Protecting the ecological environment is an important goal of the world sustainable development. Rapid and quantitative evaluation of regional ecological environment is the technical support and necessary condition for this goal. The ecological environment index model (RSEI) which used to assess ecological environment is the most popular now. But it changed into two completely opposite models in the application. Most researchers choose which model to use based on the desired results. This article concludes the reason by studying the operating mechanism of the model and finds that it is the eigenvector direction in the principal component analysis causes this to happen. Taking Pingyu County as an example, this article calculates RSEI with Landsat 8 images in different periods in Google Earth Engine using the two existing models respectively and finds that two models show two opposite result trends in spatial distribution. Using any model to calculate the same image, the results are also opposite if changing the input order of the indicators. It is the eigenvector direction determines the spatial distribution by comparing and analyzing the eigenvector of each image and its corresponding RSEI. Then, this paper improves the model by fixing the eigenvector direction based on the actual effects on ecological environment of the four indicators, taking absolute values of the eigenvectors of NDVI and Wet which have a positive effect on the ecological environment and the opposite of absolute values of the eigenvectors of LST and NDSI which have a negative effect on the ecological environment, in order to improve the RSEI model. Using the improved model calculate each image, the results are consistently accurate. Furthermore, this paper also proposed a model for users who calculating the principal components through software where the eigenvector direction cannot be altered artificially. This paper proposes the improved model which is suitable for all users whether using software or conducting programming. The improved model is suitable for all images of any input order of the indicators. It provides the possibility of applying remote sensing big data to the ecological environment. At the same time, the study of the mechanism of the model provides a scientific basis for future scholars to calculate in batches.

中文翻译：

---

生态环境指标模型RSEI的改进

保护生态环境是世界可持续发展的重要目标。快速定量地评价区域生态环境是实现这一目标的技术支持和必要条件。如今，用于评估生态环境的生态环境指数模型（RSEI）最受欢迎。但是在应用程序中它变成了两个完全相反的模型。大多数研究人员根据期望的结果选择使用哪种模型。本文通过研究模型的运行机理总结了原因，并发现这是主成分分析中特征向量方向的原因。以平榆县为例 本文分别使用两个现有模型，分别使用Google Earth Engine中不同时期的Landsat 8图像来计算RSEI，发现两个模型在空间分布上显示出两个相反的结果趋势。使用任何模型来计算同一图像，如果更改指标的输入顺序，结果也将相反。特征向量方向是通过比较和分析每个图像及其对应的RSEI的特征向量来确定空间分布的。然后，根据四个指标对生态环境的实际影响，通过固定特征向量方向来改进模型。取NDVI和Wet特征向量的绝对值对生态环境有正影响，而LST和NDSI特征向量的绝对值对生态环境有负影响，以改善RSEI模型。使用改进的模型计算每张图像，结果始终准确。此外，本文还为用户提供了一个模型，该模型可通过软件计算特征向量方向不能人为改变的主分量。本文提出了一种改进的模型，该模型适用于所有使用软件或进行编程的用户。改进后的模型适用于指标的任何输入顺序的所有图像。它提供了将遥感大数据应用于生态环境的可能性。