Land Surface Modeling: A Mini-Workshop

5. Plenary Discussion: Recommendations

5.1 Research and Modeling Priorities

The plenary discussion identified several areas in which land surface models need improvement to achieve overall research objectives. The topics discussed included:

  1. snow cover: a key climate feedback for both long time scales and seasonal-to-interannual climate; a related aspect of the modeling concerns permafrost and freeze-thaw processes, which need to be represented well models are to achieve realistic run-off and be suitable for applications involving soil properties.
  2. run-off: provides a principal driver for the ocean in coupled models, while changes in runoff can have substantial social impacts. More importantly, its proper parameterization is critical to the accurate representation of other boundary fluxes (evaporation, sensible heat) on seasonal-to-interannual timescales.
  3. wetlands and lakes: poor representation of these can have a large effect on the realism of the simulated atmosphere, river run-off, and evaporation; the ability to model wetlands is fundamental to analyses of the global methane cycle.
  4. dynamic vegetation: physiological and structural variations in vegetation exert an impact on climate on a range of time scales from seasonal to centennial.
  5. spatial heterogeneity: how to account for the great variations of physical properties and processes within a climate model grid-box remains a major challenge and a focus of model development.
  6. deep ground water: mechanisms for transfer of water from the surface to deep layers need to be realistic; among other things this suggests the need to account for percolation properties of karst formations.

Some of these topics can be relatively isolated, e.g., snow modeling, which is thus a good candidate for development of a parameterization that could be shared by many groups. Other topics, such as runoff, are a product of many parts of the land and atmosphere models, and thus the most profitable cooperation may lie in the provision of well-defined data sets that allow intercomparison, testing and improvement of the models.

We did not assign relative priorities to these research areas, which were all judged to be important for the general objective of studying interannual to multi-decadal climate variations. Prioritization seemed inappropriate because of the range of specific objectives among the different research groups. On the other hand, the various research foci of individual institutions provide complementary capabilities and interests that suggest potentially valuable collaborations in land surface model development and applications.

We have assembled in Table 1 some connections among the research foci and research groups, with the objective of encouraging communication and collaboration among groups. We include comments about the anticipated gains with improved modeling, as well as comments on the strengths and weaknesses of the current modeling and validation.

5.2 Validation and Other Data Sets

Collaboration also needs to include the development and sharing of data sets, such as global land characteristics needed as model input and for model assessment. Data sets discussed at the workshop included soil moisture, snow cover, river runoff, topography, watershed routing and other variables. Table 2 summarizes some of the important categories and generic sources of data sets. It is clear that many useful data are scattered throughout published field studies, although some are integrated datasets (e.g., BOREAS).

Mechanisms for provision and exchange of data sets are readily available via the internet, electronic file transfer procedures, and cataloguing of available information by data centers such as those supported by the NASA Earth Observing System Data Information System (EOSDIS).

Effective utilization of this potential requires cooperation and coordination among the relevant researchers and groups, the final topic discussed at the workshop.

5.3 Mechanisms for Interaction and Collaboration

There was general agreement on the need to improve interactions and cooperation among individual researchers and groups in land surface modeling, in global climate modeling and in modeling applications. This topic has received widespread attention recently because of the perception that the extensive talents and abilities of researchers and modeling groups in the United States are not being utilized as effectively as possible toward advancing our understanding of global climate change.

The most effective approach to rapid improvement in the modeling and applications may be to find mechanisms to improve the interaction and collaboration among researchers and modeling groups, especially to involve the academic research community more extensively in large-scale modeling at government laboratories. Several ideas were discussed that have the potential to yield substantial advances while requiring only moderate resources.

Model intercomparisons. There has been substantial effort in comparison of land surface models in recent years (e.g., PILPS). However, it is felt that there is need for intercomparisons at the level of particular processes in the models, where it is more feasible to identify cause and effect in analyzing model performance. Such intercomparisons might be identified in more specific mini-workshops, as discussed below.

Computer code modularization. The tremendous range of capabilities of researchers in different aspects of the global modeling and analysis could be utilized more effectively if, instead of each global modeling group trying to cover all bases, there was the ability to share intellectual resources. It is understood that the notion of "plug compatibility" of all routines in global models is over simplified. Nevertheless, some advantages could be obtained if modeling groups would work toward much more commonality in the infrastructure of the models, including software and data. These topics were discussed in more detail at a workshop held at the National Center for Environmental Prediction on August 5-6, and it was subsequently announced that an initial working meeting will be held in Tucson on October 15-16.

Student and post-doctoral exchanges. Capabilities of academic research communities could be used more effectively, to the benefit of both global modeling groups and university researchers, if there were more support for students and post-doctoral scientists to work with the modeling groups while working under the direction of, or in collaboration with, university scientists. The benefit-to-cost ratio of such exchanges is considered to be high.

Joint global modeling experiments. A point related to student and post-doctoral exchanges is collaboration on use of global models for land surface and global climate experiments. Seldom, if ever, is it possible to treat a global model as a "black box" for successful studies of how the real world works. It usually requires cooperation with those who build models to understand how the models work and to interpret the results of experiments. But this understanding can often be obtained via personnel exchanges, including students and post-doctoral researchers.

Workshops. Progress in the modeling and implementation of joint activities such as discussed above would be facilitated by workshops and mini-workshops on specific topics, which can be organized by different researchers and groups with appropriate interests. An example is the workshop on modeling infrastructure to be held at the University of Arizona on October 15-16, 1998. The merits of having a workshop aimed at understanding present and future levels of atmospheric methane were briefly mentioned. Two topics that are felt to be of high priority for the GISS model are sea ice modeling, and wetland/lake/river modeling; both of these areas will probably be the subject of mini-workshops at GISS within the next year.

Table 1. Priority research foci for land-surface models in GCMs, including expected gains and current limitations and strengths as well as model groups currently focusing on these topics.
Development / Area / Group Model Component / Improvements Anticipated Gains Current Strengths (S) and Weaknesses (W)
snow submodels surface temperature, albedo, (surface fluxes) S: models developed; W: validation data improving but regional gaps remain
soil physics/snow in LS
soil temperature
distribution of boreal veg. for current & other climates; soil T S: --; W: scarce validation data
PBL interface:
turbulence model
coupling methods
improved fluxes into the atmosphere S: recent advances in turbulence theory; W: ad hoc couplings to surface
Lakes & Wetlands:
linked surface-water model
predictive lakes and wetlands
river flow and discharge, ocean salinity/circulation, runoff, evaporation, lake area, volume, temperature, hydrology under various climates S: equilibrium wetland/lake model; W: seasonal wetland/lake model; vertical resolution of topography data
(+atm. chemistry)
predictive wetlands
(seasonal/interannual area and location)
prediction of seasonal and longer-term methane/climate interactions S: methane production, transport, & emission models
; W: seasonal wetland model; data on seasonal/interannual wetland variation
Runoff, Riverflow:
sub-grid scale effects improved energy and water budgets S: new approaches currently under development; W: longterm and large-area validation data
Soil Moisture:
more realistic percolation, vertical layering improved energy and water budgets S: new approaches currently under development; Robock field data set; W: highly variable, remote sensing approaches not proven effective
Dynamic Veg:
climate vegetation interactions seasonal-interannual-decadal C, H2O, energy exchange; feedbacks between atm-bios. transient change S: theory and offline models developed; remote and other veg. observations available; W: hydrology/energy still simple
Land Use:
LS: C response to & recovery from disturbance improved simulation of atmospheric concentration of CO2 S: required DVMs ~ available; W: need historical data (temporal, spatial, type)
CPEP = Climate, People, and the Environment Program at University of Wisconsin-Madison
CSU = Colorado State University
GSFC = Goddard Space Flight Center
GISS = Goddard Institute for Space Studies

Table 2. Summary of types of data needed and availability.
Category Source
runoffriver discharge, river gauge
wetlands, lakes, riversmicrowave obs., monthly, interannual
soil moisturefield measurements, ? GEWEX
evaporationfield measurements, ?
distribution traditional obs., remote sensing obs., paleo obs./analogues
continuous characteristics (woody, herb, bare) indirect - remote sensing & mixture modeling
phenologyremote sensing observations
leaf area direct - field observations
indirect - NDVI, f (veg. type)
height (roughness length)field measurements, altimeter?
root depth, activity, profilesfield measurements
leaf orientationfield measurements
natural disturbances (e.g., fire)remote sensing
land-use change historical and current from combined satellite, traditional, statistical, and environmental data
progressive degradation remote sensing
N fertilization?
Carbon Exchange
seasonality, distribution, magnitude indirect: atmospheric CO2 and CH4 observations (global, seasonal, decadal, isotopic)
field measurements: soil respiration, photosynthesis autotropic respiration
Surface Energy
surface reflectance satellite obs. (spectral, global, weekly/monthly, interannual)
surface temperature station and satellite obs. (global, interannual)
photosynthetically active radiation (PAR) remote sensing
latent and heat sensible fluxesindirect
areal extent station obs., optical/microwave satellite obs. depth
station and microwave satellite obs.
phenology station obs., optical satellite obs.

Workshop Homepage * Summary * Agenda
Background * Presentations * Discussion, Recommendations
References * Participants