DTSIF：黄淮海地区高时空分辨率（4天和8天，500米）日光诱导叶绿素荧光数据集V1.0（2002-2025）

黄淮海地区高时空分辨率日光诱导叶绿素荧光（SIF）数据集（DTSIF），主要采用MODIS反射率数据（波段1-7）、光合有效辐射比（FPAR）和叶面积指数（LAI）数据、ERA5再分析数据中的温度、辐射、土壤湿度和再计算得到的饱和水汽压差数据作为模型输入解释变量，以及采用2018-2021年TROPOMI SIF数据作为模型输入被解释变量，综合对比8种机器学习模型的训练和测试精度，最终择优选取模拟效果最好的ExtraTrees机器学习模型（测试R2为0.896，RMSE为0.0787 W·m-2·μ-1·sr-1）对TROPOMI SIF进行重构，数据集时间跨度2002-2025年，时间分辨率4天和8天，空间分辨率500m，区域上覆盖北京、天津、山东、河北、河南、安徽和江苏等7省（市）。该数据集经过8个ChinaFlux站点观测的GPP数据和3个ChinaSpec站点观测的SIF数据以及现有卫星反演SIF数据（如CSIF和GOSIF）和MODSI GPP数据分别从站点和区域尺度进行效果检验，总体模拟效果较好，数据精度较高，可为黄淮海地区植被光合作用监测、物候研究、生态气象监测评估、生态质量气象评价、碳通量估算、农业气象灾害监测预警、作物产量预测等气候变化研究以及精细化生态气象与卫星遥感和农业气象科研和业务服务提供新的高精度数据支持。

This is the high spatiotemporal resolution solar-induced chlorophyll fluorescence (SIF) dataset (DTSIF) for the Huang-Huai-Hai Region. Primarily, this dataset uses MODIS reflectance data (bands 1–7), fraction of photosynthetically active radiation (FPAR) and leaf area index (LAI) data, temperature, radiation, and soil moisture from the ERA5 reanalysis dataset, as well as recalculated vapor pressure deficit data as explanatory variables for model input, and adopts TROPOMI SIF data from 2018 to 2021 as the dependent variable for model training. After comprehensively comparing the training and testing accuracies of eight machine learning models, the ExtraTrees machine learning model with the optimal simulation performance was finally selected (with a testing R² of 0.896 and RMSE of 0.0787 W·m⁻²·μ⁻¹·sr⁻¹) to reconstruct TROPOMI SIF. The dataset spans the period from 2002 to 2025, with temporal resolutions of 4 days and 8 days, a spatial resolution of 500 m, and covers seven provinces and municipalities including Beijing, Tianjin, Shandong, Hebei, Henan, Anhui, and Jiangsu. This dataset was validated at both site and regional scales using GPP data from 8 ChinaFlux sites, SIF data from 3 ChinaSpec sites, existing satellite-retrieved SIF datasets (such as CSIF and GOSIF), and MODIS GPP data. The overall simulation performance is favorable, and the data accuracy is high. It provides new high-precision data support for climate change-related research in the Huang-Huai-Hai Region, including vegetation photosynthesis monitoring, phenology research, ecological meteorological monitoring and assessment, ecological quality meteorological evaluation, carbon flux estimation, agricultural meteorological disaster monitoring and early warning, and crop yield prediction, as well as refined ecological meteorology, satellite remote sensing, and agricultural meteorology research and operational services.

搜集汇总

背景与挑战

背景概述

DTSIF是一个高时空分辨率的日光诱导叶绿素荧光数据集，覆盖黄淮海地区2002至2025年，时间分辨率为4天和8天，空间分辨率为500米。该数据集采用ExtraTrees机器学习模型重构，精度较高（测试R2为0.896），主要用于植被光合作用监测、生态气象评估和气候变化研究等领域。