Reliable, robust and realistic: the three R's of next-generation land-surface modelling

Abstract

Abstract. Land-surface models (LSMs) are increasingly called upon to represent not only the exchanges of energy, water and momentum across the land-atmosphere interface (their original purpose in climate models), but also how ecosystems and water resources respond to climate, atmospheric environment, land-use and land-use change, and how these responses in turn influence land-atmosphere fluxes of carbon dioxide (CO 2 ), trace gases and other species that affect the composition and chemistry of the atmosphere. However, the LSMs embedded in state-of-the-art climate models differ in how they represent fundamental aspects of the hydrological and carbon cycles, resulting in large inter-model differences and sometimes faulty predictions. These "thirdgeneration" LSMs respect the close coupling of the carbon and water cycles through plants, but otherwise tend to be under-constrained, and have not taken full advantage of robust hydrological parameterizations that were independently developed in offline models. Benchmarking, combining multiple sources of atmospheric, biospheric and hydrological data, should be a required component of LSM development, but this field has been relatively poorly supported and intermittently pursued. Moreover, benchmarking alone is not sufficient to ensure that models improve. Increasing complexity may increase realism but decrease reliability and robustness, by increasing the number of poorly known model parameters. In contrast, simplifying the representation of complex processes by stochastic parameterization (the representation of unresolved processes by statistical distributions of values) has been shown to improve model reliability and realism in both atmospheric and land-surface modelling contexts. We provide examples for important processes in hydrology (the generation of runoff and flow routing in heterogeneous catchments) and biology (carbon uptake by speciesdiverse ecosystems). We propose that the way forward for next-generation complex LSMs will include: (a) representations of biological and hydrological processes based on the implementation of multiple internal constraints; (b) systematic application of benchmarking and data assimilation techniques to optimize parameter values and thereby test the structural adequacy of models; and (c) stochastic parameterization of unresolved variability, applied in both the hydrological and the biological domains.

Quantification and attribution of errors in the simulated annual gross primary production and latent heat fluxes by two global land surface models

“…A significant part of those differences can be attributed to the differences in their model components, model parameter values [Friedlingstein et al, 2013;Prentice et al, 2015], and model input [Fekete et al, 2004]. For example, the one standard deviation of the simulated net carbon uptake from 1900 to 2100 by different global land models increased steadily from 117 to 259 Gt C, while the mean amount of C uptake as estimated by different models remained between 326 and 410 Gt C over the last three successive assessment reports by the IPCC.…”

“…Large discrepancies have been found among the simulated gross primary production (GPP) and latent heat (LE) fluxes by the most advanced earth system models in the most recent assessment report (AR) by the International Panel on Climate Change (IPCC) [Ciais et al, 2013], and those differences have not decreased over the last three successive assessments by the IPCC [Prentice et al, 2015]. A significant part of those differences can be attributed to the differences in their model components, model parameter values [Friedlingstein et al, 2013;Prentice et al, 2015], and model input [Fekete et al, 2004].…”

“…For example, the one standard deviation of the simulated net carbon uptake from 1900 to 2100 by different global land models increased steadily from 117 to 259 Gt C, while the mean amount of C uptake as estimated by different models remained between 326 and 410 Gt C over the last three successive assessment reports by the IPCC. Among the 12 earth system models for the fifth assessment report of IPCC, only the mean annual gross primary production (GPP) and latent heat flux (LE) by two models fall within the ranges based on observations [Prentice et al, 2015]. Therefore, quantification and attribution of those errors in land surface models are essential for future model improvement and development [Luo et al, 2012].…”

Description and validation of the Simple, Efficient, Dynamic, Global, Ecological Simulator (SEDGES v.1.0)

“…Within the context of the history of land surface modeling, the SEDGES framework (defined as SEDGES and the type of land surface model that it presupposes that it forms a part of) combines aspects of first-and thirdgeneration models (Sellers et al, 1997;Pitman, 2003;Prentice et al, 2015), which include, respectively, the use of a simple single-layer "bucket" model for soil hydrology and the full coupling of photosynthesis and transpiration through interactive canopy conductance.…”

“…Choosing the appropriate level of complexity of represented processes within a land surface or dynamic global vegetation model often depends on the context in which the model is to be used and involves subjective weighting of trade-offs between increased accuracy and realism on the one hand and, on the other hand, robustness to the chosen values of poorly constrained parameters and model reliability in a wide range of situations (Prentice et al, 2015). This paper presents the SEDGES model's structure, equations, and ability to simulate ecological and hydrological variables when used in conjunction with a simple soil hydrological scheme and parameterization for aerodynamic conductance and forced by reanalysis data.…”