性发展规划的基础,是生态学研究的主要方向之一。以土地利用/覆盖数据和Landsat OLI数据为基础,分别在ArcGIS和GEE平台上进行景观多样性指数（Landscape Diversity Index,LDI）与归一化植被指数（Normalized Difference Vegetation Index,NDVI）、湿度（Wet Index,WI）、归一化裸土和建筑指数（Normalized Difference Building-Soil Index,NDBSI）、遥感生态指数（Remote Sensing Based Ecological Index,RSEI）和改进遥感生态指数（Modified Remote Sensing Ecological Index,MRSEI）的计算。在LDI最佳尺度约束下分析表明,宁夏沿黄平原区景观多样性指数具有显著尺度依赖特征（P<0.001）,阈值出现在3000 m×3000 m。主成分分析（Principal Component Analysis,PCA）解释了研究区改进遥感生态指数主要受到NDVI和LDI影响,其中NDVI是PC1（特征值贡献率68.98%）的决定因子,特征向量为0.8901;LDI为次要决定因子,特征向量为-0.4146,该分量在MRSEI计算中分值较高。LDI是PC2（特征值贡献率28.76%）的决定因子,特征向量为0.9100;NDVI为次要决定因子,特征向量为0.4056,该分量在MRSEI计算中分值较低。从MRSEI在应用中可信性来看,其在分析中采用LDI替代LST有效地避免了RSEI分析中NDBSI和LST之间存在的生态学意义重复表达和多因子向量投影中的高度聚集。研究区空间异质性主要以"差"和"较差"级别分布在不同土地利用/覆盖类型交错区且以环绕研究区为主要特点。在"差"到"好"梯度上,斑块密度为减少趋势由8.3个/km²减少到5.9个/km²,而平均斑块面积呈增加趋势由0.120 km²增加到0.169 km²。综合来看宁夏沿黄平原区生态环境质量总体MRSEI值为0.0117刚好超过"较好"水平下限。

Eco-environmental quality assessment of the Ningxia plain along the Yellow River based on the modified remote sensing ecological index

Regional eco-environmental quality assessment is not only the basis of national economic construction and sustainable development planning, but also one of the main directions of ecological research. For an assessment of eco-environmental quality in the Ningxia Plain along the Yellow River, based on land use/cover data and Landsat OLI data, this paper calculated Landscape Diversity Index (LDI) and Normalized Difference Vegetation Index (NDVI), Wet Index (WI), Normalized Difference Building-Soil Index (NDBSI), Remote Sensing Based Ecological Index (RSEI), and Modified RSEI (MRSEI) on ArcGIS and GEE platforms, respectively. The analysis under the best scale constraint of LDI showed that the LDI in the study area had significant scale dependence (P<0.001), and its threshold value occurred at 3000 m×3000 m of analysis window. Principal Component Analysis (PCA) explained that the MRSEI of the study area was mainly affected by NDVI and LDI, in which NDVI was the most important factor of PC1 (eigenvalue contribution of 68.98%) with eigenvector 0.8901 while LDI was the minor determinant with eigenvector -0.4146, which had a high value in the MRSEI calculation. LDI was the determinant of PC2 (eigenvalue contribution of 28.76%) with eigenvector 0.9100 while NDVI was the minor determinant with eigenvector 0.4056, and this component had a low score in MRSEI calculation. Regarding the application of MRSEI, its use of LDI instead of LST in the analysis effectively avoided the duplicate expression of ecological significance and high aggregation in the multi-factor vector projection between NDBSI and LST in the RSEI analysis. In terms of the MRSEI, the spatial heterogeneity was mainly distributed at the "poor" and "worse" levels in the ecotones of different land use/cover types and mainly around the study area. On the "poor" to "excelllent" gradient, the patch density tended to decrease from 8.3 to 5.9 patches/km², while the average patch area tended to increase from 0.120 to 0.169 km². The overall MRSEI value of 0.0117 was just above the lower limit of "good" level for eco-environmental quality in the study area.