全球月尺度生态系统呼吸数据集v1.1（1980-2014）

该数据集为1980-2014年全球月尺度生态系统呼吸数据集。生态系统呼吸是生态系统尺度上初级生产者和消费者将有机碳转化为二氧化碳的过程。传统方法估算白天生态系统呼吸时，忽略了叶片呼吸在光照条件下被抑制的过程，导致传统方法（例如夜间法）高估了白天生态系统呼吸。通过构建物理约束机器学习方法来准确识别包含光抑制过程的白天生态系统呼吸，并结合夜间生态系统呼吸计算了月平均生态系统呼吸。使用机器学习方法训练月平均生态系统呼吸与气象数据（气温、饱和水气压差、土壤含水量、风速和CO2浓度）之间的关系，结合CMIP6耦合模式比较计划的4个模型输出的气象数据模拟了1980-2014年全球月尺度生态系统呼吸。空间分辨率为0.5度，时间分辨率为每月。

This dataset is a global monthly-scale ecosystem respiration dataset spanning from 1980 to 2014. Ecosystem respiration refers to the process at the ecosystem scale where primary producers and consumers convert organic carbon into carbon dioxide. When estimating daytime ecosystem respiration using traditional methods, the process of light-induced inhibition of leaf respiration is often overlooked, leading to overestimation of daytime ecosystem respiration by traditional approaches (e.g., the nighttime method). We developed a physically constrained machine learning method to accurately identify daytime ecosystem respiration that includes the light inhibition process, and calculated the monthly average ecosystem respiration by combining with nighttime ecosystem respiration. We trained the relationship between monthly average ecosystem respiration and meteorological data (air temperature, saturation vapor pressure deficit, soil moisture content, wind speed, and CO₂ concentration) using machine learning methods, and simulated the global monthly-scale ecosystem respiration from 1980 to 2014 by combining meteorological data outputs from four models of the Coupled Model Intercomparison Project Phase 6 (CMIP6). Its spatial resolution is 0.5 degrees, with a monthly temporal resolution.

搜集汇总

背景与挑战

背景概述

该数据集提供了1980年至2014年全球月尺度的生态系统呼吸数据，采用物理约束机器学习方法，通过结合气象数据和CMIP6模型输出模拟生成，空间分辨率为0.5度，时间分辨率为每月。其创新点在于解决了传统方法高估白天生态系统呼吸的问题，为全球碳循环研究提供了高时空分辨率的长期观测支持。