‘LandsatTS': an R package to facilitate retrieval, cleaning, cross‐calibration, and phenological modeling of Landsat time series data
‘LandsatTS': an R package to facilitate retrieval, cleaning, cross‐calibration, and phenological modeling of Landsat time series data
复制标题
DOI：
10.1111/ecog.06768
复制
发表时间：
2023-06
期刊：
影响因子：
5.9
通讯作者：
L. Berner;Jakob Johan Assmann;S. Normand;S. Goetz
中科院分区：
文献类型：
--
作者：
L. Berner;Jakob Johan Assmann;S. Normand;S. Goetz
陆地卫星提供了近几十年的近全球地表反射率测量数据，这些数据越来越多地用于评估陆地生态系统功能的年际变化。这些评估通常依赖于与植被绿度和生产力相关的光谱指数（如归一化植被指数NDVI）。然而，多种因素阻碍了使用Landsat卫星数据进行多年代际光谱指数评估，包括数据访问和清理的便利性，以及交叉传感器校准的遗留问题和无云采集时间不规律的挑战。为了帮助解决这些问题，我们为r开发了“LandsatTS”软件包。该软件包便于基于样本的地表反射率和光谱指数的时间序列分析，这些分析来自Landsat传感器。该软件包包括使用谷歌Earth Engine直接从r中访问的功能，可以从收集2中提取完整的Landsat 5、7和8记录，用于点样本位置或小研究区域。此外，该软件包还包括1)严格的数据清洗，2)交叉传感器校准，3)物象建模和4)时间序列分析。作为一个示例应用，我们展示了如何使用“LandsatTS”来评估2000年至2022年美国阿拉斯加州北部Noatak国家保护区年最大植被绿化率的变化。总的来说，该软件提供了一套功能，可以更广泛地使用Landsat卫星数据来评估和监测近几十年来在地方到全球地理范围内的陆地生态系统功能。
The Landsat satellites provide decades of near‐global surface reflectance measurements that are increasingly used to assess interannual changes in terrestrial ecosystem function. These assessments often rely on spectral indices related to vegetation greenness and productivity (e.g. Normalized Difference Vegetation Index, NDVI). Nevertheless, multiple factors impede multi‐decadal assessments of spectral indices using Landsat satellite data, including ease of data access and cleaning, as well as lingering issues with cross‐sensor calibration and challenges with irregular timing of cloud‐free acquisitions. To help address these problems, we developed the ‘LandsatTS' package for R. This software package facilitates sample‐based time series analysis of surface reflectance and spectral indices derived from Landsat sensors. The package includes functions that enable the extraction of the full Landsat 5, 7, and 8 records from Collection 2 for point sample locations or small study regions using Google Earth Engine accessed directly from R. Moreover, the package includes functions for 1) rigorous data cleaning, 2) cross‐sensor calibration, 3) phenological modeling, and 4) time series analysis. For an example application, we show how ‘LandsatTS' can be used to assess changes in annual maximum vegetation greenness from 2000 to 2022 across the Noatak National Preserve in northern Alaska, USA. Overall, this software provides a suite of functions to enable broader use of Landsat satellite data for assessing and monitoring terrestrial ecosystem function during recent decades across local to global geographic extents.