Spatiotemporal Contextual Uncertainties in Green Space Exposure Measures: Exploring a Time Series of the Normalized Difference Vegetation Indices
Spatiotemporal Contextual Uncertainties in Green Space Exposure Measures: Exploring a Time Series of the Normalized Difference Vegetation Indices
复制标题
DOI：
10.3390/ijerph16050852
复制
发表时间：
2019-03-01
影响因子：
--
通讯作者：
Helbich, Marco
中科院分区：
文献类型：
--
作者：
Helbich, Marco
关于绿色空间的环境健康研究可能会受到源于环境暴露的暂时性和空间背景如何界定的背景不确定性的影响。归一化差异植被指数(NDVI)是一种常用的室外绿地度量指标，它反映了植被冠层中叶绿素的含量。这项研究评估了(1)居民的NDVI暴露是否随着时间的推移而变化，以及(2)这些NDVI得分的时间序列如何在空间背景范围内变化。荷兰2006-2017年期间的多时相NDVI数据来自中分辨率成像光谱仪(MODIS)卫星平台。以10,000个荷兰住宅地址的随机抽样为中心，确定了多个缓冲区大小(即300、600和1000米)的年度NDVI暴露。除了描述性统计外，还使用配对Wilcoxon检验和Fligner-Killeen检验来确定不同缓冲区宽度的年度NDVI得分的均值和方差差异。热图将关联矩阵可视化。对多重假设检验的显著性水平进行了调整。结果表明，年NDVI指数之间存在显着的相关性，但其大小在0.60~0.97之间变化较大。年NDVI暴露的许多均值和方差差异显著。似乎不同的缓冲区(即300m和1000m)的相关性不那么强，可能是因为较大的缓冲区降低了方差异质性。这些结果多年来大体上是一致的，并通过了基于蒙特卡洛的敏感性测试。总而言之，除了评估不同缓冲区大小的绿地暴露，我们的发现建议绿地健康研究应该使用与流行病学数据很好地一致的NDVI数据。即使是每年的时间不相容，也可能掩盖或扭曲绿色空间与健康的联系。这两种战略都可能减少环境暴露评估中的背景不确定性。
Environmental health studies on green space may be affected by contextual uncertainties originating from the temporality of environmental exposures and by how the spatial context is delimitated. The Normalized Difference Vegetation Index (NDVI) is frequently used as an outdoor green space metric capturing the chlorophyll content in the vegetation canopy. This study assessed (1) whether residential NDVI exposures vary over time, and (2) how these time series of NDVI scores vary across spatial context delimitations. Multi-temporal NDVI data for the period 2006-2017 for the Netherlands were obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite platform. Annual NDVI exposures were determined across multiple buffer sizes (i.e., 300, 600, and 1000 m) centered on a random sample of 10,000 Dutch residential addresses. Besides the descriptive statistics, pairwise Wilcoxon tests and Fligner-Killeen tests were used to determine mean and variance differences in annual NDVI scores across buffer widths. Heat maps visualized the correlation matrices. Significance levels were adjusted for multiple hypotheses testing. The results indicated that annual NDVI metrics were significantly correlated but their magnitude varied notably between 0.60 to 0.97. Numerous mean and variance differences in annual NDVI exposures were significant. It seems that the disparate buffers (i.e., 300 and 1000 m) were less strongly correlated, possibly because variance heterogeneity is reduced in larger buffers. These results have been largely consistent over the years and have passed Monte Carlo-based sensitivity tests. In conclusion, besides assessing green space exposures along different buffer sizes, our findings suggest that green space-health studies should employ NDVI data that are well-aligned with epidemiological data. Even an annual temporal incompatibility may obscure or distort green space-health associations. Both strategies may diminish contextual uncertainties in environmental exposure assessments.