Forest Disturbance in China from 1986 to 2020: 30 m disturbance attributes and 1 km annual density products

Abstract: This repository hosts a comprehensive forest disturbance dataset for China, spanning the period from 1986 to 2020. The dataset was generated using the Continuous Change Detection and Classification (CCDC) algorithm on the Google Earth Engine (GEE) platform, utilizing all available Landsat time-series imagery from 1985 to 2021. This dataset provides a dual-scale perspective on forest dynamics, offering both fine-scale disturbance attributes (30 m resolution) and aggregated disturbance density (1 km resolution) to support diverse ecological and climatic applications. The dataset consists of two major components: 1. 30 m Resolution Forest Disturbance Attributes: These raster files record the specific characteristics of detected forest disturbance events at a 30 m spatial resolution. (1) CFD_year.tif: Records the year of the forest disturbance event. The pixel values are stored as indices relative to the base year 1985. Value Range: 1 represents 1986, 2 represents 1987, ..., and 35 represents 2020. Nodata value: -128 (indicating no disturbance or non-forest areas). (2) CFD_mag.tif: This dataset records the maximum disturbance magnitude (abbreviated as "mag"), representing the maximum spectral change detected within a specific time segment and spectral band. It is quantified by the difference between the end-point of the pre-disturbance segment and the start-point of the post-disturbance segment, where a larger value indicates a more severe disturbance. To facilitate interpretation, the raw magnitude values were reclassified into five discrete classes using the Jenks Natural Breaks method. The final pixel values correspond to the following ranges: Class 1 (0 < mag ≤ 0.09), Class 2 (0.09 <mag ≤ 0.19), Class 3 (0.19 < mag ≤ 0.30), Class 4 (0.30 <mag ≤ 0.45), and Class 5 (mag > 0.45). (3) CFD_num.tif: Records the total number of disturbance events detected within the pixel during the study period. 2. 1 km Resolution Annual Forest Disturbance Density (CFDD): Derived from the 30 m disturbance year data, this product quantifies the spatiotemporal density of disturbances. CFDD[Year].tif (e.g., CFDD2000.tif): Represents the annual forest disturbance density. Unit: The pixel value indicates the ratio of disturbed forest area within each 1 km × 1 km grid cell (Unit: % or dimensionless fraction).摘要：本仓库托管了一套覆盖1986至2020年的中国全域森林扰动综合数据集。本数据集依托谷歌地球引擎（Google Earth Engine，GEE）平台的连续变化检测与分类（Continuous Change Detection and Classification，CCDC）算法，通过整合1985至2021年的全部可用Landsat时间序列影像生成。该数据集从双尺度视角展现森林动态特征，提供了精细尺度的扰动属性（空间分辨率30米）与聚合式扰动密度（空间分辨率1千米）两类数据，可支撑多样化的生态与气候相关应用研究。 本数据集包含两大核心组成部分： 1. 30米分辨率森林扰动属性：此类栅格文件记录了空间分辨率30米尺度下探测到的森林扰动事件的具体特征。 (1) CFD_year.tif：记录森林扰动事件发生年份。像素值以1985年为基准年进行索引编码：取值1对应1986年，取值2对应1987年，以此类推，取值35对应2020年。无数据值为-128，代表无扰动区域或非森林区域。 (2) CFD_mag.tif：该数据集记录了最大扰动幅度（简称“mag”），即特定时段与光谱波段内探测到的最大光谱变化量。其量化方式为扰动前时段端点与扰动后时段起始点的光谱差值，数值越大代表扰动程度越剧烈。为便于解译，原始幅度值通过Jenks自然断点法被重分类为5个离散等级，最终像素值对应区间如下：等级1（0 < mag ≤ 0.09）、等级2（0.09 < mag ≤ 0.19）、等级3（0.19 < mag ≤ 0.30）、等级4（0.30 < mag ≤ 0.45）以及等级5（mag > 0.45）。 (3) CFD_num.tif：记录研究时段内该像素点探测到的扰动事件总次数。 2. 1千米分辨率年度森林扰动密度（CFDD）：该产品由30米分辨率的扰动年份数据聚合生成，用于量化扰动的时空密度。CFDD[Year].tif（例如CFDD2000.tif）代表对应年度的森林扰动密度。单位：像素值表示每个1千米×1千米网格内受扰动森林面积占比，单位为百分比或无量纲分数。