scholarly journals Deep reinforcement learning in World-Earth system models to discover sustainable management strategies

2020 ◽  
Author(s):  
Felix Strnad ◽  
Wolfram Barfuss ◽  
Jonathan Donges ◽  
Jobst Heitzig
Keyword(s):  
Sustainable Development ◽  
Reinforcement Learning ◽  
Sustainable Management ◽  
System Analysis ◽  
Management Strategies ◽  
Earth System ◽  
Management Options ◽  
System Models ◽  
The World ◽  
Earth System Models

<p>The identification of pathways leading to robust mitigation of dangerous anthropogenic climate change is nowadays of particular interest <br>not only to the scientific community but also to policy makers and the wider public. </p><p>Increasingly complex, non-linear World-Earth system models are used for describing the dynamics of the biophysical Earth system and the socio-economic and socio-cultural World of human societies and their interactions. Identifying pathways towards a sustainable future in these models is a challenging and widely investigated task in the field of climate research and broader Earth system science.  This problem is especially difficult when caring for both environmental limits and social foundations need to be taken into account.</p><p>In this work, we propose to combine recently developed machine learning techniques, namely deep reinforcement learning (DRL), with classical analysis of trajectories in the World-Earth system as an approach to extend the field of Earth system analysis by a new method. Based on the concept of the agent-environment interface, we develop a method for using a DRL-agent that is able to act and learn in variable manageable environment models of the Earth system in order to discover management strategies for sustainable development.</p><p>We demonstrate the potential of our framework by applying DRL algorithms to stylized World-Earth system models. The agent can apply management options to an environment, an Earth system model, and learn by rewards provided by the environment. We train our agent with a deep Q-neural network extended by current state-of-the-art algorithms. Conceptually, we thereby explore the feasibility of finding novel global governance policies leading into a safe and just operating space constrained by certain planetary and socio-economic boundaries.  </p><p>We find that the agent is able to learn novel, previously undiscovered policies that navigate the system into sustainable regions of the underlying conceptual models of the World-Earth system. In particular, the artificially intelligent agent learns that the timing of a specific mix of taxing carbon emissions and subsidies on renewables is of crucial relevance for finding World-Earth system trajectories that are sustainable in the long term. Overall, we show in this work how concepts and tools from artificial intelligence can help to address the current challenges on the way towards sustainable development.</p><p>Underlying publication</p><p>[1] Strnad, F. M.; Barfuss, W.; Donges, J. F. & Heitzig, J. Deep reinforcement learning in World-Earth system models to discover sustainable management strategies Chaos: An Interdisciplinary Journal of Nonlinear Science, AIP Publishing LLC, 2019, 29, 123122</p>

Is large-scale terrestrial hydrological cycling well represented in Earth System Models?

2021 ◽  
Author(s):  
Navid Ghajarnia ◽  
Zahra Kalantari ◽  
Georgia Destouni
Keyword(s):  
Climate Change ◽  
Observational Data ◽  
Large Scale ◽  
Hydrological Cycle ◽  
Good Representation ◽  
Earth System ◽  
System Models ◽  
The World ◽  
Terrestrial Hydrology ◽  
Earth System Models

<p>This paper addresses how large-scale terrestrial water cycling is represented in the land surface schemes of Earth System Models (ESMs). Good representation is essential, for example in regional planning for climate change adaptation and in preparation for hydro-climatic extremes that have recently set records world-wide in devastating consequences for societies and deaths of thousands of people. ESMs provide simulations and projections for the climate system and its interactions with the terrestrial hydrological cycle, and are widely used to study and prepare for associated impacts of climate change. However, the reliability of ESMs is unclear with regard to their representation of large-scale terrestrial hydrology and its changes and interactions between its key variables‎. Despite being crucial for model realism, analysis of co-variations among terrestrial hydrology variables is still largely missing in ESM performance evaluations. To bridge this research gap, we have studied and identified large-scale co-variation patterns between soil moisture (SM) and the main freshwater fluxes of runoff (R), precipitation (P), and evapotranspiration (ET) from observational data and across 6405 hydrological catchments in different parts and climates of the world. Furthermore, we have compared the identified observation-based relationships with those emerging from ESMs and reanalysis products. Our results show that the most strongly correlated freshwater variables based on observational data are also the most misrepresented hydrological patterns in ESMs and reanalysis simulations. In particular, we find SM and R to have the generally strongest large-scale correlations according to the observation-based data, across the numerous studied catchments with widely different hydroclimatic characteristics. Compared to the SM-R correlation signals, the observation-based correlations are overall weaker for the commonly expected closer dependencies of: R on P; ET on P; SM on P; and ET on SM. Nevertheless, this strongest SM-R correlation and the P-R correlation are the most misrepresented hydrological patterns in reanalysis products and ESMs. Our results also show that ESM outputs can perform relatively well in simulating individual hydrological variables, while exhibiting essential inconsistencies in simulated co-variations between variables. Such investigations of large-scale terrestrial hydrology representation by ESMs can enhance our understanding of fundamental ESM biases and uncertainties while providing important insights for systematic ESM improvement with regard to the large-scale hydrological cycling over the world’s continents and regional land areas.</p>

scholarly journals The technosphere in Earth System analysis: A coevolutionary perspective

2017 ◽  
Vol 4 (1) ◽  
pp. 23-33 ◽  
Author(s):  
Jonathan F Donges ◽  
Wolfgang Lucht ◽  
Finn Müller-Hansen ◽  
Will Steffen
Keyword(s):  
Sustainable Development ◽  
System Analysis ◽  
Adaptive Networks ◽  
Earth System ◽  
Management Options ◽  
Technological Processes ◽  
The Earth ◽  
Joint Dynamics ◽  
Complex Adaptive ◽  
Earth System Analysis

Earth System analysis is the study of the joint dynamics of biogeophysical, social and technological processes on our planet. To advance our understanding of possible future development pathways and identify management options for navigating to safe operating spaces while avoiding undesirable domains, computer models of the Earth System are developed and applied. These models hardly represent dynamical properties of technological processes despite their great planetary-scale influence on the biogeophysical components of the Earth System and the associated risks for human societies posed, e.g. by climatic change or novel entities. In this contribution, we reflect on the technosphere from the perspective of Earth System analysis with a threefold focus on agency, networks and complex coevolutionary dynamics. First, we argue that Haff’s conception of the technosphere takes an extreme position in implying a strongly constrained human agency in the Earth System. Assuming that the technosphere develops according to dynamics largely independently of human intentions, Haff’s perspective appears incompatible with a humanistic view that underlies the sustainability discourse at large and, more specifically, current frameworks such as UN sustainable development goals and the safe and just operating space for humanity. Second, as an alternative to Haff’s static three-stratum picture, we propose complex adaptive networks as a concept for describing the interplay of social agents and technospheric entities and their emergent dynamics for Earth System analysis. Third, we argue that following a coevolutionary approach in conceptualising and modelling technospheric dynamics, also including the socio-cultural and biophysical spheres of the Earth System, could resolve the apparent conflict between the discourses on sustainability and the technosphere. Hence, this coevolutionary approach may point the way forward in modelling technological influences in the Earth System and may lead to a considerably deeper understanding of pathways to sustainable development in the future.

scholarly journals The evaluation of Earth System Models: discussion summary

2011 ◽  
Vol 6 ◽  
pp. 216-221
Author(s):  
Sönke Zaehle ◽  
Colin Prentice ◽  
Sarah Cornell
Keyword(s):  
Earth System ◽  
System Models ◽  
Earth System Models

scholarly journals Using field observations to inform thermal hydrology models of permafrost dynamics with ATS (v0.83)

2015 ◽  
Vol 8 (4) ◽  
pp. 3235-3292 ◽  
Author(s):  
A. L. Atchley ◽  
S. L. Painter ◽  
D. R. Harp ◽  
E. T. Coon ◽  
C. J. Wilson ◽  
...  
Keyword(s):  
Carbon Sink ◽  
Thermal Response ◽  
Field Measurements ◽  
Early Summer ◽  
Active Layer Thickness ◽  
Earth System ◽  
Model Refinement ◽  
System Models ◽  
Developed Surface ◽  
Earth System Models

Abstract. Climate change is profoundly transforming the carbon-rich Arctic tundra landscape, potentially moving it from a carbon sink to a carbon source by increasing the thickness of soil that thaws on a seasonal basis. However, the modeling capability and precise parameterizations of the physical characteristics needed to estimate projected active layer thickness (ALT) are limited in Earth System Models (ESMs). In particular, discrepancies in spatial scale between field measurements and Earth System Models challenge validation and parameterization of hydrothermal models. A recently developed surface/subsurface model for permafrost thermal hydrology, the Advanced Terrestrial Simulator (ATS), is used in combination with field measurements to calibrate and identify fine scale controls of ALT in ice wedge polygon tundra in Barrow, Alaska. An iterative model refinement procedure that cycles between borehole temperature and snow cover measurements and simulations functions to evaluate and parameterize different model processes necessary to simulate freeze/thaw processes and ALT formation. After model refinement and calibration, reasonable matches between simulated and measured soil temperatures are obtained, with the largest errors occurring during early summer above ice wedges (e.g. troughs). The results suggest that properly constructed and calibrated one-dimensional thermal hydrology models have the potential to provide reasonable representation of the subsurface thermal response and can be used to infer model input parameters and process representations. The models for soil thermal conductivity and snow distribution were found to be the most sensitive process representations. However, information on lateral flow and snowpack evolution might be needed to constrain model representations of surface hydrology and snow depth.

scholarly journals Evolution of the Arabian Sea Upwelling from the Last Millennium to the Future as Simulated by Earth System Models

Climate ◽  
2021 ◽  
Vol 9 (5) ◽  
pp. 72
Author(s):  
Xing Yi ◽  
Birgit Hünicke ◽  
Eduardo Zorita
Keyword(s):  
Arabian Sea ◽  
Positive Trend ◽  
Last Millennium ◽  
Earth System ◽  
System Models ◽  
Sediment Records ◽  
The Future ◽  
Water Stratification ◽  
Arabian Sea Upwelling ◽  
Earth System Models

Arabian Sea upwelling in the past has been generally studied based on the sediment records. We apply two earth system models and analyze the simulated water vertical velocity to investigate coastal upwelling in the western Arabian Sea over the last millennium. In addition, two models with slightly different configurations are also employed to study the upwelling in the 21st century under the strongest and the weakest greenhouse gas emission scenarios. With a negative long-term trend caused by the orbital forcing of the models, the upwelling over the last millennium is found to be closely correlated with the sea surface temperature, the Indian summer Monsoon and the sediment records. The future upwelling under the Representative Concentration Pathway (RCP) 8.5 scenario reveals a negative trend, in contrast with the positive trend displayed by the upwelling favorable along-shore winds. Therefore, it is likely that other factors, like water stratification in the upper ocean layers caused by the stronger surface warming, overrides the effect from the upwelling favorable wind. No significant trend is found for the upwelling under the RCP2.6 scenario, which is likely due to a compensation between the opposing effects of the increase in upwelling favorable winds and the water stratification.

GFDL’s ESM2 Global Coupled Climate–Carbon Earth System Models. Part I: Physical Formulation and Baseline Simulation Characteristics

2012 ◽  
Vol 25 (19) ◽  
pp. 6646-6665 ◽  
Author(s):  
John P. Dunne ◽  
Jasmin G. John ◽  
Alistair J. Adcroft ◽  
Stephen M. Griffies ◽  
Robert W. Hallberg ◽  
...  
Keyword(s):  
Climate Changes ◽  
Southern Oscillation ◽  
Heat Content ◽  
Ocean Dynamics ◽  
Thermocline Depth ◽  
Earth System ◽  
System Models ◽  
Model Version ◽  
Earth System Models

Abstract The physical climate formulation and simulation characteristics of two new global coupled carbon–climate Earth System Models, ESM2M and ESM2G, are described. These models demonstrate similar climate fidelity as the Geophysical Fluid Dynamics Laboratory’s previous Climate Model version 2.1 (CM2.1) while incorporating explicit and consistent carbon dynamics. The two models differ exclusively in the physical ocean component; ESM2M uses Modular Ocean Model version 4p1 with vertical pressure layers while ESM2G uses Generalized Ocean Layer Dynamics with a bulk mixed layer and interior isopycnal layers. Differences in the ocean mean state include the thermocline depth being relatively deep in ESM2M and relatively shallow in ESM2G compared to observations. The crucial role of ocean dynamics on climate variability is highlighted in El Niño–Southern Oscillation being overly strong in ESM2M and overly weak in ESM2G relative to observations. Thus, while ESM2G might better represent climate changes relating to total heat content variability given its lack of long-term drift, gyre circulation, and ventilation in the North Pacific, tropical Atlantic, and Indian Oceans, and depth structure in the overturning and abyssal flows, ESM2M might better represent climate changes relating to surface circulation given its superior surface temperature, salinity, and height patterns, tropical Pacific circulation and variability, and Southern Ocean dynamics. The overall assessment is that neither model is fundamentally superior to the other, and that both models achieve sufficient fidelity to allow meaningful climate and earth system modeling applications. This affords the ability to assess the role of ocean configuration on earth system interactions in the context of two state-of-the-art coupled carbon–climate models.

Assessing the capacity of Earth System Models to simulate spatiotemporal variability in fire weather indicators

2021 ◽  
Author(s):  
Carolina Gallo Granizo ◽  
Jonathan Eden ◽  
Bastien Dieppois ◽  
Matthew Blackett
Keyword(s):  
Climate Change ◽  
Spatial Scales ◽  
Fire Danger ◽  
Spatiotemporal Variability ◽  
Global Evaluation ◽  
Fire Weather ◽  
Earth System ◽  
Fire Weather Index ◽  
System Models ◽  
Earth System Models

<p>Weather and climate play an important role in shaping global fire regimes and geographical distributions of burnable areas. At the global scale, fire danger is likely to increase in the near future due to warmer temperatures and changes in precipitation patterns, as projected by the Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC). There is a need to develop the most reliable projections of future climate-driven fire danger to enable decision makers and forest managers to take both targeted proactive actions and to respond to future fire events.</p><p>Climate change projections generated by Earth System Models (ESMs) provide the most important basis for understanding past, present and future changes in the climate system and its impacts. ESMs are, however, subject to systematic errors and biases, which are not fully taken into account when developing risk scenarios for wild fire activity. Projections of climate-driven fire danger have often been limited to the use of single models or the mean of multi-model ensembles, and compared to a single set of observational data (e.g. one index derived from one reanalysis).</p><p>Here, a comprehensive global evaluation of the representation of a series of fire weather indicators in the latest generation of ESMs is presented. Seven fire weather indices from the Canadian Forest Fire Weather Index System were generated using daily fields realisations simulated by 25 ESMs from the 6<sup>th</sup> Coupled Model Intercomparison Project (CMIP6). With reference to observational and reanalysis datasets, we quantify the capacity of each model to realistically simulate the variability, magnitude and spatial extent of fire danger. The highest-performing models are identified and, subsequently, the limitations of combining models based on independency and equal performance when generating fire danger projections are discussed. To conclude, recommendations are given for the development of user- and policy-driven model evaluation at spatial scales relevant for decision-making and forest management.</p>

Sign in / Sign up

Export Citation Format

Share Document