Skip to main content

Re-inventing Ecosystem And Land-surface Models

Periodic Reporting for period 2 - REALM (Re-inventing Ecosystem And Land-surface Models)

Reporting period: 2020-04-01 to 2021-09-30

Terrestrial ecosystems respond to changes in climate and the atmospheric environment, which they in turn help to regulate. As global change has become an international concern, high expectations have been laid on Earth Ssystem Mmodels (ESMs) with embedded ecosystem and land-surface components to deliver reliable, quantitative predictions of large-scale changes in ecosystems and their feedbacks to climate. But the lack of established quantitative theory for many key processes – including the long-term effects of temperature on primary production and carbon allocation, the sustainability and nutrient requirements of CO2 ‘fertilization’, and the regulation of green vegetation cover and its water use – has made such expectations impossible to fulfil. As a result, numerical models of land ecosystem processes continue to disagree both with one another, and with benchmark data sets. This impasse can be overcome, but not without re-thinking modelling practice. Theory must be re-instated as the required link between observations and models. Multidisciplinary data resources now available should be used more extensively and creatively. Observational and experimental results should be made integral to model development, and not restricted to ‘end-of-pipe’ benchmarking of complex, poorly constrained models.
REALM is developing a comprehensive, next-generation vegetation model using eco-evolutionary optimality hypotheses to generate quantitative, testable predictions, and multiple data sources to provide tests. We expect this work to lead to greatly improved predictability, both of ecosystems’ reactions to change, and of climate itself. Ultimately, it will transform the practice of global vegetation and land-surface modelling, thereby establishing the foundations for a more robust understanding of the role of terrestrial ecosystems in Earth System dynamics and climate change.
The benefits of REALM
There are three reasons why REALM will lead to three main improvements in Earth System Models (ESMs):
1) First-principles approaches based on eco-evolutionary optimality hypotheses are generating new, surprisingly simple, robust and well-founded governing equations for the key plant and ecosystem processes that are represented in ESMs.
2) Such equations can now be tested, drawing on recent huge increases in the amount and variety of relevant observations available to the research community. The REALM approach willis thus allowing the new-found wealth of available data to become an essential component of ESM development.
3) Current models of plant and ecosystem processes remain essentially physical, the role of biology being limited to providing (over-)generalized parameter values for ‘plant functional types’ based loosely on published measurements. The REALM approach, by contrast, puts biology at the centre – harnessing the power of natural selection, which is the “missing law” that has been largely neglected in decades of research on land-surface modelling. In this new framework, key adaptive processes – such as photosynthetic and respiratory acclimation are– are no longer treated as ‘add-ons’ that further complicate models. instead, they are integral to the way that the new models represent plant and ecosystem function.
By establishing firmer theoretical and empirical foundations for modelling, REALM is addressing some still unanswered, yet fundamental uncertainties about how ecosystems work:
• The relationship of green vegetation cover to CO2, water and energy supply. There is no general quantitative theory for this, and current models predict vegetation cover very poorly.
• The temperature dependence of gross primary production (GPP) under field conditions. Even the sign of this relationship is unclear from the literature, and differs among models.
• The influence of nutrient supplies on the carbon cycle. This has been a recent target for model ‘improvement’ but there is considerable confusion about the mechanisms involved.
• The competitive advantage of plants with different morphologies and phenologies. There is still no accepted explanation, for example, for the global distribution of vegetation dominated by deciduous versus evergreen trees.
The REALM Challenges
REALM is structured around four challenges:
1) Explaining and predicting the carbon and water exchanges of terrestrial ecosystems.
2) Accounting for the functional diversity of plant communities.
3) Explaining and predicting global phytogeographic patterns.
4) Modelling vegetation processes and carbon cycling in a changing environment.
The following is a summary of key REALM findings that represent the most substantial advances beyond the state-of-the-art as it was at the beginning of the project:
• It is possible to represent subdaily GPP variations in an optimality framework using far fewer parameters the current models.
• The controls on GPP have been quantified by applying regression methods to flux data which provides a much more informative model benchmark.
• Shown that forests produce biomass more efficiently at higher temperatures. This is consistent with optimality theory, but opposite in sign to what is simulated by state-of-the-art models.
• Preliminary results support an entirely new theory that predicts the optimal allocation of carbon to leaves as a function of potential gross primary production.
• The plasticity of different leaf traits has been quantified, for the first time, based on extensive re-sampling of species along environmental gradients.
• A unifying theory comprising several novel elements predicts the existence of the leaf economics spectrum and how leaf mass-per-area varies, quantitatively, across environments.
• Observed trends in GPP at flux sites can be correctly simulated using the P model – a first for any process-based model, as far as we know.
By the end of the project we expect to have achieved the following:
• A global implementation of the extended P model in the ECMWF framework, and assessments of its performance using both in situ data and global atmospheric CO2 inversions.
• A comprehensive and well-tested scheme to predict the allocation of carbon to leaves, stems and roots (including plant height, root-to-shoot ratio and leaf area index) by plants in different environments.
• A ‘plant-centred’ representation of root-zone size and its influence on GPP and hydrology.
• A first-principles explanation of the distribution of biomes, implemented in a global model.
• A demonstration of niche complementarity and its relation to biodiversity in complex forests.
• An understanding of recent trends in the amplitude of the seasonal cycle of atmospheric CO2.
• A well-founded assessment of the potential consequences of alternative CO2 and climate-change scenarios for greening/browning, terrestrial CO2 uptake, and the global distributions of different types of plants.
REALM team members at Silwood Park Campus, Imperial College London.