Search

Soils contain a large and dynamic fraction of global terrestrial carbon stocks. The distribution of soil carbon (SC) with depth varies among ecosystems and land uses and is an important factor in calculating SC stocks and their vulnerabilities. Systematic analysis of SC depth distributions across databases of SC profiles has been challenging due to the heterogeneity of soil profile measurements, which vary in depth sampling. Here, we fit over 40,000 SC depth profiles to an exponential decline relationship with depth to determine SC concentration at the top of the mineral soil, minimum SC concentration at depth, and the characteristic “length” of SC concentration decline with depth. Fitting these parameters allowed profile characteristics to be analyzed across a large and heterogeneous dataset. We then assessed the differences in these depth parameters across soil orders and land cover types and between soil profiles with or without a history of tillage, as represented by the presence of an Ap horizon. We found that historically tilled soils had more gradual decreases of SC with depth (greater e-folding depth or Z∗), deeper SC profiles, lower SC concentrations at the top of the mineral soil, and lower total SC stocks integrated to 30 cm. The large database of profiles allowed these results to be confirmed across different land cover types and spatial areas within the Continental United States, providing robust evidence for systematic impacts of historical tillage on SC stocks and depth distributions.

This study investigates and compares soil moisture and hydrology projections of broadly used land models with permafrost processes and highlights the causes and impacts of permafrost zone soil moisture projections. Climate models project warmer temperatures and increases in precipitation (P) which will intensify evapotranspiration (ET) and runoff in land models. However, this study shows that most models project a long-term drying of the surface soil (0-20 cm) for the permafrost region despite increases in the net air-surface water flux (P-ET). Drying is generally explained by infiltration of moisture to deeper soil layers as the active layer deepens or permafrost thaws completely. Although most models agree on drying, the projections vary strongly in magnitude and spatial pattern. Land models tend to agree with decadal runoff trends but underestimate runoff volume when compared to gauge data across the major Arctic river basins, potentially indicating model structural limitations. Coordinated efforts to address the ongoing challenges presented in this study will help reduce uncertainty in our capability to predict the future Arctic hydrological state and associated land-atmosphere biogeochemical processes across spatial and temporal scales.

The enhanced vegetation growth by climate warming plays a pivotal role in amplifying the seasonal cycle of atmospheric CO 2 at northern lands (>50° N) since 1960s. However, the correlation between vegetation growth, temperature and seasonal amplitude of atmospheric CO 2 concentration have become elusive with the slowed increasing trend of vegetation growth and weakened temperature control on CO 2 uptake since late 1990s. Here, based on in situ atmospheric CO 2 concentration records from the Barrow observatory site, we found a slowdown in the increasing trend of the atmospheric CO 2 amplitude from 1990s to mid-2000s. This phenomenon was associated with the paused decrease in the minimum CO 2 concentration ([CO 2 ] min ), which was significantly correlated with the slowdown of vegetation greening and growing-season length extension. We then showed that both the vegetation greenness and growing-season length were positively correlated with spring but not autumn temperature over the northern lands. Furthermore, such asymmetric dependences of vegetation growth upon spring and autumn temperature cannot be captured by the state-of-art terrestrial biosphere models. These findings indicate that the responses of vegetation growth to spring and autumn warming are asymmetric, and highlight the need of improving autumn phenology in the models for predicting seasonal cycle of atmospheric CO 2 concentration.

Realistic projection of future climate-carbon (C) cycle feedbacks requires better understanding and an improved representation of the C cycle in permafrost regions in the current generation of Earth system models. Here we evaluated 10 terrestrial ecosystem models for their estimates of net primary productivity (NPP) and responses to historical climate change in permafrost regions in the Northern Hemisphere. In comparison with the satellite estimate from the Moderate Resolution Imaging Spectroradiometer (MODIS; 246 ± 6 g C m−2 yr−1), most models produced higher NPP (309 ± 12 g C m−2 yr−1) over the permafrost region during 2000–2009. By comparing the simulated gross primary productivity (GPP) with a flux tower-based database, we found that although mean GPP among the models was only overestimated by 10% over 1982–2009, there was a twofold discrepancy among models (380 to 800 g C m−2 yr−1), which mainly resulted from differences in simulated maximum monthly GPP (GPPmax). Most models overestimated C use efficiency (CUE) as compared to observations at both regional and site levels. Further analysis shows that model variability of GPP and CUE are nonlinearly correlated to variability in specific leaf area and the maximum rate of carboxylation by the enzyme Rubisco at 25°C (Vcmax_25), respectively. The models also varied in their sensitivities of NPP, GPP, and CUE to historical changes in climate and atmospheric CO2 concentration. These results indicate that model predictive ability of the C cycle in permafrost regions can be improved by better representation of the processes controlling CUE and GPPmax as well as their sensitivity to climate change.

Accomplishing the objective of the current climate policies will require establishing carbon budget and flux estimates in each region and county of the globe by comparing and reconciling multiple estimates including the observations and the results of top-down atmospheric carbon dioxide (CO2) inversions and bottom-up dynamic global vegetation models. With this in view, this study synthesizes the carbon source/sink due to net ecosystem productivity (NEP), land cover land use change (E LUC), fires and fossil burning (E FIRE) for the South Asia (SA), Southeast Asia (SEA) and South and Southeast Asia (SSEA = SA + SEA) and each country in these regions using the multiple top-down and bottom-up modeling results. The terrestrial net biome productivity (NBP = NEP - E LUC - E FIRE) calculated based on bottom-up models in combination with E FIRE based on GFED4s data show net carbon sinks of 217 ±147, 10 ±55, and 227 ±279 TgC yr-1 for SA, SEA, and SSEA. The top-down models estimated NBP net carbon sinks were 20 ±170, 4 ±90 and 24 ±180 TgC yr-1. In comparison, regional emissions from the combustion of fossil fuels were 495, 275, and 770 TgC yr-1, which are many times higher than the NBP sink estimates, suggesting that the contribution of the fossil fuel emissions to the carbon budget of SSEA results in a significant net carbon source during the 2000s. When considering both NBP and fossil fuel emissions for the individual countries within the regions, Bhutan and Laos were net carbon sinks and rest of the countries were net carbon source during the 2000s. The relative contributions of each of the fluxes (NBP, NEP, E LUC, and E FIRE, fossil fuel emissions) to a nation's net carbon flux varied greatly from country to country, suggesting a heterogeneous dominant carbon fluxes on the country-level throughout SSEA.

We examined the net terrestrial carbon flux to the atmosphere (FTA) simulated by nine models from the TRENDY dynamic global vegetation model project for its seasonal cycle and amplitude trend during 1961-2012. While some models exhibit similar phase and amplitude compared to atmospheric inversions, with spring drawdown and autumn rebound, others tend to rebound early in summer. The model ensemble mean underestimates the magnitude of the seasonal cycle by 40g% compared to atmospheric inversions. Global FTA amplitude increase (19g±g8g%) and its decadal variability from the model ensemble are generally consistent with constraints from surface atmosphere observations. However, models disagree on attribution of this long-term amplitude increase, with factorial experiments attributing 83g±g56g%, ĝ'3g±g74 and 20g±g30g% to rising CO2, climate change and land use/cover change, respectively. Seven out of the nine models suggest that CO2 fertilization is the strongest control - with the notable exception of VEGAS, which attributes approximately equally to the three factors. Generally, all models display an enhanced seasonality over the boreal region in response to high-latitude warming, but a negative climate contribution from part of the Northern Hemisphere temperate region, and the net result is a divergence over climate change effect. Six of the nine models show that land use/cover change amplifies the seasonal cycle of global FTA: some are due to forest regrowth, while others are caused by crop expansion or agricultural intensification, as revealed by their divergent spatial patterns. We also discovered a moderate cross-model correlation between FTA amplitude increase and increase in land carbon sink (R2 Combining double low line g0.61). Our results suggest that models can show similar results in some benchmarks with different underlying mechanisms; therefore, the spatial traits of CO2 fertilization, climate change and land use/cover changes are crucial in determining the right mechanisms in seasonal carbon cycle change as well as mean sink change.

Coordinated experimental design and implementation has become a cornerstone of global climate modelling. Model Intercomparison Projects (MIPs) enable systematic and robust analysis of results across many models, by reducing the influence of ad hoc differences in model set-up or experimental boundary conditions. As it enters its 6th phase, the Coupled Model Intercomparison Project (CMIP6) has grown significantly in scope with the design and documentation of individual simulations delegated to individual climate science communities. The Coupled Climate-Carbon Cycle Model Intercomparison Project (C4MIP) takes responsibility for design, documentation, and analysis of carbon cycle feedbacks and interactions in climate simulations. These feedbacks are potentially large and play a leading-order contribution in determining the atmospheric composition in response to human emissions of CO2 and in the setting of emissions targets to stabilize climate or avoid dangerous climate change. For over a decade, C4MIP has coordinated coupled climate-carbon cycle simulations, and in this paper we describe the C4MIP simulations that will be formally part of CMIP6. While the climate-carbon cycle community has created this experimental design, the simulations also fit within the wider CMIP activity, conform to some common standards including documentation and diagnostic requests, and are designed to complement the CMIP core experiments known as the Diagnostic, Evaluation and Characterization of Klima (DECK). C4MIP has three key strands of scientific motivation and the requested simulations are designed to satisfy their needs: (1) pre-industrial and historical simulations (formally part of the common set of CMIP6 experiments) to enable model evaluation, (2) idealized coupled and partially coupled simulations with 1% per year increases in CO2 to enable diagnosis of feedback strength and its components, (3) future scenario simulations to project how the Earth system will respond to anthropogenic activity over the 21st century and beyond. This paper documents in detail these simulations, explains their rationale and planned analysis, and describes how to set up and run the simulations. Particular attention is paid to boundary conditions, input data, and requested output diagnostics. It is important that modelling groups participating in C4MIP adhere as closely as possible to this experimental design.

A realistic simulation of snow cover and its thermal properties are important for accurate modelling of permafrost. We analyse simulated relationships between air and near-surface (20 cm) soil temperatures in the Northern Hemisphere permafrost region during winter, with a particular focus on snow insulation effects in nine land surface models, and compare them with observations from 268 Russian stations. There are large cross-model differences in the simulated differences between near-surface soil and air temperatures (ΔT; 3 to 14 °C), in the sensitivity of soil-to-air temperature (0.13 to 0.96 °C °C-1), and in the relationship between ΔT and snow depth. The observed relationship between ΔT and snow depth can be used as a metric to evaluate the effects of each model's representation of snow insulation, hence guide improvements to the model's conceptual structure and process parameterisations. Models with better performance apply multilayer snow schemes and consider complex snow processes. Some models show poor performance in representing snow insulation due to underestimation of snow depth and/or overestimation of snow conductivity. Generally, models identified as most acceptable with respect to snow insulation simulate reasonable areas of near-surface permafrost (13.19 to 15.77 million km2). However, there is not a simple relationship between the sophistication of the snow insulation in the acceptable models and the simulated area of Northern Hemisphere near-surface permafrost, because several other factors, such as soil depth used in the models, the treatment of soil organic matter content, hydrology and vegetation cover, also affect the simulated permafrost distribution.

Global environmental change is rapidly altering the dynamics of terrestrial vegetation, with consequences for the functioning of the Earth system and provision of ecosystem services. Yet how global vegetation is responding to the changing environment is not well established. Here we use three long-term satellite leaf area index (LAI) records and ten global ecosystem models to investigate four key drivers of LAI trends during 1982-2009. We show a persistent and widespread increase of growing season integrated LAI (greening) over 25% to 50% of the global vegetated area, whereas less than 4% of the globe shows decreasing LAI (browning). Factorial simulations with multiple global ecosystem models suggest that CO2 fertilization effects explain 70% of the observed greening trend, followed by nitrogen deposition (9%), climate change (8%) and land cover change (LCC) (4%). CO2 fertilization effects explain most of the greening trends in the tropics, whereas climate change resulted in greening of the high latitudes and the Tibetan Plateau. LCC contributed most to the regional greening observed in southeast China and the eastern United States. The regional effects of unexplained factors suggest that the next generation of ecosystem models will need to explore the impacts of forest demography, differences in regional management intensities for cropland and pastures, and other emerging productivity constraints such as phosphorus availability.

We present a synthesis of the land-atmosphere carbon flux from land use and land cover change (LULCC) in Asia using multiple data sources and paying particular attention to deforestation and forest regrowth fluxes. The data sources are quasi-independent and include the U.N. Food and Agriculture Organization-Forest Resource Assessment (FAO-FRA 2015; country-level inventory estimates), the Emission Database for Global Atmospheric Research (EDGARv4.3), the 'Houghton' bookkeeping model that incorporates FAO-FRA data, an ensemble of 8 state-of-the-art Dynamic Global Vegetation Models (DGVM), and 2 recently published independent studies using primarily remote sensing techniques. The estimates are aggregated spatially to Southeast, East, and South Asia and temporally for three decades, 1980-1989, 1990-1999 and 2000-2009. Since 1980, net carbon emissions from LULCC in Asia were responsible for 20%-40% of global LULCC emissions, with emissions from Southeast Asia alone accounting for 15%-25% of global LULCC emissions during the same period. In the 2000s and for all Asia, three estimates (FAO-FRA, DGVM, Houghton) were in agreement of a net source of carbon to the atmosphere, with mean estimates ranging between 0.24 to 0.41 Pg C yr-1, whereas EDGARv4.3 suggested a net carbon sink of -0.17 Pg C yr-1. Three of 4 estimates suggest that LULCC carbon emissions declined by at least 34% in the preceding decade (1990-2000). Spread in the estimates is due to the inclusion of different flux components and their treatments, showing the importance to include emissions from carbon rich peatlands and land management, such as shifting cultivation and wood harvesting, which appear to be consistently underreported.

A significant portion of the large amount of carbon (C) currently stored in soils of the permafrost region in the Northern Hemisphere has the potential to be emitted as the greenhouse gases CO2 and CH4 under a warmer climate. In this study we evaluated the variability in the sensitivity of permafrost and C in recent decades among land surface model simulations over the permafrost region between 1960 and 2009. The 15 model simulations all predict a loss of near-surface permafrost (within 3 m) area over the region, but there are large differences in the magnitude of the simulated rates of loss among the models (0.2 to 58.8 × 103 km2 yr−1). Sensitivity simulations indicated that changes in air temperature largely explained changes in permafrost area, although interactions among changes in other environmental variables also played a role. All of the models indicate that both vegetation and soil C storage together have increased by 156 to 954 Tg C yr−1 between 1960 and 2009 over the permafrost region even though model analyses indicate that warming alone would decrease soil C storage. Increases in gross primary production (GPP) largely explain the simulated increases in vegetation and soil C. The sensitivity of GPP to increases in atmospheric CO2 was the dominant cause of increases in GPP across the models, but comparison of simulated GPP trends across the 1982–2009 period with that of a global GPP data set indicates that all of the models overestimate the trend in GPP. Disturbance also appears to be an important factor affecting C storage, as models that consider disturbance had lower increases in C storage than models that did not consider disturbance. To improve the modeling of C in the permafrost region, there is the need for the modeling community to standardize structural representation of permafrost and carbon dynamics among models that are used to evaluate the permafrost C feedback and for the modeling and observational communities to jointly develop data sets and methodologies to more effectively benchmark models.

Global temperature rise and extremes accompanying drought threaten forests and their associated climatic feedbacks. Our ability to accurately simulate drought-induced forest impacts remains highly uncertain in part owing to our failure to integrate physiological measurements, regional-scale models, and dynamic global vegetation models (DGVMs). Here we show consistent predictions of widespread mortality of needleleaf evergreen trees (NET) within Southwest USA by 2100 using state-of-the-art models evaluated against empirical data sets. Experimentally, dominant Southwest USA NET species died when they fell below predawn water potential (pd) thresholds (April-August mean) beyond which photosynthesis, hydraulic and stomatal conductance, and carbohydrate availability approached zero. The evaluated regional models accurately predicted NET pd, and 91% of predictions (10 out of 11) exceeded mortality thresholds within the twenty-first century due to temperature rise. The independent DGVMs predicted ≥50% loss of Northern Hemisphere NET by 2100, consistent with the NET findings for Southwest USA. Notably, the global models underestimated future mortality within Southwest USA, highlighting that predictions of future mortality within global models may be underestimates. Taken together, the validated regional predictions and the global simulations predict widespread conifer loss in coming decades under projected global warming.

We perform a land-surface model intercomparison to investigate how the simulation of permafrost area on the Tibetan Plateau (TP) varies among six modern stand-alone land-surface models (CLM4.5, CoLM, ISBA, JULES, LPJ-GUESS, UVic). We also examine the variability in simulated permafrost area and distribution introduced by five different methods of diagnosing permafrost (from modeled monthly ground temperature, mean annual ground and air temperatures, air and surface frost indexes). There is good agreement (99 to 135 × 104km2) between the two diagnostic methods based on air temperature which are also consistent with the observation-based estimate of actual permafrost area (101 ×104km2). However the uncertainty (1 to 128 × 104km2) using the three methods that require simulation of ground temperature is much greater. Moreover simulated permafrost distribution on the TP is generally only fair to poor for these three methods (diagnosis of permafrost from monthly, and mean annual ground temperature, and surface frost index), while permafrost distribution using air-temperature-based methods is generally good. Model evaluation at field sites highlights specific problems in process simulations likely related to soil texture specification, vegetation types and snow cover. Models are particularly poor at simulating permafrost distribution using the definition that soil temperature remains at or below 0°C for 24 consecutive months, which requires reliable simulation of both mean annual ground temperatures and seasonal cycle, and hence is relatively demanding. Although models can produce better permafrost maps using mean annual ground temperature and surface frost index, analysis of simulated soil temperature profiles reveals substantial biases. The current generation of land-surface models need to reduce biases in simulated soil temperature profiles before reliable contemporary permafrost maps and predictions of changes in future permafrost distribution can be made for the Tibetan Plateau.

Soil temperature (Ts/change is a key indicator of the dynamics of permafrost. On seasonal and interannual timescales, the variability of Ts determines the activelayer depth, which regulates hydrological soil properties and biogeochemical processes. On the multi-decadal scale, increasing Ts not only drives permafrost thaw/retreat but can also trigger and accelerate the decomposition of soil organic carbon. The magnitude of permafrost carbon feedbacks is thus closely linked to the rate of change of soil thermal regimes. In this study, we used nine process-based ecosystem models with permafrost processes, all forced by different observation-based climate forcing during the period 1960-2000, to characterize the warming rate of Ts in permafrost regions. There is a large spread of Ts trends at 20 cm depth across the models, with trend values ranging from 0.010 ± 0.003 to 0.031 ± 0.005 °C yr-1. Most models show smaller increase in Ts with increasing depth. Air temperature (Ta/and longwave downward radiation (LWDR) are the main drivers of Ts trends, but their relative contributions differ amongst the models. Different trends of LWDR used in the forcing of models can explain 61 % of their differences in Ts trends, while trends of Ta only explain 5 % of the differences in Ts trends. Uncertain climate forcing contributes a larger uncertainty in Ts trends (0.021 ± 0.008 °C yr-1, mean ± standard deviation) than the uncertainty of model structure (0.012 ± 0.001 °C yr-1/, diagnosed from the range of response between different models, normalized to the same forcing. In addition, the loss rate of near-surface permafrost area, defined as total area where the maximum seasonal active-layer thickness (ALT) is less than 3 m loss rate, is found to be significantly correlated with the magnitude of the trends of Ts at 1 m depth across the models (R D-0:85, P = 0:003), but not with the initial total nearsurface permafrost area (R =-0:30, P = 0:438). The sensitivity of the total boreal near-surface permafrost area to Ts at 1 m is estimated to be of-2.80 ± 0.67 million km2 °C-1. Finally, by using two long-term LWDR data sets and relationships between trends of LWDR and Ts across models, we infer an observation-constrained total boreal near-surface permafrost area decrease comprising between 39 ± 14 × 103 and 75 ± 14 × 103 km2 yr-1 from 1960 to 2000. This corresponds to 9-18 % degradation of the current permafrost area.

Soil carbon (C) is a critical component of Earth system models (ESMs), and its diverse representations are a major source of the large spread across models in the terrestrial C sink from the third to fifth assessment reports of the Intergovernmental Panel on Climate Change (IPCC). Improving soil C projections is of a high priority for Earth system modeling in the future IPCC and other assessments. To achieve this goal, we suggest that (1) model structures should reflect real-world processes, (2) parameters should be calibrated to match model outputs with observations, and (3) external forcing variables should accurately prescribe the environmental conditions that soils experience. First, most soil C cycle models simulate C input from litter production and C release through decomposition. The latter process has traditionally been represented by first-order decay functions, regulated primarily by temperature, moisture, litter quality, and soil texture. While this formulation well captures macroscopic soil organic C (SOC) dynamics, better understanding is needed of their underlying mechanisms as related to microbial processes, depth-dependent environmental controls, and other processes that strongly affect soil C dynamics. Second, incomplete use of observations in model parameterization is a major cause of bias in soil C projections from ESMs. Optimal parameter calibration with both pool- and flux-based data sets through data assimilation is among the highest priorities for near-term research to reduce biases among ESMs. Third, external variables are represented inconsistently among ESMs, leading to differences in modeled soil C dynamics. We recommend the implementation of traceability analyses to identify how external variables and model parameterizations influence SOC dynamics in different ESMs. Overall, projections of the terrestrial C sink can be substantially improved when reliable data sets are available to select the most representative model structure, constrain parameters, and prescribe forcing fields.

Accurate assessment of anthropogenic carbon dioxide (CO2) emissions and their redistribution among the atmosphere, ocean, and terrestrial biosphere is important to better understand the global carbon cycle, support the development of climate policies, and project future climate change. Here we describe data sets and a methodology to quantify all major components of the global carbon budget, including their uncertainties, based on the combination of a range of data, algorithms, statistics, and model estimates and their interpretation by a broad scientific community. We discuss changes compared to previous estimates as well as consistency within and among components, alongside methodology and data limitations. CO2 emissions from fossil fuels and industry (EFF) are based on energy statistics and cement production data, while emissions from land-use change (ELUC), mainly deforestation, are based on combined evidence from land-cover-change data, fire activity associated with deforestation, and models. The global atmospheric CO2 concentration is measured directly and its rate of growth (GATM) is computed from the annual changes in concentration. The mean ocean CO2 sink (SOCEAN) is based on observations from the 1990s, while the annual anomalies and trends are estimated with ocean models. The variability in SOCEAN is evaluated with data products based on surveys of ocean CO2 measurements. The global residual terrestrial CO2 sink (SLAND) is estimated by the difference of the other terms of the global carbon budget and compared to results of independent dynamic global vegetation models forced by observed climate, CO2, and land-cover change (some including nitrogen-carbon interactions). We compare the mean land and ocean fluxes and their variability to estimates from three atmospheric inverse methods for three broad latitude bands. All uncertainties are reported as ±1σ, reflecting the current capacity to characterise the annual estimates of each component of the global carbon budget. For the last decade available (2005-2014), EFF was 9.0 ± 0.5 GtC yrg'1, ELUC was 0.9 ± 0.5 GtC yrg'1, GATM was 4.4 ± 0.1 GtC yrg'1, SOCEAN was 2.6 ± 0.5 GtC yrg'1, and SLAND was 3.0 ± 0.8 GtC yrg'1. For the year 2014 alone, EFF grew to 9.8 ± 0.5 GtC yrg'1, 0.6 % above 2013, continuing the growth trend in these emissions, albeit at a slower rate compared to the average growth of 2.2 % yrg'1 that took place during 2005-2014. Also, for 2014, ELUC was 1.1 ± 0.5 GtC yrg'1, GATM was 3.9 ± 0.2 GtC yrg'1, SOCEAN was 2.9 ± 0.5 GtC yrg'1, and SLAND was 4.1 ± 0.9 GtC yrg'1. GATM was lower in 2014 compared to the past decade (2005-2014), reflecting a larger SLAND for that year. The global atmospheric CO2 concentration reached 397.15 ± 0.10 ppm averaged over 2014. For 2015, preliminary data indicate that the growth in EFF will be near or slightly below zero, with a projection of g'0.6 [range of g'1.6 to +0.5] %, based on national emissions projections for China and the USA, and projections of gross domestic product corrected for recent changes in the carbon intensity of the global economy for the rest of the world. From this projection of EFF and assumed constant ELUC for 2015, cumulative emissions of CO2 will reach about 555 ± 55 GtC (2035 ± 205 GtCO2) for 1870-2015, about 75 % from EFF and 25 % from ELUC. This living data update documents changes in the methods and data sets used in this new carbon budget compared with previous publications of this data set (Le Quéré et al., 2015, 2014, 2013). All observations presented here can be downloaded from the Carbon Dioxide Information Analysis Center (doi:10.3334/CDIAC/GCP-2015).

Land models, which have been developed by the modeling community in the past few decades to predict future states of ecosystems and climate, have to be critically evaluated for their performance skills of simulating ecosystem responses and feedback to climate change. Benchmarking is an emerging procedure to measure performance of models against a set of defined standards. This paper proposes a benchmarking framework for evaluation of land model performances and, meanwhile, highlights major challenges at this infant stage of benchmark analysis. The framework includes (1) targeted aspects of model performance to be evaluated, (2) a set of benchmarks as defined references to test model performance, (3) metrics to measure and compare performance skills among models so as to identify model strengths and deficiencies, and (4) model improvement. Land models are required to simulate exchange of water, energy, carbon and sometimes other trace gases between the atmosphere and land surface, and should be evaluated for their simulations of biophysical processes, biogeochemical cycles, and vegetation dynamics in response to climate change across broad temporal and spatial scales. Thus, one major challenge is to select and define a limited number of benchmarks to effectively evaluate land model performance. The second challenge is to develop metrics of measuring mismatches between models and benchmarks. The metrics may include (1) a priori thresholds of acceptable model performance and (2) a scoring system to combine data–model mismatches for various processes at different temporal and spatial scales. The benchmark analyses should identify clues of weak model performance to guide future development, thus enabling improved predictions of future states of ecosystems and climate. The near-future research effort should be on development of a set of widely acceptable benchmarks that can be used to objectively, effectively, and reliably evaluate fundamental properties of land models to improve their prediction performance skills.