Articles | Volume 14, issue 7
https://doi.org/10.5194/gmd-14-4495-2021
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/gmd-14-4495-2021
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Methods for assessment of models
| 
22 Jul 2021
Methods for assessment of models |
| 22 Jul 2021
| 22 Jul 2021
Testing the reliability of interpretable neural networks in geoscience using the Madden–Julian oscillation
Benjamin A. Toms
CORRESPONDING AUTHOR
Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado, USA
Karthik Kashinath
Lawrence Berkeley National Laboratory, Berkeley, California, USA
Prabhat
Lawrence Berkeley National Laboratory, Berkeley, California, USA
Lawrence Berkeley National Laboratory, Berkeley, California, USA
Department of Agricultural and Environmental Sciences, University of California, Davis, Davis, California, USA
Related authors
No articles found.
Ashesh Chattopadhyay, Mustafa Mustafa, Pedram Hassanzadeh, Eviatar Bach, and Karthik Kashinath
Geosci. Model Dev., 15, 2221–2237, https://doi.org/10.5194/gmd-15-2221-2022, https://doi.org/10.5194/gmd-15-2221-2022, 2022
Short summary
Short summary
There is growing interest in data-driven weather forecasting, i.e., to predict the weather by using a deep neural network that learns from the evolution of past atmospheric patterns. Here, we propose three components to add to the current data-driven weather forecast models to improve their performance. These components involve a feature that incorporates physics into the neural network, a method to add data assimilation, and an algorithm to use several different time intervals in the forecast.
Prabhat, Karthik Kashinath, Mayur Mudigonda, Sol Kim, Lukas Kapp-Schwoerer, Andre Graubner, Ege Karaismailoglu, Leo von Kleist, Thorsten Kurth, Annette Greiner, Ankur Mahesh, Kevin Yang, Colby Lewis, Jiayi Chen, Andrew Lou, Sathyavat Chandran, Ben Toms, Will Chapman, Katherine Dagon, Christine A. Shields, Travis O'Brien, Michael Wehner, and William Collins
Geosci. Model Dev., 14, 107–124, https://doi.org/10.5194/gmd-14-107-2021, https://doi.org/10.5194/gmd-14-107-2021, 2021
Short summary
Short summary
Detecting extreme weather events is a crucial step in understanding how they change due to climate change. Deep learning (DL) is remarkable at pattern recognition; however, it works best only when labeled datasets are available. We create
ClimateNet– an expert-labeled curated dataset – to train a DL model for detecting weather events and predicting changes in extreme precipitation. This work paves the way for DL-based automated, high-fidelity, and highly precise analytics of climate data.
Travis A. O'Brien, Mark D. Risser, Burlen Loring, Abdelrahman A. Elbashandy, Harinarayan Krishnan, Jeffrey Johnson, Christina M. Patricola, John P. O'Brien, Ankur Mahesh, Prabhat, Sarahí Arriaga Ramirez, Alan M. Rhoades, Alexander Charn, Héctor Inda Díaz, and William D. Collins
Geosci. Model Dev., 13, 6131–6148, https://doi.org/10.5194/gmd-13-6131-2020, https://doi.org/10.5194/gmd-13-6131-2020, 2020
Short summary
Short summary
Researchers utilize various algorithms to identify extreme weather features in climate data, and we seek to answer this question: given a
plausibleweather event detector, how does uncertainty in the detector impact scientific results? We generate a suite of statistical models that emulate expert identification of weather features. We find that the connection between El Niño and atmospheric rivers – a specific extreme weather type – depends systematically on the design of the detector.
Grzegorz Muszynski, Karthik Kashinath, Vitaliy Kurlin, Michael Wehner, and Prabhat
Geosci. Model Dev., 12, 613–628, https://doi.org/10.5194/gmd-12-613-2019, https://doi.org/10.5194/gmd-12-613-2019, 2019
Short summary
Short summary
We present the automated method for recognizing atmospheric rivers in climate data, i.e., climate model output and reanalysis product. The method is based on topological data analysis and machine learning, both of which are powerful tools that the climate science community often does not use. An advantage of the proposed method is that it is free of selection of subjective threshold conditions on a physical variable. This method is also suitable for rapidly analyzing large amounts of data.
Christine A. Shields, Jonathan J. Rutz, Lai-Yung Leung, F. Martin Ralph, Michael Wehner, Brian Kawzenuk, Juan M. Lora, Elizabeth McClenny, Tashiana Osborne, Ashley E. Payne, Paul Ullrich, Alexander Gershunov, Naomi Goldenson, Bin Guan, Yun Qian, Alexandre M. Ramos, Chandan Sarangi, Scott Sellars, Irina Gorodetskaya, Karthik Kashinath, Vitaliy Kurlin, Kelly Mahoney, Grzegorz Muszynski, Roger Pierce, Aneesh C. Subramanian, Ricardo Tome, Duane Waliser, Daniel Walton, Gary Wick, Anna Wilson, David Lavers, Prabhat, Allison Collow, Harinarayan Krishnan, Gudrun Magnusdottir, and Phu Nguyen
Geosci. Model Dev., 11, 2455–2474, https://doi.org/10.5194/gmd-11-2455-2018, https://doi.org/10.5194/gmd-11-2455-2018, 2018
Short summary
Short summary
ARTMIP (Atmospheric River Tracking Method Intercomparison Project) is a community effort with the explicit goal of understanding the uncertainties, and the implications of those uncertainties, in atmospheric river science solely due to detection algorithm. ARTMIP strives to quantify these differences and provide guidance on appropriate algorithmic choices for the science question posed. Project goals, experimental design, and preliminary results are provided.
Related subject area
Climate and Earth system modeling
Structural k-means (S k-means) and clustering uncertainty evaluation framework (CUEF) for mining climate data
The emergence of the Gulf Stream and interior western boundary as key regions to constrain the future North Atlantic carbon uptake
Evaluating wind profiles in a numerical weather prediction model with Doppler lidar
Evaluation of bias correction methods for a multivariate drought index: case study of the Upper Jhelum Basin
The impact of lateral boundary forcing in the CORDEX-Africa ensemble over southern Africa
Effects of complex terrain on the shortwave radiative balance: a sub-grid-scale parameterization for the GFDL Earth System Model version 4.1
Understanding AMOC stability: the North Atlantic Hosing Model Intercomparison Project
Assessing methods for representing soil heterogeneity through a flexible approach within the Joint UK Land Environment Simulator (JULES) at version 3.4.1
Nudging allows direct evaluation of coupled climate models with in situ observations: a case study from the MOSAiC expedition
Importance of ice nucleation and precipitation on climate with the Parameterization of Unified Microphysics Across Scales version 1 (PUMASv1)
UKESM1.1: development and evaluation of an updated configuration of the UK Earth System Model
Porting the WAVEWATCH III (v6.07) wave action source terms to GPU
Yeti 1.0: a generalized framework for constructing bottom-up emission inventories from traffic sources at road-link resolutions
Analysis of systematic biases in tropospheric hydrostatic delay models and construction of a correction model
A new precipitation emulator (PREMU v1.0) for lower-complexity models
Simulating marine neodymium isotope distributions using Nd v1.0 coupled to the ocean component of the FAMOUS–MOSES1 climate model: sensitivities to reversible scavenging efficiency and benthic source distributions
CMIP6 simulations with the compact Earth system model OSCAR v3.1
Application of a satellite-retrieved sheltering parameterization (v1.0) for dust event simulation with WRF-Chem v4.1
Various ways of using Empirical Orthogonal Functions for Climate Model evaluation
The pseudo-global-warming (PGW) approach: methodology, software package PGW4ERA5 v1.1, validation, and sensitivity analyses
AttentionFire_v1.0: interpretable machine learning fire model for burned-area predictions over tropics
Cell tracking of convective rainfall: sensitivity of climate-change signal to tracking algorithm and cell definition (Cell-TAO v1.0)
ICON-Sapphire: simulating the components of the Earth system and their interactions at kilometer and subkilometer scales
Ocean Modeling with Adaptive REsolution (OMARE; version 1.0) – refactoring the NEMO model (version 4.0.1) with the parallel computing framework of JASMIN – Part 1: Adaptive grid refinement in an idealized double-gyre case
Monthly-scale extended predictions using the atmospheric model coupled with a slab ocean
stoPET v1.0: a stochastic potential evapotranspiration generator for simulation of climate change impacts
URANOS v1.0 – the Ultra Rapid Adaptable Neutron-Only Simulation for Environmental Research
Combining regional mesh refinement with vertically enhanced physics to target marine stratocumulus biases as demonstrated in the Energy Exascale Earth System Model version 1
Evaluation of native Earth system model output with ESMValTool v2.6.0
The sea level simulator v1.0: a model for integration of mean sea level change and sea level extremes into a joint probabilistic framework
WRF–ML v1.0: a bridge between WRF v4.3 and machine learning parameterizations and its application to atmospheric radiative transfer
The Euro-Mediterranean Center on Climate Change (CMCC) decadal prediction system
Climate impacts of parameterizing subgrid variation and partitioning of land surface heat fluxes to the atmosphere with the NCAR CESM1.2
Accelerated photosynthesis routine in LPJmL4
C-Coupler3.0: an integrated coupler infrastructure for Earth system modeling
Improving scalability of Earth system models through coarse-grained component concurrency – a case study with the ICON v2.6.5 modelling system
Temperature forecasting by deep learning methods
Pathfinder v1.0.1: a Bayesian-inferred simple carbon–climate model to explore climate change scenarios
Inclusion of a cold hardening scheme to represent frost tolerance is essential to model realistic plant hydraulics in the Arctic–boreal zone in CLM5.0-FATES-Hydro
Implementation and evaluation of the GEOS-Chem chemistry module version 13.1.2 within the Community Earth System Model v2.1
Assessment of JSBACHv4.30 as a land component of ICON-ESM-V1 in comparison to its predecessor JSBACHv3.2 of MPI-ESM1.2
Global biomass burning fuel consumption and emissions at 500 m spatial resolution based on the Global Fire Emissions Database (GFED)
Testing the reconstruction of modelled particulate organic carbon from surface ecosystem components using PlankTOM12 and Machine Learning
Impact of increased resolution on the representation of the Canary upwelling system in climate models
Assessing Responses and Impacts of Solar climate intervention on the Earth system with stratospheric aerosol injection (ARISE-SAI): protocol and initial results from the first simulations
Introducing the VIIRS-based Fire Emission Inventory version 0 (VFEIv0)
Impact of physical parameterizations on wind simulation with WRF V3.9.1.1 under stable conditions at planetary boundary layer gray-zone resolution: a case study over the coastal regions of North China
Advancing precipitation prediction using a new-generation storm-resolving model framework – SIMA-MPAS (V1.0): a case study over the western United States
SURFER v2.0: a flexible and simple model linking anthropogenic CO2 emissions and solar radiation modification to ocean acidification and sea level rise
A new bootstrap technique to quantify uncertainty in estimates of ground surface temperature and ground heat flux histories from geothermal data
Quang-Van Doan, Toshiyuki Amagasa, Thanh-Ha Pham, Takuto Sato, Fei Chen, and Hiroyuki Kusaka
Geosci. Model Dev., 16, 2215–2233, https://doi.org/10.5194/gmd-16-2215-2023, https://doi.org/10.5194/gmd-16-2215-2023, 2023
Short summary
Short summary
This study proposes (i) the structural k-means (S k-means) algorithm for clustering spatiotemporally structured climate data and (ii) the clustering uncertainty evaluation framework (CUEF) based on the mutual-information concept.
Nadine Goris, Klaus Johannsen, and Jerry Tjiputra
Geosci. Model Dev., 16, 2095–2117, https://doi.org/10.5194/gmd-16-2095-2023, https://doi.org/10.5194/gmd-16-2095-2023, 2023
Short summary
Short summary
Climate projections of a high-CO2 future are highly uncertain. A new study provides a novel approach to identifying key regions that dynamically explain the model uncertainty. To yield an accurate estimate of the future North Atlantic carbon uptake, we find that a correct simulation of the upper- and interior-ocean volume transport at 25–30° N is key. However, results indicate that models rarely perform well for both indicators and point towards inconsistencies within the model ensemble.
Pyry Pentikäinen, Ewan J. O'Connor, and Pablo Ortiz-Amezcua
Geosci. Model Dev., 16, 2077–2094, https://doi.org/10.5194/gmd-16-2077-2023, https://doi.org/10.5194/gmd-16-2077-2023, 2023
Short summary
Short summary
We used Doppler lidar to evaluate the wind profiles generated by a weather forecast model. We first compared the Doppler lidar observations with co-located radiosonde profiles, and they agree well. The model performs best over marine and coastal locations. Larger errors were seen in locations where the surface was more complex, especially in the wind direction. Our results show that Doppler lidar is a suitable instrument for evaluating the boundary layer wind profiles in atmospheric models.
Rubina Ansari, Ana Casanueva, Muhammad Usman Liaqat, and Giovanna Grossi
Geosci. Model Dev., 16, 2055–2076, https://doi.org/10.5194/gmd-16-2055-2023, https://doi.org/10.5194/gmd-16-2055-2023, 2023
Short summary
Short summary
Bias correction (BC) has become indispensable to climate model output as a post-processing step to render output more useful for impact assessment studies. The current work presents a comparison of different state-of-the-art BC methods (univariate and multivariate) and BC approaches (direct and component-wise) for climate model simulations from three initiatives (CMIP6, CORDEX, and CORDEX-CORE) for a multivariate drought index (i.e., standardized precipitation evapotranspiration index).
Maria Chara Karypidou, Stefan Pieter Sobolowski, Lorenzo Sangelantoni, Grigory Nikulin, and Eleni Katragkou
Geosci. Model Dev., 16, 1887–1908, https://doi.org/10.5194/gmd-16-1887-2023, https://doi.org/10.5194/gmd-16-1887-2023, 2023
Short summary
Short summary
Southern Africa is listed among the climate change hotspots; hence, accurate climate change information is vital for the optimal preparedness of local communities. In this work we assess the degree to which regional climate models (RCMs) are influenced by the global climate models (GCMs) from which they receive their lateral boundary forcing. We find that although GCMs exert a strong impact on RCMs, RCMs are still able to display substantial improvement relative to the driving GCMs.
Enrico Zorzetto, Sergey Malyshev, Nathaniel Chaney, David Paynter, Raymond Menzel, and Elena Shevliakova
Geosci. Model Dev., 16, 1937–1960, https://doi.org/10.5194/gmd-16-1937-2023, https://doi.org/10.5194/gmd-16-1937-2023, 2023
Short summary
Short summary
In this paper we develop a methodology to model the spatial distribution of solar radiation received by land over mountainous terrain. The approach is designed to be used in Earth system models, where coarse grid cells hinder the description of fine-scale land–atmosphere interactions. We adopt a clustering algorithm to partition the land domain into a set of homogeneous sub-grid
tiles, and for each tile we evaluate solar radiation received by land based on terrain properties.
Laura C. Jackson, Eduardo Alastrué de Asenjo, Katinka Bellomo, Gokhan Danabasoglu, Helmuth Haak, Aixue Hu, Johann Jungclaus, Warren Lee, Virna L. Meccia, Oleg Saenko, Andrew Shao, and Didier Swingedouw
Geosci. Model Dev., 16, 1975–1995, https://doi.org/10.5194/gmd-16-1975-2023, https://doi.org/10.5194/gmd-16-1975-2023, 2023
Short summary
Short summary
The Atlantic meridional overturning circulation (AMOC) has an important impact on the climate. There are theories that freshening of the ocean might cause the AMOC to cross a tipping point (TP) beyond which recovery is difficult; however, it is unclear whether TPs exist in global climate models. Here, we outline a set of experiments designed to explore AMOC tipping points and sensitivity to additional freshwater input as part of the North Atlantic Hosing Model Intercomparison Project (NAHosMIP).
Heather S. Rumbold, Richard J. J. Gilham, and Martin J. Best
Geosci. Model Dev., 16, 1875–1886, https://doi.org/10.5194/gmd-16-1875-2023, https://doi.org/10.5194/gmd-16-1875-2023, 2023
Short summary
Short summary
The Joint UK Land Environment Simulator (JULES) uses a tiled representation of land cover but can only model a single dominant soil type within a grid box; hence there is no representation of sub-grid soil heterogeneity. This paper evaluates a new surface–soil tiling scheme in JULES and demonstrates the impacts of the scheme using several soil tiling approaches. Results show that soil tiling has an impact on the water and energy exchanges due to the way vegetation accesses the soil moisture.
Felix Pithan, Marylou Athanase, Sandro Dahlke, Antonio Sánchez-Benítez, Matthew D. Shupe, Anne Sledd, Jan Streffing, Gunilla Svensson, and Thomas Jung
Geosci. Model Dev., 16, 1857–1873, https://doi.org/10.5194/gmd-16-1857-2023, https://doi.org/10.5194/gmd-16-1857-2023, 2023
Short summary
Short summary
Evaluating climate models usually requires long observational time series, but we present a method that also works for short field campaigns. We compare climate model output to observations from the MOSAiC expedition in the central Arctic Ocean. All models show how the arrival of a warm air mass warms the Arctic in April 2020, but two models do not show the response of snow temperature to the diurnal cycle. One model has too little liquid water and too much ice in clouds during cold days.
Andrew Gettelman, Hugh Morrison, Trude Eidhammer, Katherine Thayer-Calder, Jian Sun, Richard Forbes, Zachary McGraw, Jiang Zhu, Trude Storelvmo, and John Dennis
Geosci. Model Dev., 16, 1735–1754, https://doi.org/10.5194/gmd-16-1735-2023, https://doi.org/10.5194/gmd-16-1735-2023, 2023
Short summary
Short summary
Clouds are a critical part of weather and climate prediction. In this work, we document updates and corrections to the description of clouds used in several Earth system models. These updates include the ability to run the scheme on graphics processing units (GPUs), changes to the numerical description of precipitation, and a correction to the ice number. There are big improvements in the computational performance that can be achieved with GPU acceleration.
Jane P. Mulcahy, Colin G. Jones, Steven T. Rumbold, Till Kuhlbrodt, Andrea J. Dittus, Edward W. Blockley, Andrew Yool, Jeremy Walton, Catherine Hardacre, Timothy Andrews, Alejandro Bodas-Salcedo, Marc Stringer, Lee de Mora, Phil Harris, Richard Hill, Doug Kelley, Eddy Robertson, and Yongming Tang
Geosci. Model Dev., 16, 1569–1600, https://doi.org/10.5194/gmd-16-1569-2023, https://doi.org/10.5194/gmd-16-1569-2023, 2023
Short summary
Short summary
Recent global climate models simulate historical global mean surface temperatures which are too cold, possibly to due to excessive aerosol cooling. This raises questions about the models' ability to simulate important climate processes and reduces confidence in future climate predictions. We present a new version of the UK Earth System Model, which has an improved aerosols simulation and a historical temperature record. Interestingly, the long-term response to CO2 remains largely unchanged.
Olawale James Ikuyajolu, Luke Van Roekel, Steven R. Brus, Erin E. Thomas, Yi Deng, and Sarat Sreepathi
Geosci. Model Dev., 16, 1445–1458, https://doi.org/10.5194/gmd-16-1445-2023, https://doi.org/10.5194/gmd-16-1445-2023, 2023
Short summary
Short summary
Wind-generated waves play an important role in modifying physical processes at the air–sea interface, but they have been traditionally excluded from climate models due to the high computational cost of running spectral wave models for climate simulations. To address this, our work identified and accelerated the computationally intensive section of WAVEWATCH III on GPU using OpenACC. This allows for high-resolution modeling of atmosphere–wave–ocean feedbacks in century-scale climate integrations.
Edward C. Chan, Joana Leitão, Andreas Kerschbaumer, and Timothy M. Butler
Geosci. Model Dev., 16, 1427–1444, https://doi.org/10.5194/gmd-16-1427-2023, https://doi.org/10.5194/gmd-16-1427-2023, 2023
Short summary
Short summary
Yeti is a Handbook Emission Factors for Road Transport-based traffic emission inventory written in the Python 3 scripting language, which adopts a generalized treatment for activity data using traffic information of varying levels of detail introduced in a systematic and consistent manner, with the ability to maximize reusability. Thus, Yeti has been conceived and implemented with a high degree of data and process symmetry, allowing scalable and flexible execution while affording ease of use.
Haopeng Fan, Siran Li, Zhongmiao Sun, Guorui Xiao, Xinxing Li, and Xiaogang Liu
Geosci. Model Dev., 16, 1345–1358, https://doi.org/10.5194/gmd-16-1345-2023, https://doi.org/10.5194/gmd-16-1345-2023, 2023
Short summary
Short summary
The traditional tropospheric zenith hydrostatic delay (ZHD) model's bias is usually thought negligible, yet it still reaches 10 mm sometimes and would lead to millimeter-level position errors for space geodetic observations. Therefore, we analyzed the bias’ characteristics and present a grid model to correct the traditional ZHD formula. When verifying the efficiency based on data from the ECMWF (European Centre for Medium-Range Weather Forecasts), ZHD biases were rectified by ~50 %.
Gang Liu, Shushi Peng, Chris Huntingford, and Yi Xi
Geosci. Model Dev., 16, 1277–1296, https://doi.org/10.5194/gmd-16-1277-2023, https://doi.org/10.5194/gmd-16-1277-2023, 2023
Short summary
Short summary
Due to computational limits, lower-complexity models (LCMs) were developed as a complementary tool for accelerating comprehensive Earth system models (ESMs) but still lack a good precipitation emulator for LCMs. Here, we developed a data-calibrated precipitation emulator (PREMU), a computationally effective way to better estimate historical and simulated precipitation by current ESMs. PREMU has potential applications related to land surface processes and their interactions with climate change.
Suzanne Robinson, Ruza F. Ivanovic, Lauren J. Gregoire, Julia Tindall, Tina van de Flierdt, Yves Plancherel, Frerk Pöppelmeier, Kazuyo Tachikawa, and Paul J. Valdes
Geosci. Model Dev., 16, 1231–1264, https://doi.org/10.5194/gmd-16-1231-2023, https://doi.org/10.5194/gmd-16-1231-2023, 2023
Short summary
Short summary
We present the implementation of neodymium (Nd) isotopes into the ocean model of FAMOUS (Nd v1.0). Nd fluxes from seafloor sediment and incorporation of Nd onto sinking particles represent the major global sources and sinks, respectively. However, model–data mismatch in the North Pacific and northern North Atlantic suggest that certain reactive components of the sediment interact the most with seawater. Our results are important for interpreting Nd isotopes in terms of ocean circulation.
Yann Quilcaille, Thomas Gasser, Philippe Ciais, and Olivier Boucher
Geosci. Model Dev., 16, 1129–1161, https://doi.org/10.5194/gmd-16-1129-2023, https://doi.org/10.5194/gmd-16-1129-2023, 2023
Short summary
Short summary
The model OSCAR is a simple climate model, meaning its representation of the Earth system is simplified but calibrated on models of higher complexity. Here, we diagnose its latest version using a total of 99 experiments in a probabilistic framework and under observational constraints. OSCAR v3.1 shows good agreement with observations, complex Earth system models and emerging properties. Some points for improvements are identified, such as the ocean carbon cycle.
Sandra L. LeGrand, Theodore W. Letcher, Gregory S. Okin, Nicholas P. Webb, Alex R. Gallagher, Saroj Dhital, Taylor S. Hodgdon, Nancy P. Ziegler, and Michelle L. Michaels
Geosci. Model Dev., 16, 1009–1038, https://doi.org/10.5194/gmd-16-1009-2023, https://doi.org/10.5194/gmd-16-1009-2023, 2023
Short summary
Short summary
Ground cover affects dust emissions by reducing wind flow over the immediate soil surface. This study reviews a method for estimating ground cover effects on wind erosion from satellite-detected terrain shadows. We conducted a case study for a US dust event using the Weather Research and Forecasting with Chemistry (WRF-Chem) model. Adding the shadow-based method for ground cover effects markedly improved simulated results and may lead to better dust modeling outcomes in vegetated drylands.
Rasmus E. Benestad, Abdelkader Mezghani, Julia Lutz, Andreas Dobler, Kajsa M. Parding, and Oskar A. Landgren
EGUsphere, https://doi.org/10.5194/egusphere-2022-1385, https://doi.org/10.5194/egusphere-2022-1385, 2023
Short summary
Short summary
A mathematical method known as 'common EOFs' is not widely used within the climate research community, but they offer innovative ways of evaluating climate models. We show how they can be used to evaluate large ensembles of global climate model simulations and distill information about their ability to reproduce salient features of the regional climate. We can say they represent a kind of machine learning (ML) for dealing with "Big data".
Roman Brogli, Christoph Heim, Jonas Mensch, Silje Lund Sørland, and Christoph Schär
Geosci. Model Dev., 16, 907–926, https://doi.org/10.5194/gmd-16-907-2023, https://doi.org/10.5194/gmd-16-907-2023, 2023
Short summary
Short summary
The pseudo-global-warming (PGW) approach is a downscaling methodology that imposes the large-scale GCM-based climate change signal on the boundary conditions of a regional climate simulation. It offers several benefits in comparison to conventional downscaling. We present a detailed description of the methodology, provide companion software to facilitate the preparation of PGW simulations, and present validation and sensitivity studies.
Fa Li, Qing Zhu, William J. Riley, Lei Zhao, Li Xu, Kunxiaojia Yuan, Min Chen, Huayi Wu, Zhipeng Gui, Jianya Gong, and James T. Randerson
Geosci. Model Dev., 16, 869–884, https://doi.org/10.5194/gmd-16-869-2023, https://doi.org/10.5194/gmd-16-869-2023, 2023
Short summary
Short summary
We developed an interpretable machine learning model to predict sub-seasonal and near-future wildfire-burned area over African and South American regions. We found strong time-lagged controls (up to 6–8 months) of local climate wetness on burned areas. A skillful use of such time-lagged controls in machine learning models results in highly accurate predictions of wildfire-burned areas; this will also help develop relevant early-warning and management systems for tropical wildfires.
Edmund P. Meredith, Uwe Ulbrich, and Henning W. Rust
Geosci. Model Dev., 16, 851–867, https://doi.org/10.5194/gmd-16-851-2023, https://doi.org/10.5194/gmd-16-851-2023, 2023
Short summary
Short summary
Cell-tracking algorithms allow for the study of properties of a convective cell across its lifetime and, in particular, how these respond to climate change. We investigated whether the design of the algorithm can affect the magnitude of the climate-change signal. The algorithm's criteria for identifying a cell were found to have a strong impact on the warming response. The sensitivity of the warming response to different algorithm settings and cell types should thus be fully explored.
Cathy Hohenegger, Peter Korn, Leonidas Linardakis, René Redler, Reiner Schnur, Panagiotis Adamidis, Jiawei Bao, Swantje Bastin, Milad Behravesh, Martin Bergemann, Joachim Biercamp, Hendryk Bockelmann, Renate Brokopf, Nils Brüggemann, Lucas Casaroli, Fatemeh Chegini, George Datseris, Monika Esch, Geet George, Marco Giorgetta, Oliver Gutjahr, Helmuth Haak, Moritz Hanke, Tatiana Ilyina, Thomas Jahns, Johann Jungclaus, Marcel Kern, Daniel Klocke, Lukas Kluft, Tobias Kölling, Luis Kornblueh, Sergey Kosukhin, Clarissa Kroll, Junhong Lee, Thorsten Mauritsen, Carolin Mehlmann, Theresa Mieslinger, Ann Kristin Naumann, Laura Paccini, Angel Peinado, Divya Sri Praturi, Dian Putrasahan, Sebastian Rast, Thomas Riddick, Niklas Roeber, Hauke Schmidt, Uwe Schulzweida, Florian Schütte, Hans Segura, Radomyra Shevchenko, Vikram Singh, Mia Specht, Claudia Christine Stephan, Jin-Song von Storch, Raphaela Vogel, Christian Wengel, Marius Winkler, Florian Ziemen, Jochem Marotzke, and Bjorn Stevens
Geosci. Model Dev., 16, 779–811, https://doi.org/10.5194/gmd-16-779-2023, https://doi.org/10.5194/gmd-16-779-2023, 2023
Short summary
Short summary
Models of the Earth system used to understand climate and predict its change typically employ a grid spacing of about 100 km. Yet, many atmospheric and oceanic processes occur on much smaller scales. In this study, we present a new model configuration designed for the simulation of the components of the Earth system and their interactions at kilometer and smaller scales, allowing an explicit representation of the main drivers of the flow of energy and matter by solving the underlying equations.
Yan Zhang, Xuantong Wang, Yuhao Sun, Chenhui Ning, Shiming Xu, Hengbin An, Dehong Tang, Hong Guo, Hao Yang, Ye Pu, Bo Jiang, and Bin Wang
Geosci. Model Dev., 16, 679–704, https://doi.org/10.5194/gmd-16-679-2023, https://doi.org/10.5194/gmd-16-679-2023, 2023
Short summary
Short summary
We construct a new ocean model, OMARE, that can carry out multi-scale ocean simulation with adaptive mesh refinement. OMARE is based on the refactorization of NEMO with a third-party, high-performance piece of middleware. We report the porting process and experiments of an idealized western-boundary current system. The new model simulates turbulent and temporally varying mesoscale and submesoscale processes via adaptive refinement. Related topics and future work with OMARE are also discussed.
Zhenming Wang, Shaoqing Zhang, Yishuai Jin, Yinglai Jia, Yangyang Yu, Yang Gao, Xiaolin Yu, Mingkui Li, Xiaopei Lin, and Lixin Wu
Geosci. Model Dev., 16, 705–717, https://doi.org/10.5194/gmd-16-705-2023, https://doi.org/10.5194/gmd-16-705-2023, 2023
Short summary
Short summary
To improve the numerical model predictability of monthly extended-range scales, we use the simplified slab ocean model (SOM) to restrict the complicated sea surface temperature (SST) bias from a 3-D dynamical ocean model. As for SST prediction, whether in space or time, the WRF-SOM is verified to have better performance than the WRF-ROMS, which has a significant impact on the atmosphere. For extreme weather events such as typhoons, the predictions of WRF-SOM are in good agreement with WRF-ROMS.
Dagmawi Teklu Asfaw, Michael Bliss Singer, Rafael Rosolem, David MacLeod, Mark Cuthbert, Edisson Quichimbo Miguitama, Manuel F. Rios Gaona, and Katerina Michaelides
Geosci. Model Dev., 16, 557–571, https://doi.org/10.5194/gmd-16-557-2023, https://doi.org/10.5194/gmd-16-557-2023, 2023
Short summary
Short summary
stoPET is a new stochastic potential evapotranspiration (PET) generator for the globe at hourly resolution. Many stochastic weather generators are used to generate stochastic rainfall time series; however, no such model exists for stochastically generating plausible PET time series. As such, stoPET represents a significant methodological advance. stoPET generate many realizations of PET to conduct climate studies related to the water balance, agriculture, water resources, and ecology.
Markus Köhli, Martin Schrön, Steffen Zacharias, and Ulrich Schmidt
Geosci. Model Dev., 16, 449–477, https://doi.org/10.5194/gmd-16-449-2023, https://doi.org/10.5194/gmd-16-449-2023, 2023
Short summary
Short summary
In the last decades, Monte Carlo codes were often consulted to study neutrons near the surface. As an alternative for the growing community of CRNS, we developed URANOS. The main model features are tracking of particle histories from creation to detection, detector representations as layers or geometric shapes, a voxel-based geometry model, and material setup based on color codes in ASCII matrices or bitmap images. The entire software is developed in C++ and features a graphical user interface.
Peter A. Bogenschutz, Hsiang-He Lee, Qi Tang, and Takanobu Yamaguchi
Geosci. Model Dev., 16, 335–352, https://doi.org/10.5194/gmd-16-335-2023, https://doi.org/10.5194/gmd-16-335-2023, 2023
Short summary
Short summary
Models that are used to simulate and predict climate often have trouble representing specific cloud types, such as stratocumulus, that are particularly thin in the vertical direction. It has been found that increasing the model resolution can help improve this problem. In this paper, we develop a novel framework that increases the horizontal and vertical resolutions only for areas of the globe that contain stratocumulus, hence reducing the model runtime while providing better results.
Manuel Schlund, Birgit Hassler, Axel Lauer, Bouwe Andela, Patrick Jöckel, Rémi Kazeroni, Saskia Loosveldt Tomas, Brian Medeiros, Valeriu Predoi, Stéphane Sénési, Jérôme Servonnat, Tobias Stacke, Javier Vegas-Regidor, Klaus Zimmermann, and Veronika Eyring
Geosci. Model Dev., 16, 315–333, https://doi.org/10.5194/gmd-16-315-2023, https://doi.org/10.5194/gmd-16-315-2023, 2023
Short summary
Short summary
The Earth System Model Evaluation Tool (ESMValTool) is a community diagnostics and performance metrics tool for routine evaluation of Earth system models. Originally, ESMValTool was designed to process reformatted output provided by large model intercomparison projects like the Coupled Model Intercomparison Project (CMIP). Here, we describe a new extension of ESMValTool that allows for reading and processing native climate model output, i.e., data that have not been reformatted before.
Magnus Hieronymus
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-295, https://doi.org/10.5194/gmd-2022-295, 2023
Revised manuscript accepted for GMD
Short summary
Short summary
A statistical model called the sea level simulator is presented and made freely available. The sea level simulator integrates mean sea level rise and sea level extremes into a joint framework that is useful for flood risk estimation. These flood risk estimates are contingent on probabilities given to different emission scenarios and the length of the planning period. The model is also useful for uncertainty quantifications and in decision and adaptation problems.
Xiaohui Zhong, Zhijian Ma, Yichen Yao, Lifei Xu, Yuan Wu, and Zhibin Wang
Geosci. Model Dev., 16, 199–209, https://doi.org/10.5194/gmd-16-199-2023, https://doi.org/10.5194/gmd-16-199-2023, 2023
Short summary
Short summary
More and more researchers use deep learning models to replace physics-based parameterizations to accelerate weather simulations. However, embedding the ML models within the weather models is difficult as they are implemented in different languages. This work proposes a coupling framework to allow ML-based parameterizations to be coupled with the Weather Research and Forecasting (WRF) model. We also demonstrate using the coupler to couple the ML-based radiation schemes with the WRF model.
Dario Nicolì, Alessio Bellucci, Paolo Ruggieri, Panos J. Athanasiadis, Stefano Materia, Daniele Peano, Giusy Fedele, Riccardo Hénin, and Silvio Gualdi
Geosci. Model Dev., 16, 179–197, https://doi.org/10.5194/gmd-16-179-2023, https://doi.org/10.5194/gmd-16-179-2023, 2023
Short summary
Short summary
Decadal climate predictions, obtained by constraining the initial condition of a dynamical model through a truthful estimate of the observed climate state, provide an accurate assessment of the near-term climate and are useful for informing decision-makers on future climate-related risks. The predictive skill for key variables is assessed from the operational decadal prediction system compared with non-initialized historical simulations so as to quantify the added value of initialization.
Ming Yin, Yilun Han, Yong Wang, Wenqi Sun, Jianbo Deng, Daoming Wei, Ying Kong, and Bin Wang
Geosci. Model Dev., 16, 135–156, https://doi.org/10.5194/gmd-16-135-2023, https://doi.org/10.5194/gmd-16-135-2023, 2023
Short summary
Short summary
All global climate models (GCMs) use the grid-averaged surface heat fluxes to drive the atmosphere, and thus their horizontal variations within the grid cell are averaged out. In this regard, a novel scheme considering the variation and partitioning of the surface heat fluxes within the grid cell is developed. The scheme reduces the long-standing rainfall biases on the southern and eastern margins of the Tibetan Plateau. The performance of key variables at the global scale is also evaluated.
Jenny Niebsch, Werner von Bloh, Kirsten Thonicke, and Ronny Ramlau
Geosci. Model Dev., 16, 17–33, https://doi.org/10.5194/gmd-16-17-2023, https://doi.org/10.5194/gmd-16-17-2023, 2023
Short summary
Short summary
The impacts of climate change require strategies for climate adaptation. Dynamic global vegetation models (DGVMs) are used to study the effects of multiple processes in the biosphere under climate change. There is a demand for a better computational performance of the models. In this paper, the photosynthesis model in the Lund–Potsdam–Jena managed Land DGVM (4.0.002) was examined. We found a better numerical solution of a nonlinear equation. A significant run time reduction was possible.
Li Liu, Chao Sun, Xinzhu Yu, Hao Yu, Qingu Jiang, Xingliang Li, Ruizhe Li, Bin Wang, Xueshun Shen, and Guangwen Yang
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-257, https://doi.org/10.5194/gmd-2022-257, 2023
Revised manuscript accepted for GMD
Short summary
Short summary
C-Coupler3.0 is an integrated coupler infrastructure with new features. i.e., a series of parallel optimization technologies, a common halo-exchange library, a common module-integration framework, a common framework for conveniently developing a weakly coupled ensemble data assimilation system, and a common framework for flexibly inputting and outputting fields in parallel. It is able to handle coupling under much finer resolutions (e.g., more than 100 million horizontal grid cells).
Leonidas Linardakis, Irene Stemmler, Moritz Hanke, Lennart Ramme, Fatemeh Chegini, Tatiana Ilyina, and Peter Korn
Geosci. Model Dev., 15, 9157–9176, https://doi.org/10.5194/gmd-15-9157-2022, https://doi.org/10.5194/gmd-15-9157-2022, 2022
Short summary
Short summary
In Earth system modelling, we are facing the challenge of making efficient use of very large machines, with millions of cores. To meet this challenge we will need to employ multi-level and multi-dimensional parallelism. Component concurrency, being a function parallel technique, offers an additional dimension to the traditional data-parallel approaches. In this paper we examine the behaviour of component concurrency and identify the conditions for its optimal application.
Bing Gong, Michael Langguth, Yan Ji, Amirpasha Mozaffari, Scarlet Stadtler, Karim Mache, and Martin G. Schultz
Geosci. Model Dev., 15, 8931–8956, https://doi.org/10.5194/gmd-15-8931-2022, https://doi.org/10.5194/gmd-15-8931-2022, 2022
Short summary
Short summary
Inspired by the success of deep learning in various domains, we test the applicability of video prediction methods by generative adversarial network (GAN)-based deep learning to predict the 2 m temperature over Europe. Our video prediction models have skill in predicting the diurnal cycle of 2 m temperature up to 12 h ahead. Complemented by probing the relevance of several model parameters, this study confirms the potential of deep learning in meteorological forecasting applications.
Thomas Bossy, Thomas Gasser, and Philippe Ciais
Geosci. Model Dev., 15, 8831–8868, https://doi.org/10.5194/gmd-15-8831-2022, https://doi.org/10.5194/gmd-15-8831-2022, 2022
Short summary
Short summary
We developed a new simple climate model designed to fill a perceived gap within the existing simple climate models by fulfilling three key requirements: calibration using Bayesian inference, the possibility of coupling with integrated assessment models, and the capacity to explore climate scenarios compatible with limiting climate impacts. Here, we describe the model and its calibration using the latest data from complex CMIP6 models and the IPCC AR6, and we assess its performance.
Marius S. A. Lambert, Hui Tang, Kjetil S. Aas, Frode Stordal, Rosie A. Fisher, Yilin Fang, Junyan Ding, and Frans-Jan W. Parmentier
Geosci. Model Dev., 15, 8809–8829, https://doi.org/10.5194/gmd-15-8809-2022, https://doi.org/10.5194/gmd-15-8809-2022, 2022
Short summary
Short summary
In this study, we implement a hardening mortality scheme into CTSM5.0-FATES-Hydro and evaluate how it impacts plant hydraulics and vegetation growth. Our work shows that the hydraulic modifications prescribed by the hardening scheme are necessary to model realistic vegetation growth in cold climates, in contrast to the default model that simulates almost nonexistent and declining vegetation due to abnormally large water loss through the roots.
Thibaud M. Fritz, Sebastian D. Eastham, Louisa K. Emmons, Haipeng Lin, Elizabeth W. Lundgren, Steve Goldhaber, Steven R. H. Barrett, and Daniel J. Jacob
Geosci. Model Dev., 15, 8669–8704, https://doi.org/10.5194/gmd-15-8669-2022, https://doi.org/10.5194/gmd-15-8669-2022, 2022
Short summary
Short summary
We bring the state-of-the-science chemistry module GEOS-Chem into the Community Earth System Model (CESM). We show that some known differences between results from GEOS-Chem and CESM's CAM-chem chemistry module may be due to the configuration of model meteorology rather than inherent differences in the model chemistry. This is a significant step towards a truly modular Earth system model and allows two strong but currently separate research communities to benefit from each other's advances.
Rainer Schneck, Veronika Gayler, Julia E. M. S. Nabel, Thomas Raddatz, Christian H. Reick, and Reiner Schnur
Geosci. Model Dev., 15, 8581–8611, https://doi.org/10.5194/gmd-15-8581-2022, https://doi.org/10.5194/gmd-15-8581-2022, 2022
Short summary
Short summary
The versions of ICON-A and ICON-Land/JSBACHv4 used for this study constitute the first milestone in the development of the new ICON Earth System Model ICON-ESM. JSBACHv4 is the successor of JSBACHv3, and most of the parameterizations of JSBACHv4 are re-implementations from JSBACHv3. We assess and compare the performance of JSBACHv4 and JSBACHv3. Overall, the JSBACHv4 results are as good as JSBACHv3, but both models reveal the same main shortcomings, e.g. the depiction of the leaf area index.
Dave van Wees, Guido R. van der Werf, James T. Randerson, Brendan M. Rogers, Yang Chen, Sander Veraverbeke, Louis Giglio, and Douglas C. Morton
Geosci. Model Dev., 15, 8411–8437, https://doi.org/10.5194/gmd-15-8411-2022, https://doi.org/10.5194/gmd-15-8411-2022, 2022
Short summary
Short summary
We present a global fire emission model based on the GFED model framework with a spatial resolution of 500 m. The higher resolution allowed for a more detailed representation of spatial heterogeneity in fuels and emissions. Specific modules were developed to model, for example, emissions from fire-related forest loss and belowground burning. Results from the 500 m model were compared to GFED4s, showing that global emissions were relatively similar but that spatial differences were substantial.
Anna Denvil-Sommer, Erik T. Buitenhuis, Rainer Kiko, Fabien Lombard, Lionel Guidi, and Corinne Le Quéré
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-224, https://doi.org/10.5194/gmd-2022-224, 2022
Revised manuscript accepted for GMD
Short summary
Short summary
Using outputs of global biogeochemical ocean model and Machine Learning methods we demonstrate that it will be possible to identify linkages between surface environmental and ecosystem structure and the export of carbon to depth by sinking organic particles using real observations. It will be possible to use this knowledge to improve both our understanding of ecosystem dynamics and of their functional representation within models.
Adama Sylla, Emilia Sanchez Gomez, Juliette Mignot, and Jorge López-Parages
Geosci. Model Dev., 15, 8245–8267, https://doi.org/10.5194/gmd-15-8245-2022, https://doi.org/10.5194/gmd-15-8245-2022, 2022
Short summary
Short summary
Increasing model resolution depends on the subdomain of the Canary upwelling considered. In the Iberian Peninsula, the high-resolution (HR) models do not seem to better simulate the upwelling indices, while in Morocco to the Senegalese coast, the HR models show a clear improvement. Thus increasing the resolution of a global climate model does not necessarily have to be the only way to better represent the climate system. There is still much work to be done in terms of physical parameterizations.
Jadwiga H. Richter, Daniele Visioni, Douglas G. MacMartin, David A. Bailey, Nan Rosenbloom, Brian Dobbins, Walker R. Lee, Mari Tye, and Jean-Francois Lamarque
Geosci. Model Dev., 15, 8221–8243, https://doi.org/10.5194/gmd-15-8221-2022, https://doi.org/10.5194/gmd-15-8221-2022, 2022
Short summary
Short summary
Solar climate intervention using stratospheric aerosol injection is a proposed method of reducing global mean temperatures to reduce the worst consequences of climate change. We present a new modeling protocol aimed at simulating a plausible deployment of stratospheric aerosol injection and reproducibility of simulations using other Earth system models: Assessing Responses and Impacts of Solar climate intervention on the Earth system with stratospheric aerosol injection (ARISE-SAI).
Gonzalo A. Ferrada, Meng Zhou, Jun Wang, Alexei Lyapustin, Yujie Wang, Saulo R. Freitas, and Gregory R. Carmichael
Geosci. Model Dev., 15, 8085–8109, https://doi.org/10.5194/gmd-15-8085-2022, https://doi.org/10.5194/gmd-15-8085-2022, 2022
Short summary
Short summary
The smoke from fires is composed of different compounds that interact with the atmosphere and can create poor air-quality episodes. Here, we present a new fire inventory based on satellite observations from the Visible Infrared Imaging Radiometer Suite (VIIRS). We named this inventory the VIIRS-based Fire Emission Inventory (VFEI). Advantages of VFEI are its high resolution (~500 m) and that it provides information for many species. VFEI is publicly available and has provided data since 2012.
Entao Yu, Rui Bai, Xia Chen, and Lifang Shao
Geosci. Model Dev., 15, 8111–8134, https://doi.org/10.5194/gmd-15-8111-2022, https://doi.org/10.5194/gmd-15-8111-2022, 2022
Short summary
Short summary
A large number of simulations are conducted to investigate how different physical parameterization schemes impact surface wind simulations under stable weather conditions over the coastal regions of North China using the Weather Research and Forecasting model with a horizontal grid spacing of 0.5 km. Results indicate that the simulated wind speed is most sensitive to the planetary boundary layer schemes, followed by short-wave/long-wave radiation schemes and microphysics schemes.
Xingying Huang, Andrew Gettelman, William C. Skamarock, Peter Hjort Lauritzen, Miles Curry, Adam Herrington, John T. Truesdale, and Michael Duda
Geosci. Model Dev., 15, 8135–8151, https://doi.org/10.5194/gmd-15-8135-2022, https://doi.org/10.5194/gmd-15-8135-2022, 2022
Short summary
Short summary
We focus on the recent development of a state-of-the-art storm-resolving global climate model and investigate how this next-generation model performs for precipitation prediction over the western USA. Results show realistic representations of precipitation with significantly enhanced snowpack over complex terrains. The model evaluation advances the unified modeling of large-scale forcing constraints and realistic fine-scale features to advance multi-scale climate predictions and changes.
Marina Martínez Montero, Michel Crucifix, Victor Couplet, Nuria Brede, and Nicola Botta
Geosci. Model Dev., 15, 8059–8084, https://doi.org/10.5194/gmd-15-8059-2022, https://doi.org/10.5194/gmd-15-8059-2022, 2022
Short summary
Short summary
We present SURFER, a lightweight model that links CO2 emissions and geoengineering to ocean acidification and sea level rise from glaciers, ocean thermal expansion and Greenland and Antarctic ice sheets. The ice sheet module adequately describes the tipping points of both Greenland and Antarctica. SURFER is understandable, fast, accurate up to several thousands of years, capable of emulating results obtained by state of the art models and well suited for policy analyses.
Francisco José Cuesta-Valero, Hugo Beltrami, Stephan Gruber, Almudena García-García, and J. Fidel González-Rouco
Geosci. Model Dev., 15, 7913–7932, https://doi.org/10.5194/gmd-15-7913-2022, https://doi.org/10.5194/gmd-15-7913-2022, 2022
Short summary
Short summary
Inversions of subsurface temperature profiles provide past long-term estimates of ground surface temperature histories and ground heat flux histories at timescales of decades to millennia. Theses estimates complement high-frequency proxy temperature reconstructions and are the basis for studying continental heat storage. We develop and release a new bootstrap method to derive meaningful confidence intervals for the average surface temperature and heat flux histories from any number of profiles.
Cited articles
Abhik, S. and Hendon, H. H.: Influence of the QBO on the MJO during coupled
model multiweek forecasts, Geophys. Res. Lett., 46, 9213–9221,
2019. a
Adames, Á. F. and Kim, D.: The MJO as a dispersive, convectively coupled
moisture wave: Theory and observations, J. Atmos. Sci.,
73, 913–941, 2016. a
Adames, Á. F. and Wallace, J. M.: Three-dimensional structure and evolution
of the MJO and its relation to the mean flow, J. Atmos.
Sci., 71, 2007–2026, 2014. a
Adames, Á. F. and Wallace, J. M.: Three-dimensional structure and evolution
of the moisture field in the MJO, J. Atmos. Sci., 72,
3733–3754, 2015. a
Bach, S., Binder, A., Montavon, G., Klauschen, F., Müller, K.-R., and
Samek, W.: On pixel-wise explanations for non-linear classifier decisions by
layer-wise relevance propagation, PloS one, 10, e0130140, https://doi.org/10.1371/journal.pone.0130140, 2015. a
Barnes, E. A., Hurrell, J. W., Ebert-Uphoff, I., Anderson, C., and Anderson,
D.: Viewing forced climate patterns through an AI Lens, Geophys. Res.
Lett., 46, 13389–13398, 2019. a
Brenowitz, N. D. and Bretherton, C. S.: Prognostic validation of a neural
network unified physics parameterization, Geophys. Res. Lett., 45,
6289–6298, 2018. a
Chen, T. and Chen, H.: Universal approximation to nonlinear operators by neural
networks with arbitrary activation functions and its application to dynamical
systems, IEEE T. Neural Netw., 6, 911–917, 1995. a
Densmore, C. R., Sanabia, E. R., and Barrett, B. S.: QBO influence on MJO
amplitude over the Maritime Continent: Physical mechanisms and seasonality,
Mon. Weather Rev., 147, 389–406, 2019. a
Ebert-Uphoff, I. and Hilburn, K. A.: Evaluation, Tuning and Interpretation of
Neural Networks for Meteorological Applications, arXiv [preprint],
arXiv:2005.03126, 2020. a, b
Gelaro, R., McCarty, W., Suárez, M. J., Todling, R., Molod, A., Takacs, L., Randles, C. A., Darmenov, A., Bosilovich, M. G., Reichle, R., Wargan, K., Coy, L., Cullather, R., Draper, C., Akella, S., Buchard, V., Conaty, A., da Silva, A. M., Gu, W., Kim, G., Koster, R., Lucchesi, R., Merkova, D., Nielsen, J. E., Partyka, G., Pawson, S., Putman, W., Rienecker, M., Schubert, S. D., Sienkiewicz, M., and Zhao, B.: The
modern-era retrospective analysis for research and applications, version 2
(MERRA-2), J. Climate, 30, 5419–5454, 2017. a, b
Goodfellow, I., Bengio, Y., and Courville, A.: Deep learning, MIT press, available at: https://books.google.com/books?hl=en&lr=&id=omivDQAAQBAJ&oi=fnd&pg=PR5&dq=Goodfellow,+I.,+Bengio,+Y.,+and+Courville,+A.:+Deep+learning&ots=MNO3arszWQ&sig=NLK3dUkGYpMXR4-kZ1z7yD9O-7g#v=onepage&q=Goodfellow%2C I.%2C Bengio%2C Y.%2C and Courville%2C A.%3A Deep learning&f=false (last access: 31 December 2020),
2016. a
Hendon, H. H. and Liebmann, B.: Organization of convection within the
Madden-Julian oscillation, J. Geophys. Res.-Atmos., 99,
8073–8083, 1994. a
Jiang, X., Adames, Á. F., Kim, D., Maloney, E. D., Lin, H., Kim, H., Zhang,
C., DeMott, C. A., and Klingaman, N. P.: Fifty Years of Research on the
Madden-Julian Oscillation: Recent Progress, Challenges, and Perspectives,
J. Geophys. Res.-Atmos., 125, e2019JD030911, https://doi.org/10.1029/2019JD030911, 2020. a
Lee, H.-T. and NOAA CDR Program: NOAA Climate Data Record (CDR) of Daily Outgoing Longwave Radiation (OLR), Version 1.2, NOAA National Climatic Data Center, https://doi.org/10.7289/V5SJ1HH2, 2011. a, b
Liebmann, B. and Smith, C. A.: Description of a complete (interpolated)
outgoing longwave radiation dataset, B. Am. Meteorol.
Soc., 77, 1275–1277, 1996. a
Madden, R. A. and Julian, P. R.: Detection of a 40–50 day oscillation in the
zonal wind in the tropical Pacific, J. Atmos. Sci., 28,
702–708, 1971. a
Martin, Z., Vitart, F., Wang, S., and Sobel, A.: The impact of the stratosphere
on the MJO in a forecast model, J. Geophys. Res.-Atmos., 125, e2019JD032106,
https://doi.org/10.1029/2019JD032106,
2020. a
Matsuno, T.: Quasi-geostrophic motions in the equatorial area, J.
Meteorol. Soc. Jpn. Ser. II, 44, 25–43, 1966. a
McGovern, A., Lagerquist, R., John Gagne, D., Jergensen, G. E., Elmore, K. L.,
Homeyer, C. R., and Smith, T.: Making the black box more transparent:
Understanding the physical implications of machine learning, B.
Am. Meteorol. Soc., 100, 2175–2199, 2019. a
Montavon, G., Lapuschkin, S., Binder, A., Samek, W., and Müller, K.-R.:
Explaining nonlinear classification decisions with deep taylor decomposition,
Pattern Recogn., 65, 211–222, 2017. a
Monteiro, J. M., Adames, Á. F., Wallace, J. M., and Sukhatme, J. S.:
Interpreting the upper level structure of the Madden-Julian oscillation,
Geophys. Res. Lett., 41, 9158–9165, 2014. a
NOAA: MJO indices, available at: https://www.psl.noaa.gov/mjo/mjoindex/, (last access: 31 December 2020. a
Olah, C., Mordvintsev, A., and Schubert, L.: Feature visualization, Distill, 2, e7,
2017. a
Rasp, S., Pritchard, M. S., and Gentine, P.: Deep learning to represent subgrid
processes in climate models, P. Natl. Acad. Sci. USA,
115, 9684–9689, 2018. a
Roundy, P. E. and Frank, W. M.: Applications of a multiple linear regression
model to the analysis of relationships between eastward-and westward-moving
intraseasonal modes, J. Atmos. Sci., 61, 3041–3048,
2004. a
Roundy, P. E., MacRitchie, K., Asuma, J., and Melino, T.: Modulation of the
global atmospheric circulation by combined activity in the Madden–Julian
oscillation and the El Niño–Southern Oscillation during boreal winter,
J. Climate, 23, 4045–4059, 2010. a
Simonyan, K., Vedaldi, A., and Zisserman, A.: Deep inside convolutional
networks: Visualising image classification models and saliency maps, arXiv
[preprint], arXiv:1312.6034, 2013. a
Sobel, A. and Maloney, E.: Moisture modes and the eastward propagation of the
MJO, J. Atmos. Sci., 70, 187–192, 2013. a
Son, S.-W., Lim, Y., Yoo, C., Hendon, H. H., and Kim, J.: Stratospheric control
of the Madden–Julian oscillation, J. Climate, 30, 1909–1922, 2017. a
Toms, B.: Data for GMD 2020-152 (Version V1), Zenodo [Data set], https://doi.org/10.5281/zenodo.3968896, 2020. a
Toms, B. A., Barnes, E. A., Maloney, E. D., and van den Heever, S. C.: The
Global Teleconnection Signature of the Madden-Julian Oscillation and its
Modulation by the Quasi-Biennial Oscillation, J. Geophys.
Res.-Atmos., 25, e2020JD032653,
https://doi.org/10.1029/2020JD032653, 2020b. a, b
Tseng, K.-C., Maloney, E., and Barnes, E.: The consistency of MJO
teleconnection patterns: an explanation using linear Rossby wave theory,
J. Climate, 32, 531–548, 2019. a
Weyn, J. A., Durran, D. R., and Caruana, R.: Can Machines Learn to Predict
Weather? Using Deep Learning to Predict Gridded 500-hPa Geopotential Height
From Historical Weather Data, J. Adv. Model. Earth Sy.,
11, 2680–2693, 2019. a
Yang, D. and Ingersoll, A. P.: Testing the hypothesis that the MJO is a mixed
Rossby–gravity wave packet, J. Atmos. Sci., 68,
226–239, 2011. a
Yang, D. and Ingersoll, A. P.: Triggered convection, gravity waves, and the
MJO: A shallow-water model, J. Atmos. Sci., 70,
2476–2486, 2013. a
Yoo, C. and Son, S.-W.: Modulation of the boreal wintertime Madden-Julian
oscillation by the stratospheric quasi-biennial oscillation, Geophys.
Res. Lett., 43, 1392–1398, 2016. a
Yosinski, J., Clune, J., Nguyen, A., Fuchs, T., and Lipson, H.: Understanding
neural networks through deep visualization, arXiv [preprint], arXiv:1506.06579,
2015. a
Zhang, C. and Dong, M.: Seasonality in the Madden–Julian Oscillation, J.
Climate, 17, 3169–3180,
https://doi.org/10.1175/1520-0442(2004)017<3169:SITMO>2.0.CO;2, 2004. a
Zhang, C. and Zhang, B.: QBO-MJO Connection, J. Geophys. Res.-Atmos., 123, 2957–2967, 2018. a
Zhang, C., Adames, Á., Khouider, B., Wang, B., and Yang, D.: Four Theories
of the Madden-Julian Oscillation, Rev. Geophys., 58.3, e2019RG000685,
https://doi.org/10.1029/2019RG000685, 2020. a
Zhao, C., Li, T., and Zhou, T.: Precursor signals and processes associated with
MJO initiation over the tropical Indian Ocean, J. Climate, 26,
291–307, 2013. a
Short summary
We test whether a type of machine learning called neural networks can be used trustfully within the geosciences. We do so by challenging the networks to understand the spatial patterns of a commonly studied geoscientific phenomenon. The neural networks can correctly identify the spatial patterns, which lends confidence that similar networks can be used for more uncertain problems. The results of this study may give geoscientists confidence when using neural networks in their research.
We test whether a type of machine learning called neural networks can be used trustfully within...
