Telesca Luciano, Aromando Angelo, Faridani Farid, Lovallo Michele, Cardettini Gianfranco, Abate Nicodemo, Papitto Giancarlo, Lasaponara Rosa
Institute of Methodologies for Environmental Analysis, National Research Council, 85050 Tito, Italy.
DICEM, Department of European and Mediterranean Cultures, Environment, and Cultural Heritage, University of Basilicata, 85100 Potenza, Italy.
Entropy (Basel). 2022 Dec 6;24(12):1784. doi: 10.3390/e24121784.
The main goal of this study was to evaluate the potential of the Fisher-Shannon statistical method applied to the MODIS satellite time series to search for and explore any small multiyear trends and changes (herein also denoted as inner anomalies) in vegetation cover. For the purpose of our investigation, we focused on the vegetation cover of three peri-urban parks close to Rome and Naples (Italy). For each of these three areas, we analyzed the 2000-2020 time variation of four MODIS-based vegetation indices: evapotranspiration (ET), normalized difference vegetation index (NDVI), leaf area index (LAI), and enhanced vegetation index (EVI). These data sets are available in the Google Earth Engine (GEE) and were selected because they are related to the interactions between soil, water, atmosphere, and plants. To account for the great variability exhibited by the seasonal variations while identifying small multiyear trends and changes, we devised a procedure composed of two steps: (i) application of the Singular Spectrum Analysis (SSA) to each satellite-based time series to detect and remove the annual cycle including the seasonality and then (ii) analysis of the detrended signals using the Fisher-Shannon method, which combines the Shannon entropy and the Fisher Information Measure (FIM). Our results indicate that among all the three pilot test areas, Castel Volturno is characterized by the highest Shannon entropy and the lowest FIM that indicate a low level of order and organization of vegetation time series. This behaviour can be linked to the degradation phenomena induced by the parasite ( that has affected dramatically the area in recent years. Our results were nicely confirmed by the comparison with in situ analyzed and independent data sets revealing the existence of subtle, small multiyear trends and changes in MODIS-based vegetation indices.
本研究的主要目标是评估将Fisher-Shannon统计方法应用于MODIS卫星时间序列，以搜索和探索植被覆盖中任何微小的多年趋势和变化（在此也称为内部异常）的潜力。为了我们的调查目的，我们重点关注了意大利罗马和那不勒斯附近三个城郊公园的植被覆盖情况。对于这三个区域中的每一个，我们分析了基于MODIS的四个植被指数在2000 - 2020年的时间变化：蒸散量（ET）、归一化植被指数（NDVI）、叶面积指数（LAI）和增强植被指数（EVI）。这些数据集可在谷歌地球引擎（GEE）中获取，之所以选择它们，是因为它们与土壤、水、大气和植物之间的相互作用有关。为了在识别微小的多年趋势和变化时考虑到季节变化所表现出的巨大变异性，我们设计了一个由两步组成的程序：（i）对每个基于卫星的时间序列应用奇异谱分析（SSA），以检测和去除包括季节性在内的年周期，然后（ii）使用结合了香农熵和费希尔信息度量（FIM）的Fisher-Shannon方法对去趋势信号进行分析。我们的结果表明，在所有三个试点测试区域中，卡斯特尔沃尔图诺的特点是香农熵最高，费希尔信息度量最低，这表明植被时间序列的有序性和组织性较低。这种行为可能与寄生虫引起的退化现象有关（近年来该地区受到了严重影响）。通过与现场分析和独立数据集的比较，很好地证实了我们的结果，揭示了基于MODIS的植被指数中存在细微的、微小的多年趋势和变化。
