A coarse pixel scale ground \"truth\" surface albedo dataset based on the global in situ site measurements from 2000 to 2021@en

In situ measurements from sparsely distributed networks worldwide are a valuable source of data for validating satellite products. However, the significant differences in spatial scale between in situ and satellite measurements pose a challenge for validation. To address this challenge, we propose a new global coarse pixel-scale ground truth dataset which provide sequential pixel scale albedo on a 500 m resolution over the period 2000-2021. The creation of data set is done using ground measurements from 368 sparsely distributed networks worldwide and high-resolution long-term time series data corresponding to site-based data. Furthermore, we carefully evaluate the effectiveness of the new dataset over sites with different degrees of spatial representativeness. The results demonstrate superior overall accuracy compared to in situ measurement, especially in regions with strong spatial heterogeneity while most in situ measurements have poor spatial representativeness and cannot represent the pixel scale. The dataset offers a unique collection of pixel-scale relative ground \"truth\" with the widest spatial distribution and longest time series. Its ability to capture both spatial and temporal variations of surface albedo at coarse spatial scales makes it an invaluable resource for validating and correcting global surface albedo products.全球稀疏分布的原位（in situ）观测网络所获取的实测数据，是验证卫星遥感产品的宝贵数据源。然而，原位观测与卫星遥感在空间尺度上的显著差异，给验证工作带来了极大挑战。为应对这一挑战，本研究提出一套全新的全球粗分辨率像元尺度地面真值（ground truth）数据集，该数据集在2000-2021年期间提供500米分辨率的时序像元尺度反照率（albedo）数据。该数据集的构建依托全球368个稀疏分布观测网络的实测数据，以及与站点观测匹配的高分辨率长时间序列数据。此外，本研究针对不同空间代表性程度的站点，系统评估了该新数据集的有效性。研究结果显示，相较于原位观测数据，该数据集整体精度更优，在空间异质性较强的区域表现尤为突出——而多数原位观测因空间代表性不足，无法表征像元尺度的地表特征。本数据集提供了一套独特的像元尺度相对地面真值集合，其空间覆盖范围最广、时间序列跨度最长。该数据集能够在粗空间尺度上捕捉地表反照率的空间与时间变化特征，因此可作为验证与修正全球地表反照率产品的宝贵资源。