5th iLEAPS Science Conference Abstracts - E3

Abstracts – Session E3

Confronting land models with data for assessment and verification

E301 ORAL-0399: Applying ILAMB to assess model development and forcing uncertainty in CLM

David Lawrence1, Keith Oleson1, Rosie Fisher1, Forrest Hoffman2, Nathan Collier2, Sean Swenson1, James Randerson3, Mingquan Mu3

1Climate and Global Dynamics Laboratory, National Center for Atmospheric Research, Boulder, Colorado, The United States of America 2Environmental Sciences Division and Climate Change Science Institute, Oak Ridge National Laboratory, Oak Ridge, The United States of America 3University of California, Irvine, Irvine, The United States of America

The International Land Model Benchmarking (ILAMB) project is a model-data intercomparison and integration project designed to assess and help improve land models. The current package includes assessment of more than 25 land variables across more than 60 datasets and includes a broad range of metrics including RMSE, mean error, spatial distributions, interannual variability, and functional relationships. Here, we will describe recent progress within the ILAMB project and will present its application for assessment of several generations and configurations of the Community Land Model (CLM4, CLM4.5, and CLM5 with and without interactive vegetation). We will also assess the impact of model forcing uncertainty in the ILAMB context by additionally assessing the impact of forcing with the GSWP3v1 and CRUNCEP historical meteorological datasets. Analysis of preliminary simulations suggests that despite considerable uncertainty related to land model forcing, improvements across a range of modeled and observed variables can be traced across the generations of CLM. Strengths and weaknesses and persistent biases across generations will also be presented.

E302 - ORAL-0079: Evaluating and benchmarking land surface models

Heather Rumbold1, Graham Weedon2, Martin Best1, Sujay Kumar3

1Met Office, Exeter, United Kingdom 2Met Office, Wallingford, United Kingdom 3NASA/GSFC , Greenbelt, The United States of America

Benchmarking of land surface models (LSMs) involves adopting widely agreed standards for judging performance. Unlike evaluation or validation, benchmarking requires comparison of outputs with pre-defined targets or thresholds; allowing meaningful intercomparisons of independent models.
 
Observations relevant to the assessment of modelled latent and sensible heat are available at a wide range of spatial scales (from the footprint area of flux towers to the large pixels of GRACE satellite products) and time steps (e.g. sub-daily to monthly). Thus, an evaluation system, as part of benchmarking system, should be flexible enough to process model output and observations at whatever time steps and spatial resolutions they are available. NASA’s LVT (Land Verification Toolkit, Kumar et al., 2012 Geosci. Model Dev.), developed as part of the LIS suite, has this flexibility. It also provides a wide range of analytical metrics with in-built assessment of the uncertainties of the metrics (e.g. 95% CIs).
 
In this presentation we will describe the development of a new land surface benchmarking system based upon LVT. Initially the system will consider fluxes of momentum, heat, moisture and carbon from point observations, but will be further developed to include satellite data for state variables such as Land Surface Temperature and soil moisture. The system will build upon early work on benchmarking identified in the PLUMBER experiment (Best et al., 2015 J. Hydrometeorol.) to identify suitable metrics that can be equally applied for both stand alone and coupled simulations. It is anticipated that the basis of this system could be expanded to cover all aspects of LSMs with the eventual aim of establishing an international standard LSM benchmarking suite.

E303 - ORAL-0131: Sensitivity of global ecosystems to climate anomalies in observations and models

Diego Miralles, Matthias Demuzere, Christina Papagiannopoulou, Willem Waegeman, Stijn Decubber, Niko Verhoest, Wouter Dorigo

Vegetation is a central player in the climate system, constraining atmospheric conditions through a series of feedbacks. This fundamental role highlights the importance of understanding regional drivers of ecological sensitivity and the response of vegetation to climatic changes. While nutrient availability and short-term disturbances can be crucial for vegetation at various spatiotemporal scales, natural vegetation dynamics are overall driven by climate. At monthly scales, the interactions between vegetation and climate become complex: some vegetation types react preferentially to specific climatic changes, with different levels of intensity, resilience and lagged response. For our current Earth System Models (ESMs) being able to capture this complexity is crucial but extremely challenging. This adds uncertainty to our projections of future climate and the fate of global ecosystems. Here, following a Granger causality framework based on a random forest predictive model, we exploit the current wealth of satellite data records to uncover the main climatic drivers of monthly vegetation variability globally. Results based on three decades of satellite data indicate that water availability is the most dominant factor driving vegetation in over 60% of the vegetated land. This overall dependency of ecosystems on water availability is larger than previously reported, partly owed to the ability of our machine-learning framework to disentangle the co-linearites between climatic drivers, and to quantify non-linear impacts of climate on vegetation. Our observation-based results are then used to benchmark ESMs on their representation of vegetation sensitivity to climate and climatic extremes. Our findings indicate that the sensitivity of vegetation to climatic anomalies is ill-reproduced by some widely-used ESMs.

E304 ORAL-0054: Ecosystem-scale light-use efficiency as a metric for benchmarking CMIP5 models

Rebecca Thursa Thomas1, Heather Graven1

1Imperial College, London, United Kingdom

Terrestrial ecosystem models show differing patterns of gross primary production (GPP) and its change over time. To better understand the differences in GPP between models and introduce a novel benchmark for models, we analyse how CMIP5 models represent the mean state of, and changes in, ecosystem light-use efficiency. We define ecosystem light-use efficiency (eLUE) as eLUE = GPP/APAR, where APAR is absorbed photosynthetically active radiation, and GPP and APAR are integrated over the growing season using monthly data on the model’s native grid. eLUE is a simple way of separating structural and physiological changes in vegetation and thus provides a novel perspective on current model skill and future model development. Mean observed eLUE was calculated for 1982-2008 using GPP-MTE, fAPAR3g and PAR from WFDEI-WATCH. We found that CMIP5 models were not able to capture mean eLUE overall, but many did represent the particularly high or low eLUE in key areas such as south east Asia and Australia. Models simulate an increase in eLUE over the last few decades, particularly in the northern hemisphere, which has generally dominated the increase in modeled GPP over this period, rather than structural changes in vegetation (greening). Modelled increases in net biome production were well correlated to increases in eLUE between models, suggesting that eLUE is driving the increase in modelled land-atmosphere carbon flux. We suggest that the observation based estimate of mean eLUE that we have produced should be used in future model benchmarking exercises to further understand modelled GPP and vegetation processes.

E305 - ORAL-0118: Assessing the sequential assimilation of satellite-derived vegetation and soil moisture products using independent   

observations databases

Clement Albergel1, Simon Munier1, Delphine Leroux1, Hélène Dewaele1, David Fairbairn2, Alina Barbu1, Catherine Meurey1, Jean-Christophe Calvet1

1CNRM UMR 3589, Météo-France/CNRS, Toulouse, France 2ECMWF, Reading, United Kingdom

The increase in the occurrence of extreme weather events (droughts, floods, heat waves) in connection with global warming is a proven fact. The latest IPCC (Intergovernmental Panel on Climate Change) simulations indicate that occurrence of droughts and warm spells are likely to increase. Observing and simulating the response of land biophysical variables to extreme events is a major scientific challenge in relation to the adaptation to climate change. The modeling of terrestrial variables can be improved through the dynamical integration of observations. Remote sensing observations are particularly useful in this context as they are now unrestrictedly available at a global scale. Many satellite-derived products relevant to the hydrological and vegetation cycles are already available. Assimilating them into land surface models (LSMs) permits their integration in the land surface processes representation in a consistent way.
 
CNRM (Centre National de Recherches Météorologiques) has developed a global Land Data Assimilation system: LDAS-Monde. LDAS-Monde is able to ingest information from satellite-derived Surface Soil Moisture (SSM) and Leaf Area Index (LAI) observations to constrain the ISBA (Interactions between Soil, Biosphere, and Atmosphere) LSM coupled with the CNRM version of the Total Runoff Integrating Pathways (ISBA-CTRIP) continental hydrological system. In this study, LDAS-Monde is tested over various domain (United States of America, Austral Africa and Australia), over 2007-2016 to increase monitoring accuracy for land surface variables. Prior to their assimilation, satellite derived SSM and LAI are compared to their model equivalent constituting a first LSM evaluations. Independent observations (from insitu measurements of soil moisture, river discharges to evapotranspiration and gross primary production estimations) are then used to assess the impact of assimilating satellite derived SSM and LAI within ISBA LSM.

E306 - ORAL-0322: Modelled and observed surface soil moisture spatio-temporal dynamics in a land-atmosphere hotspot

Romina Ruscica1, Jan Polcher2, Maria Piles3, Mercedes Salvia4, Anna Sörensson1, Haydee Karszenbaum4

1 Centro de Investigaciones del Mar y la Atmosfera (CIMA/CONICET-UBA). Instituto Franco-Argentino para el Estudio del Clima y sus Impactos (UMI IFAECI/CNRS-CONICET-UBA), Ciudad de Buenos Aires, Argentina 2Laboratoire de Météorologie Dynamique du CNRS/IPSL Ecole Polytechnique, Paris, France 3Image Processing Lab (IPL) Universitat de València, Valencia, Spain 4IAFE-CONICET-UBA , Ciudad de Buenos Aires, Argentina

Knowledge of surface soil moisture (SSM) spatio-temporal dynamics is essential for many practical applications such as weather forecasting, floods/drought monitoring and water resource management. SSM is particularly relevant over climate transition regions such as south-eastern South America (SESA), a recognized land-atmosphere interaction hotspot by studies using climate models and more recently remote sensing products (RSPs).
 
SESA has the largest population density of the continent and is the most productive region in terms of agriculture, livestock and industry. It comprises the low and flat Pampas plains where the most severe subtropical storms of the globe are developed, making SESA an interesting region for studying SSM.
A novel framework for studying SSM dynamics over the SESA hotspot is presented. SSM dry-downs during non-rainy days after precipitation events are studied on scales of the landscape. This is an essential knowledge to better understand how the surface will respond to changes in the characteristics of rainfall. The dry-down is critical for soil moisture stress of plants, surface Bowen ratio, surface warming and atmospheric response.
 
The ORCHIDEE land surface model and the recent version of SMOS RSP (v.620) were chosen since they are particularly suited for this study. ORCHIDEE provides a high vertical resolution of the soil surface, making it suitable for comparison with RSPs. SMOS uses L-band, which is suitable for densely vegetated areas like SESA and the v.620 includes a new parameterization for forested areas that reduces uncertainties and improves the data quality filtering.
ORCHIDEE and SMOS daily data are employed at 25 km for summers in 2010-2014. Results are analyzed in terms of soil and land cover characteristics, sampling frequency, observational uncertainties and also compared to other metrics.

E307 - ORAL-0240: Evaluating modelled large-scale soil moisture limited evaporation regimes using satellite land surface temperature

Phil Harris1, 2, Sebastien Garrigues3, Stefan Hagemann4, Christopher Taylor1, 2, Belen Gallego-Elvira1

1Centre for Ecology & Hydrology, Wallingford, United Kingdom 2National Centre for Earth Observation, Wallingford, United Kingdom 3INRA, Avignon, France 4Max Planck Institute for Meteorology, Hamburg, Germany

Soil moisture availability exerts a strong control over land evaporation in many regions. However, global climate models (GCMs) disagree on when and where evaporation is limited by soil moisture. Evaluation of the relevant modelled processes has suffered from a lack of reliable, global observations of land evaporation at the GCM grid box scale. Satellite observations of land surface temperature (LST) offer spatially extensive but indirect information about the surface energy partition and soil moisture limitation on evaporation. Specifically, as soil moisture decreases during rain-free dry spells, evaporation may become limited leading to increases in LST and sensible heat flux.
 
Offline land surface model (LSM) simulations offer an inexpensive way to evaluate the surface processes of GCMs, and have the benefit that multiple models can be compared on a common grid and using unbiased forcing. Here, we assess global simulations of several LSMs (e.g., JULES, ISBA, JSBACH) driven by the WATCH Forcing Data-ERA Interim. We use MODIS Terra and Aqua observations of LST for 2000 to 2013 and aggregated from 1 km to the 0.5° WFDEI grid to evaluate composite changes in the surface energy partition during dry spells lasting 10 days or longer. This temperature-based approach can diagnose the typical strength of short term changes in surface heat fluxes and, by extension, changes in soil moisture limitation on evaporation. This offline analysis also avoids some of the difficulties in analysing free-running simulations in which land and atmosphere are coupled and, as such, it provides a flexible intermediate step in the assessment of surface processes in GCMs.

E308 ORAL-0271: Evaluation of the land surface schemes in the Conformal Cubic Atmospheric Model over Africa

Azwitamisi Eric Mudau1, 2, George Djolov1, Francois Engelbrecht2, Marcus Thatcher3, Gregor Feig4

1University of Pretoria, Department of Geography, Geoinformatics and Meteorology, Pretoria, South Africa 2Council for Scientific and Industrial Research, Natural Resources and Environment, Pretoria, South Africa 3Commonwealth Scientific and Industrial Research Organisation, Aspendale, Australia 4Council for Scientific and Industrial Research, Pretoria, South Africa

The land surface plays a major role in numerical weather prediction (NWP) models, and climate models, by providing the lower boundary to the atmosphere. It also significantly affects the land surface processes between the land surface and lower atmosphere. The land surface processes strongly influence the local, regional and global scale climate on a wide range of time scales. The two land surface schemes, namely the Mk3 land surface model and the Community Atmosphere Biosphere Land Exchange (CABLE) model used in the Conformal Cubic Atmospheric Model (CCAM) have not been well-tested and validated over Africa. CCAM is is being used as the atmospheric component of the Variable-resolution Earth System Model (VRESM) that is currently being developed by the CSIR. In this study we evaluated the simulated sensible and latent heat flux by CCAM coupled to either the MK3 land surface model or the CABLE model against FLUXNET: MTE (Multi-Tree Ensemble) and Water, Energy, and Carbon with Artificial Neural Networks (WECANN) datasets. Furthermore we evaluated the simulated rainfall and temperature against both station data and gridded observational dataset (CRU).
 
All the simulations described have been performed on the Lengau cluster of the CHPC, through the resources allocated to the VRESM model development project.

E309 ORAL-0065: Uncertainties in simulated evapotranspiration from land surface models over a 14-yearMediterranean crop succession

Sebastien Garrigues1, Anne Verhoef2, Pier-Luigi Vidale2, Albert Olioso1, Aaron Boone3, Bertrand Decharme3, Clement Albergel3, Jean-Christophe Calvet3, Sophie Moulin1

1INRA, Avignon, France 2University of Reading, Reading, United Kingdom 3CNRM, Meteo France, Toulouse, France

Evapotranspiration (ET) has been recognized as one of the most uncertain terms in the surface water balance simulated by Land Surface Models (LSM), particularly for Mediterranean regions. This study aims at assessing multi-year simulations of ET from the ISBA-A-gs and the JULES LSMs over a 14-year Mediterranean crop succession. We investigate the uncertainties related to:
  • climate, vegetation and irrigation drivers,
  • soil water transfer parametrization,
  • type of LSM.
The main outcomes are:
  • Errors in the soil hydraulic parameters and the lack of irrigation in the simulation have the largest influence on ET compared to uncertainties in climate data sets and LAI climatology. Among climate variables, the errors in yearly ET are mainly related to the errors in yearly rainfall.
  • Errors in the soil parameters derived from the pedotransfer functions lead to underestimation of the water content available for the crop and the soil hydraulic diffusivity which results in a large underestimation of ET (1300 mm over 14 years).
  • Soil evaporation represents 70% of cumulative evapotranspiration over 14 years of crop succession which explains the high sensitivity of simulated evapotranspiration to uncertainties in the soil evaporation parameters.
  • The spatiotemporal uncertainties in the soil parameters generate smaller uncertainties in ET when it is simulated with the multi-layer soil diffusion scheme (374 mm over 12 years) than when it is simulated using the force-restore scheme (962 mm over 12 years).
  • The departure between the ISBA and JULES simulations highlight other sources of uncertainties related to the representation of the plant water stress, for example.
The main challenges for LSM over Mediterranean cropland concern the representation of:
  • the spatial distribution of the soil hydraulic parameters,
  • the variability of irrigation practices,
  • the spatiotemporal variability of rainfall.

E310 ORAL-0312: How can we represent the radiation balance of sparse forests in land surface models?

Rachael Turton1, 2, Eleanor Blyth1, Richard Essery2

1CEH, Wallingford, United Kingdom 2School of GeoSciences, University of Edinburgh, Edinburgh, United Kingdom

Representing the three dimensional nature of sparse forests within a land surface model represents a significant challenge. At present high latitude land surface processes and the surface-climate feedbacks are poorly represented in hydrological and climate models. This study proposes a solution to representing the radiation balance of sparse deciduous canopies in sub-Arctic Sweden for the land surface model JULES (Joint UK Land Environment Simulator) used in the UK Hadley Centre GCM suite.
 
In early spring, the sparse leafless canopies at high latitudes appear dense and impenetrable due to the low solar elevation. Long shadows are cast across the surface of the seasonal snow pack. The sparse canopy intercepts the shortwave radiation and reduces the shortwave radiation penetration down to the snow surface. Yet this incident shortwave radiation on the surface of the trees, also acts to increase the below canopy longwave radiation to the snow surface. As spring progresses and the solar elevation changes the time and duration which the shortwave radiation can penetrate down through the canopy to the snow surface. However, shortwave radiation penetration is spatially and temporally variable due to the heterogeneous canopy cover.
 
Measurements of canopy cover, shortwave and longwave radiation are used here to develop and parameterise a shaded gap tile, which accounts for the temporal variability of the radiation balance. This novel approach improves the land-surface snow interactions of JULES and improves the timing of modelled snow melt with respect to observations.

E311 POSTER-0072: Evaluation of global fire models within the Fire Model Intercomparison Project (FireMIP)

Stijn Hantson1, Douglas Kelley, Almut Arneth, Sandy P. Harrison, Sam S. Rabin, Dominique Bachelet, Matthew Forrest, Silvia Kloster, Gitta Lasslop, Fang Li, Stephane Mangeon, Joe Melton, Chao Yue, Tim Sheehan

1Karlsruhe Institute of Technology, Institute of Meteorology and Climate research, Atmospheric Environmental Research, Garmisch-Partenkirchen, Germany

A large number of global fire-vegetation models have been developed and widely used to simulate global fire occurrence under past and future climate and land use scenarios. Important discrepancies exist between model outcomes, resulting in a large uncertainty of past and future fire occurrence. This is in part because no systematic assessment of these fire-vegetation models has been performed so far. In the Fire Model Intercomparison Project (FireMIP), we are examining the ability of global fire models to simulate temporal and spatial patterns of the global fire regime.
 
Nine global fire-vegetation models of different complexity performed a set of transient, 20th century runs using standardized climate, atmospheric CO2 concentration, lightning occurrence, population density and land cover data. Here we outline the benchmarking framework used to evaluate the participating models. Benchmarking scores of each model will be presented, with special attention to fire relevant variables. Issues encountered with the current benchmarking scheme will be discussed. The results from the benchmarking exercise will improve our understanding of key processes and biases in global fire models and will help us improve our ability to project changes in fire occurrence and to guide future development of global fire-vegetation models.

E312 -POSTER-0161: Observing and Simulating Spatial Variations of Forest C Fluxes and Stocks in Complex Terrain

Yuting He1, Kenneth Davis1, Yuning Shi2, David Eissenstat2, Jason Kaye2

1Department of Meteorology and Atmospheric Science, the Pennsylvania State University, State College, The United States of America 2Department of Ecosystem Science and Management, the Pennsylvania State University, State College, The United States of America

Terrestrial carbon (C) cycle remains the least constrained component in the global C cycle, partly due to the difficulty to quantify C sources and sinks in complex terrain. In this study, we used observations at Shale Hills Critical Zone Observatory and a biogeochemistry model - Biome-BGC to examine the spatial distribution of C stocks and fluxes in a first-order watershed. We fed the model with observed soil moisture and soil temperature to reduce the uncertainties in simulating water and energy cycle. With only three parameters (whole-plant mortality, N input and maximum decomposition rates of soil and litter C pool) constrained by observations, the model could represent the average C pools and fluxes in the watershed. We then applied this tuned model to six sites along the topography, and the model was able to produce the general spatial patterns of C pools in the watershed, with higher biomass and soil C in the valley and lower on the ridgetop, even though the model underestimated the spatial contrast along the topography. We also examined the effects of four environmental factors on the spatial distribution of C pools. These four environmental factors are soil moisture, soil temperature, N availability and solar radiation. Results from this part of the study highlighted the importance of accurate hydrological simulations to ecosystem modelling. Following this study, we also analyzed the sensitivity of Biome-BGC parameters to C stocks and fluxes in complex terrain. We found that C dynamics are sensitive to a small number of the eco-physiological parameters (e.g. specific leaf area and C allocation ratio between stem and leaf), but the spatial distribution of C in complex terrain is dominated by parameters related to soil properties (i.e. van Genuchten parameters). Results from this part of the study guides construction of a watershed C data assimilation system.

E313 - POSTER-0182: Application of vegetation integrated simulator for trace-gases (VISIT) to impact assessment for 1.5/2.0K to high-end warming

Akihiko Ito2, 1

1NIES-CGER, Tsukuba, Japan 2Japan Agency for Marine-Earth Science and Technology (JAMSTEC), Yokohama, Japan

Terrestrial ecosystems are expected to experience different magnitude of impacts from climatic warming, leading to serious changes in ecosystem functions and services. We have developed an integrated model of atmosphere–ecosystem exchange of trace gases, including greenhouse gases (CO2, CH4, and N2O), biomass burning emissions (black carbon, CO, PM2.5 etc.), and biogenic volatile organic compounds (BVOCs). In addition to analyses of current budgets of these trace gases, we have applied the model to impact assessments of climatic change on the basis of multiple socioeconomic scenarios. In the ISI-MIP (phase 2b), we are conducting a scenario-based assessment on the different climatic impacts of 1.5K and 2.0K warming, aiming at making a contribution to form scientific base for the Paris Agreement. In the previous phase (2a), we conducted a benchmarking of biome models (e.g. on photosynthetic CO2 uptake) for verification using a couple of observational data. Currently, we are focusing on the arctic and alpine regions, where severer climatic warming is expected. Also in the IMPRESSIONS project, the model is used to assess the impacts of near-high-end warming, focusing on several areas in Europe. Considering combinations of different levels of temperature and precipitation changes, it is possible to estimate the likelihood of impacts on, for example, vegetation productivity and carbon stock. Additionally, we show several preliminary results of the impact assessment of countermeasure deployment such as climatic engineering and biofuel production on terrestrial ecosystem functions. Finally, we would like to discuss future directions of model development and validation for better estimation.

E314 POSTER-0317: Verification of IAP-DGVM with spatial analysis methods

Xiaodong Zeng1, Juan Li2, Xiang Song1

1International Center for Climate and Environment Sciences, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China 2Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

The Dynamic Global Vegetation Models (DGVMs) are developed to simulate the dynamics of the distribution and structure of natural vegetation in response to global climate change. Plants are usually classified into plant functional types (PFTs) according to their physical, phylogenetic and phenological characteristics. The PFT fractional coverage is one of the most important state variables, serving as the weight for integrating subgrid physical, biophysical and biochemical processes. In this work, spatial analysis methods of multiple intensities and scales were applied to the verification of IAP-DGVM. Vegetation coverage over each gridcell was summed into four categories, i.e., tree (7 PFTs), shrub (2 PFTs), grass (3 PFTs), and bare soil. The CLM4 surface vegetation dataset was used as the observational data of the global vegetation distribution. Nine intensities ranging from 10% to 90% respectively were applied. For each of the four categories, areas with fractional coverages larger than given intensity in both the DGVM simulation and observation were further separated into different clusters (objects) according connectivity. The skill score of each intensity as well as each DGVM object were calculated as the ratio of number of overlapping grids to the average number of grids of simulation and observation. Result showed that the skill score usually decreased as intensity increased. Among the four categories, bare soil possessed the highest skill scores, slightly decreased from 90% to 80% as intensity increased from 10% to 90%. The skill score of trees were also high, and were smaller than that of bare soil when intensity was larger than 70%. The skill scores of shrubs and grasses, on the other hand, were lower for all intensities, and dramatics decrease occurred as intensity was larger than 60%. Climatic information are applied in further analysis.

E315 POSTER-0378: Scaling light use and carbon dynamics from individual plants to the globe with LiDAR-derived vegetation demography in the Ent Terrrestrial Biosphere Model

Nancy Kiang1, Igor Aleinov2, Carlo Montes1, 3, Wenge Ni-Meister4, Crystal Schaaf5, Qingsong Sun5, Wenze Yang6, 7, Tian Yao8, Zhuosen Wang9, Dominique Carrer10

1NASA Goddard Institute for Space Studies, New York, NY, The United States of America 2Center for Climate Systems Research, Columbia University, New York, NY, The United States of America 3Universities Space Research Association, New York, NY, The United States of America 4Hunter College of The City University of New York, New York, NY, The United States of America 5University of Massachusetts Boston, Boston, MA, The United States of America 6National Oceanographic and Atmospheric Administration, College Park, MD, The United States of America 7I.M. Systems Group, Inc., College Park, MD, The United States of America 8Boston University, Boston, The United States of America 9NASA Goddard Space Flight Center, Greenbelt, The United States of America 10CNRM, Meteo France, Toulouse, France

Earth System Models (ESMs) have been limited in their simulation of land carbon and ecological dynamics, due to simplified representations of plant communities as “big leaf” or homogeneous stands, because biomass and plant activity scale non-linearly with woody height. Plant demography - mixed vegetation communities - is only beginning to be introduced into ESMs, by representing canopies in cohorts binned by height class. However, unrealistic canopy radiative transfer schemes result in over-suppression of understory plants. Also, global data sets to validate simulated vegetation communities are not available. The Ent Terrestrial Biosphere Model (Ent TBM) is a demographic dynamic global vegetation model (DGVM) coupled to the NASA Goddard Institute for Space Studies (GISS) ESM, utilizing the Analytical Clumped Two-Stream (ACTS) canopy radiative transfer scheme derived from remote sensing theory. To evaluate the model against global observations, we have constructed a first version of the Ent Global Vegetation Structure Data Set (Ent GVSD), which describes plant densities derived from allometric relations between satellite-derived canopy heights (Simard et al., 2011) and leaf area index (LAI, Yuan et al., 2011), with soil albedo beneath canopies (Carrer et al., 2014). This first version of the Ent GVSD serves as the reference case of homogeneous canopies but with global variation in canopy heights and plant densities, as boundary conditions to vegetation physical properties. We evaluate estimated biomass at the spatial resolutions of 1 km and GISS ESM grid of 2.5°x2°, and present climate sensitivity of carbon balances with the Ent TBM coupled to the GISS ESM. We analyze strengths and weaknesses pertaining to particular plant functional types, seasonal diagnostics, and model biases. Next versions of the Ent GVSD will incorporate canopy height stratification of mixed communities and forest types with demography derived from a combination of LiDAR observations and ecological theory.

E316 - POSTER-0391: Assimilation of remotely sensed Leaf Area Index into the Community Land Model with explicit carbon and nitrogen components using Data Assimilation Research Testbed

Xiaolu Ling1, Congbin Fu3, 2, Zongliang Yang4, Weidong Guo5

1Joint International Research Laboratory of Atmospheric and Earth System Sciences of Ministry of Education, School of Atmospheric Sciences, Nanjing University, Nanjing, China 2Joint International Research Laboratory of Atmospheric and Earth System Sciences of Ministry of Education, School of Atmosphric Sciences, Nanjing University, Nanjing, China 3Nanjing University, Nanjing, China 4Department of Geological Sciences, John A. and Katherine G. Jackson School of Geosciences, University of Texas at Austin, Austin, The United States of America 5Joint International Research Laboratory of Atmospheric and Earth System Sciences, School of Atmospheric Sciences, Nanjing University, Nanjing, China

Information of the spatial and temporal patterns of leaf area index (LAI) is crucial to understand the exchanges of momentum, carbon, energy, and water between the terrestrial ecosystem and the atmosphere, while both in-situ observation and model simulation usually show distinct deficiency in terms of LAI coverage and value. Land data assimilation, combined with observation and simulation together, is a promising way to provide variable estimation. Meanwhile, satellite-derived dataset could provide the data basis for land data assimilation at global scale. The Data Assimilation Research Testbed (DART) developed and maintained by the National Centre for Atmospheric Research (NCAR) provides a powerful tool to facilitate the combination of assimilation algorithms, models, and real (as well as synthetic) observations to better understanding of all three. Here we systematically investigated the effects of data assimilation on improving LAI simulation based on NCAR Community Land Model with the prognostic carbon–nitrogen option (CLM4CN) linked with DART using the deterministic Ensemble Adjustment Kalman Filter (EAKF). The Global Land Surface Satellite LAI data (GLASS LAI) LAI is assimilated into the CLM4CN at a frequency of 8 days, and LAI (and leaf carbon / nitrogen) are adjusted at each time step. The results show that assimilating remotely sensed LAI into the CLM4CN is an effective method for improving model performance. In detail, the CLM4-CN simulated LAI systematically overestimates global LAI, especially in low latitude with the largest bias of 5 m2/m2. While if updating both LAI and leaf carbon and leaf nitrogen simultaneously during assimilation, the analyzed LAI can be corrected, especially in low latitude regions with the bias controlled around ±1 m2/m2. In addition, the best method for LAI assimilation should include the EAKF method, the accepted percentage of all observation, as well as the carbon-nitrogen control.

E317 POSTER-0405: Inter-Comparison and Functional Benchmarking of Ecosystem Models for NASA’s ABoVE Field Campaign.

Wouter Hantson1, Daniel J. Hayes1, Eric Stofferahn2, Christopher Schwalm3, Deborah Huntzinger4, Joshua Fisher2

1University of Maine, Orono, The United States of America 2Jet Propulsion Laboratory, California Institute of Technology, Pasadena, The United States of America 3Woods Hole Research Center, Woods Hole, The United States of America 4Northern Arizona University, Flagstaff, Arizona, The United States of America

The Arctic Boreal region (ABR) is the source of among the largest uncertainties to global climate projections1. High uncertainties for carbon fluxes, defined as the variance among multiple models2, means that Arctic-Boreal Ecosystem Models can exhibit nearly every possible combination of net carbon flux source / sink pattern as shown by Fisher et al. (2014) for Alaska3. A key challenge is that there are few data available to benchmark models and guide improvements, and thus to reduce uncertainty.
 
Within our ABoVE project, we are evaluating, identifying and quantifying the sensitivity – vulnerability and resilience – of ecosystem dynamics in the ABR as captured by terrestrial biosphere models (TBMs). Results will be used to improve TBM performance in representing, simulating, and scaling the key indicators of ABR ecosystem dynamics and their associated uncertainties. By leveraging existing TBM simulation results from the MsTMIP-project4, existing model outputs can be compared to and evaluated against field and remote sensing data among the ABoVE ecosystem dynamics indicators i.e., (vegetation, carbon, permafrost, water, wildlife/habitat). These data include remote sensing observations and products for: phenology (MODIS NDVI, EVI, fAPAR), GPP and NPP (MODIS), fire (MODIS), albedo (MODIS), biomass (ICESat/GLAS), canopy height (ICESat/GLAS), evapotranspiration (MODIS), soil moisture (SMOS), total water storage and derived groundwater (GRACE), and land surface temperature (MODIS). This suite of data provides a robust constraint on interacting biogeochemical components of dynamic ABR ecosystems. Initial model inter- comparison and evaluation against Arctic/Boreal specific benchmarks, derived from field and remote sensing data, show model performance, sensitivities and uncertainties in key ecological and carbon cycle indices across the ABR.
 
References: 1 Koven et al.; IPCC, 2014; Schaefer et al., 2014; Sneyder and Liess, 2014; 2 IPCC, 2014; 3 Fisher et al., 2014; 4 Huntzinger et al., 2013.