CLASSIC v1.0: the open-source community successor to the Canadian Land Surface Scheme (CLASS) and the Canadian Terrestrial Ecosystem Model (CTEM) – Part 2: Global benchmarking
CLASSIC v1.0: the open-source community successor to the Canadian Land Surface Scheme (CLASS) and the Canadian Terrestrial Ecosystem Model (CTEM) – Part 2: Global benchmarking
复制标题
CLASSIC v1.0：加拿大陆地表面计划 (CLASS) 和加拿大陆地生态系统模型 (CTEM) 的开源社区继承者 - 第 2 部分：全球基准测试
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Libo Wang
中科院分区：
文献类型：
--
作者：
C. Seiler;J. Melton;V. Arora;Libo Wang
抽象的。加拿大陆地表面计划，包括生物地质化学循环(CLASSIC)，是一个开源社区模型，旨在解决探索陆地表面在全球气候系统中的作用的研究问题。在这里，我们评估了当强迫使用准观测气象数据时，经典模式再现能量、水和碳循环的效果如何。模型技能分数汇总了模型输出在多个统计指标上与基于观察的参考数据的一致性程度。缺乏共识可能是由于模型、其强制数据和/或参考数据中的缺陷。为了解决强迫中的不确定性，我们评估了基于三个气象数据集的经典运行集合。为了考虑观测的不确定性，我们计算了基准技能分数，这些分数量化了独立参考数据集之间的一致性水平。基准分数表明，在观察中存在不确定性的情况下，模型可能实际达到的评分值。我们的结果表明，与强迫和观测相关的不确定性相当大。例如，在这项研究中评估的19个变量中，有10个变量的偏差符号取决于所使用的强迫和参考数据。基准得分远低于预期，这意味着观测的不确定性很大。模型和基准分值基本相似，表明经典模型在考虑观测不确定性时表现良好。未来的模型开发应解决：(I)北半球温带和青藏高原部分地区的正反照率偏差和由此产生的短波辐射偏差，(Ii)南美洲和非洲潮湿热带地区的非同相季节初级生产力总循环，(Iii)年平均净生态系统交换与立地水平测量缺乏空间相关性，(Iv)对北方森林的部分燃烧面积和相应排放的低估，(V)高纬度土壤有机碳的负偏差，以及(Vi)NH温带部分地区的季节叶面积指数最大值的时间滞后。我们的结果将作为指导和监测未来经典开发的基线。
Abstract. The Canadian Land Surface Scheme Including Biogeochemical Cycles (CLASSIC) is an open-source community model designed to address research questions that explore the role of the land surface in the global climate system. Here, we evaluate how well CLASSIC reproduces the energy, water, and carbon cycle when forced with quasi-observed meteorological data. Model skill scores summarize how well model output agrees with observation-based reference data across multiple statistical metrics. A lack of agreement may be due to deficiencies in the model, its forcing data, and/or reference data. To address uncertainties in the forcing, we evaluate an ensemble of CLASSIC runs that is based on three meteorological data sets. To account for observational uncertainty, we compute benchmark skill scores that quantify the level of agreement among independent reference data sets. The benchmark scores demonstrate what score values a model may realistically achieve given the uncertainties in the observations. Our results show that uncertainties associated with the forcing and observations are considerably large. For instance, for 10 out of 19 variables assessed in this study, the sign of the bias changes depending on what forcing and reference data are used. Benchmark scores are much lower than expected, implying large observational uncertainties. Model and benchmark score values are mostly similar, indicating that CLASSIC performs well when considering observational uncertainty. Future model development should address (i) a positive albedo bias and resulting shortwave radiation bias in parts of the Northern Hemisphere (NH) extratropics and Tibetan Plateau, (ii) an out-of-phase seasonal gross primary productivity cycle in the humid tropics of South America and Africa, (iii) a lacking spatial correlation of annual mean net ecosystem exchange with site-level measurements, (iv) an underestimation of fractional area burned and corresponding emissions in the boreal forests, (v) a negative soil organic carbon bias in high latitudes, and (vi) a time lag in seasonal leaf area index maxima in parts of the NH extratropics. Our results will serve as a baseline for guiding and monitoring future CLASSIC development.