全球500m4天分辨率植被覆盖率数据集（2010-2020）

植被覆盖度（Fractional Vegetation Cover，FVC）定义为植被冠层或叶面在地面的垂直投影面积占植被区总面积的比例，是刻画地表植被覆盖程度的一个重要参数，也是指示生态环境变化的重要指标之一。使用基于孔隙率理论算法进行FVC定量遥感反演，实现长时间序列、高精度FVC产品生产。算法以多天合成叶面积指数（LAI）和聚集指数（CI）产品作为输入，基于孔隙率理论生产2010-2020年全球500m/4天FVC产品，具有良好的全球适用性。与ImagineS地面实测数据相比，FVC产品算法精度为RMSE=0.13和BIAS=-0.05，与地面实测FVC数据具有更好的一致性，与GEOV1 FVC神经网络算法和MuSyQ FVC像元二分模型估算方法相比，通过引入CI参数，考虑植被结构和空间分布特征，能够有效抑制浓密植被FVC高估问题。该数据集采用Sinusoidal Tile Grid分幅，每景覆盖的经纬度范围为10°x10°，该数据集存储为*.tif 格式。

Fractional Vegetation Cover (FVC) is defined as the ratio of the vertical projected area of vegetation canopies or foliage onto the ground surface to the total area of the vegetated region. As a critical parameter characterizing terrestrial vegetation coverage, FVC is also one of the key indicators for monitoring ecological environment changes. This dataset employs a porosity theory-based algorithm for quantitative remote sensing inversion of FVC, enabling the production of long-term time-series and high-precision FVC products. Taking multi-day composite Leaf Area Index (LAI) and Clumping Index (CI) products as inputs, the algorithm generates global 500m/4-day FVC products spanning 2010 to 2020 based on the porosity theory, demonstrating excellent global applicability. Compared with ImagineS in-situ FVC measurements, the proposed FVC product achieves an accuracy of RMSE=0.13 and BIAS=-0.05, showing better consistency with in-situ FVC observations. Compared with the GEOV1 FVC neural network algorithm and the MuSyQ FVC dimidiate pixel model estimation method, this algorithm effectively mitigates the overestimation of FVC in dense vegetation regions by introducing the CI parameter and accounting for vegetation structure and spatial distribution characteristics. This dataset uses the Sinusoidal Tile Grid for tiling, where each tile covers a 10°×10° geographic extent, and is stored in the *.tif format.

搜集汇总

背景与挑战

背景概述

该数据集是全球范围的500米空间分辨率和4天时间分辨率的植被覆盖率（FVC）产品，覆盖2010年至2020年。基于孔隙度理论算法，利用多日合成叶面积指数和聚集指数作为输入，生成高精度FVC数据，具有较好的全球适用性，并通过引入聚集指数参数有效抑制了密集植被覆盖率高估的问题。