Abstract

This paper presents a set of analytical tools to evaluate the performance of three land surface models (LSMs) that are used in global climate models (GCMs). Predictions of the fluxes of sensible heat, latent heat, and net CO₂ exchange obtained using process-based LSMs are benchmarked against two statistical models that only use incoming solar radiation, air temperature, and specific humidity as inputs to predict the fluxes. Both are then compared to measured fluxes at several flux stations located on three continents. Parameter sets used for the LSMs include default values used in GCMs for the plant functional type and soil type surrounding each flux station, locally calibrated values, and ensemble sets encompassing combinations of parameters within their respective uncertainty ranges. Performance of the LSMs is found to be generally inferior to that of the statistical models across a wide variety of performance metrics, suggesting that the LSMs underutilize the meteorological information used in their inputs and that model complexity may be hindering accurate prediction. The authors show that model evaluation is purpose specific; good performance in one metric does not guarantee good performance in others. Self-organizing maps are used to divide meteorological “‘forcing space” into distinct regions as a mechanism to identify the conditions under which model bias is greatest. These new techniques will aid modelers to identify the areas of model structure responsible for poor performance.

Abstract

Data assimilation in the field of predictive land surface modeling is generally limited to using observational data to estimate optimal model states or restrict model parameter ranges. To date, very little work has attempted to systematically define and quantify error resulting from a model's inherent inability to simulate the natural system. This paper introduces a data assimilation technique that moves toward this goal by accounting for those deficiencies in the model itself that lead to systematic errors in model output. This is done using a supervised artificial neural network to “learn” and simulate systematic trends in the model output error. These simulations in turn are used to correct the model's output each time step. The technique is applied in two case studies, using fluxes of latent heat flux at one site and net ecosystem exchange (NEE) of carbon dioxide at another. Root-mean-square error (rmse) in latent heat flux per time step was reduced from 27.5 to 18.6 W m⁻² (32%) and monthly from 9.91 to 3.08 W m⁻² (68%). For NEE, rmse per time step was reduced from 3.71 to 2.70 μmol m⁻² s⁻¹ (27%) and annually from 2.24 to 0.11 μmol m⁻² s⁻¹ (95%). In both cases the correction provided significantly greater gains than single criteria parameter estimation on the same flux.

Abstract

Satellite and gridded meteorological data can be used to estimate evaporation (E) from land surfaces using simple diagnostic models. Two satellite datasets indicate a positive trend (first time derivative) in global available energy from 1983 to 2006, suggesting that positive trends in evaporation may occur in “wet” regions where energy supply limits evaporation. However, decadal trends in evaporation estimated from water balances of 110 wet catchments do not match trends in evaporation estimated using three alternative methods: 1) , a model-tree ensemble approach that uses statistical relationships between E measured across the global network of flux stations, meteorological drivers, and remotely sensed fraction of absorbed photosynthetically active radiation; 2) , a Budyko-style hydrometeorological model; and 3) , the Penman–Monteith energy-balance equation coupled with a simple biophysical model for surface conductance. Key model inputs for the estimation of and are remotely sensed radiation and gridded meteorological fields and it is concluded that these data are, as yet, not sufficiently accurate to explain trends in E for wet regions. This provides a significant challenge for satellite-based energy-balance methods. Trends in for 87 “dry” catchments are strongly correlated to trends in precipitation (R ² = 0.85). These trends were best captured by , which explicitly includes precipitation and available energy as model inputs.

Savannas are highly significant global ecosystems that consist of a mix of trees and grasses and that are highly spatially varied in their physical structure, species composition, and physiological function (i.e., leaf area and function, stem density, albedo, and roughness). Variability in ecosystem characteristics alters biophysical and biogeochemical processes that can affect regional to global circulation patterns, which are not well characterized by land surface models. We initiated a multidisciplinary field campaign called Savanna Patterns of Energy and Carbon Integrated across the Landscape (SPECIAL) during the dry season in Australian savannas to understand the spatial patterns and processes of land surface–atmosphere exchanges (radiation, heat, moisture, CO₂, and other trace gasses). We utilized a combination of multiscale measurements including fixed flux towers, aircraft-based flux transects, aircraft boundary layer budgets, and satellite remote sensing to quantify the spatial variability across a continental-scale rainfall gradient (transect). We found that the structure of vegetation changed along the transect in response to declining average rainfall. Tree basal area decreased from 9.6 m² ha⁻¹ in the coastal woodland savanna (annual rainfall 1,714 mm yr⁻¹) to 0 m² ha⁻¹ at the grassland site (annual rainfall 535 mm yr⁻¹), with dry-season green leaf area index (LAI) ranging from 1.04 to 0, respectively. Leaf-level measurements showed that photosynthetic properties were similar along the transect. Flux tower measurements showed that latent heat fluxes (LEs) decreased from north to south with resultant changes in the Bowen ratios (H/LE) from a minimum of 1.7 to a maximum of 15.8, respectively. Gross primary productivity, net carbon dioxide exchange, and LE showed similar declines along the transect and were well correlated with canopy LAI, and fluxes were more closely coupled to structure than floristic change.