基于连续变化检测和分类算法的动态遥感生态指数构建
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

基金项目:

国家自然科学基金项目(41901121,42206236);宁波市自然科学基金项目(2022J075,2022J092)


Construction of dynamic remote sensing ecological index using the Continuous Change Detection and Classification (CCDC) algorithm
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 文章评论
    摘要:

    沿海地区经济社会高速发展,是生态环境变化的焦点区域。然而,沿海地区云雨天气频发,遥感信息获取能力受限,导致遥感生态质量指数(RSEI)评价结果受成像日期变化而波动,可比性较差。针对以上问题,研究利用连续变化检测和分类(CCDC)算法构建时间序列模型,通过合成任意时刻影像、重构遥感生态指数以及改进指数归一化方式,研发了一种动态遥感生态指数(DRSEI),细化了RSEI在区域生态质量监测的时间尺度,并应用于沿海城市宁波生态质量时空变化监测。结果表明:(1)RSEI对时间差异较为敏感,当影像年内成像时间相差逾1个月,RSEI差异可达0.147,这种差异会对长期生态质量动态监测的稳定性和准确性造成影响。(2)基于合成影像的DRSEI平均绝对偏差为0.097,接近成像时间相差半个月的RSEI差异(0.072),误差相对较小,一定程度上减小了真实影像时相差异引起的误差。(3)DRSEI能够表征任意时刻生态质量,通过年际(1986-2019年)和半月际(2019年)DRSEI分析揭示了宁波市生态质量总体下降趋势和时空异质性加剧过程。具体地,1986-2019年宁波市南部和西部森林区域的DRSEI持续上升,而近郊农田快速转化为建成区导致DRSEI不断下降。研究提出的DRSEI能够精确描述区域生态质量变化趋势,准确定位生态质量变化转折点,有望服务海岸带地区的生态质量定期监测与评估工作,支持沿海城市高质量发展与生态环境保护。

    Abstract:

    Coastal areas have become the focal regions of ecological environment changes during rapid urbanization and economic development. However, in coastal areas, the cloudy and rainy weather is frequent, leading quite limited remote sensing images with high quality. And due to variant dates of images, the prevalent Remote Sensing based Ecological Index (RSEI) from different periods can be hardly comparable. To solve this problem, we firstly synthesized Landsat images at any time using the Continuous Change Detection and Classification (CCDC) algorithm; afterwards, we reconstructed parts of remote sensing indices and modified the index normalization mode using time-series methods; and based on which, we finally developed a dynamic RSEI (DRSEI) and applied it on the coastal city of Ningbo to monitor the spatio-temporal changes of ecological quality as a case study. The results demonstrated that:(1) the RSEI was sensitive to timespan-the mean absolute deviation (MAD) of the RSEI reached 0.147 when the acquisition date of two real Landsat images was more than 1 month. This sensitivity would affect the stability and accuracy of long-term ecological quality monitoring. (2) In contrast, the MAD of the DRSEI from synthetic images based on CCDC algorithm was minor (0.097), relatively equivalent to the value between the two real images with timespan of half-a-month (0.072). That means, the sensitivity of RSEI to the timespan of real images were offset by the DRSEI using synthetic images to a certain extent. (3) The DRSEI was capable of monitoring regional ecological quality at any time. As an example, the annual (1986-2019) and semi-monthly (2019) DRSEI revealed the overall declining trend with increasingly spatio-temporal heterogeneity of the ecological quality in Ningbo City. To be specific, the rising DRSEI in forest areas and the falling DRSEI from the urbanization were two aspects for the spatial heterogeneity of ecological quality-the gradual increase of DRSEI in the southern and western forest areas could be seen in Ningbo City during 1986-2019 while the sudden decrease of DRSEI usually came from the conversion from farmland to built-up areas in suburbs during the urbanization. The DRSEI proposed by our study can accurately describe the overall trend and locate the turning points of regional ecological quality. Therefore, it can be used to correspond different stages and protection strategies during resource development, which is expected to serve the regularly ecological quality monitoring and assessment, supporting the high-quality development and ecological conservation for coastal areas.

    参考文献
    相似文献
    引证文献
引用本文

张书,孙超,胡茗,郑嘉豪,刘永超.基于连续变化检测和分类算法的动态遥感生态指数构建.生态学报,2024,44(2):497~510

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数: