Articles | Volume 15, issue 5
https://doi.org/10.5194/gmd-15-2221-2022
© Author(s) 2022. 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-15-2221-2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Development and technical paper
| 
16 Mar 2022
Development and technical paper |
| 16 Mar 2022
| 16 Mar 2022
Towards physics-inspired data-driven weather forecasting: integrating data assimilation with a deep spatial-transformer-based U-NET in a case study with ERA5
Ashesh Chattopadhyay
Department of Mechanical Engineering, Rice University, Houston, TX, USA
Lawrence Berkeley National Laboratory, Berkeley, CA, USA
Mustafa Mustafa
Lawrence Berkeley National Laboratory, Berkeley, CA, USA
Department of Mechanical Engineering, Rice University, Houston, TX, USA
Department of Earth, Environmental and Planetary Sciences, Rice University, Houston, TX, USA
Eviatar Bach
Department of Atmospheric and Oceanic Science and Institute for Physical Science and Technology, University of Maryland, College Park, USA
Geosciences Department and Laboratoire de Météorologie Dynamique (CNRS and IPSL), École Normale Supérieure and PSL University, Paris, France
Karthik Kashinath
Lawrence Berkeley National Laboratory, Berkeley, CA, USA
Related authors
Ashesh Chattopadhyay, Pedram Hassanzadeh, and Devika Subramanian
Nonlin. Processes Geophys., 27, 373–389, https://doi.org/10.5194/npg-27-373-2020, https://doi.org/10.5194/npg-27-373-2020, 2020
Short summary
Short summary
The performance of three machine-learning methods for data-driven modeling of a multiscale chaotic Lorenz 96 system is examined. One of the methods is found to be able to predict the future evolution of the chaotic system well from just knowing the past observations of the large-scale component of the multiscale state vector. Potential applications to data-driven and data-assisted surrogate modeling of complex dynamical systems such as weather and climate are discussed.
Eviatar Bach and Michael Ghil
Nonlin. Processes Geophys. Discuss., https://doi.org/10.5194/npg-2021-35, https://doi.org/10.5194/npg-2021-35, 2021
Preprint withdrawn
Short summary
Short summary
Data assimilation (DA) is the process of combining model forecasts with observations in order to provide an optimal estimate of the system state. When models are imperfect, the uncertainty in the forecasts may be underestimated, requiring inflation of the corresponding error covariance. Here, we present a simple method for estimating the magnitude and structure of the model error covariance matrix. We demonstrate the efficacy of this method with idealized experiments.
Benjamin A. Toms, Karthik Kashinath, Prabhat, and Da Yang
Geosci. Model Dev., 14, 4495–4508, https://doi.org/10.5194/gmd-14-4495-2021, https://doi.org/10.5194/gmd-14-4495-2021, 2021
Short summary
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.
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.
Ashesh Chattopadhyay, Pedram Hassanzadeh, and Devika Subramanian
Nonlin. Processes Geophys., 27, 373–389, https://doi.org/10.5194/npg-27-373-2020, https://doi.org/10.5194/npg-27-373-2020, 2020
Short summary
Short summary
The performance of three machine-learning methods for data-driven modeling of a multiscale chaotic Lorenz 96 system is examined. One of the methods is found to be able to predict the future evolution of the chaotic system well from just knowing the past observations of the large-scale component of the multiscale state vector. Potential applications to data-driven and data-assisted surrogate modeling of complex dynamical systems such as weather and climate are discussed.
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
Earth and space science informatics
The Common Community Physics Package (CCPP) Framework v6
Causal deep learning models for studying the Earth system
A methodological framework for improving the performance of data-driven models: a case study for daily runoff prediction in the Maumee domain, USA
SHAFTS (v2022.3): a deep-learning-based Python package for simultaneous extraction of building height and footprint from sentinel imagery
Bayesian atmospheric correction over land: Sentinel-2/MSI and Landsat 8/OLI
A generalized spatial autoregressive neural network (GSARNN) method for three-dimensional spatial interpolation
Twenty-five years of the IPCC Data Distribution Centre at the DKRZ and the Reference Data Archive for CMIP data
Effectiveness and computational efficiency of absorbing boundary conditions for full-waveform inversion
LAND-SUITE V1.0: a suite of tools for statistically based landslide susceptibility zonation
Fast infrared radiative transfer calculations using graphics processing units: JURASSIC-GPU v2.0
CSDMS: a community platform for numerical modeling of Earth surface processes
A new methodological framework for geophysical sensor combinations associated with machine learning algorithms to understand soil attributes
Model calibration using ESEm v1.1.0 – an open, scalable Earth system emulator
Turbidity maximum zone index: a novel model for remote extraction of the turbidity maximum zone in different estuaries
dh2loop 1.0: an open-source Python library for automated processing and classification of geological logs
Copula-based synthetic data augmentation for machine-learning emulators
Automated geological map deconstruction for 3D model construction using map2loop 1.0 and map2model 1.0
A spatially explicit approach to simulate urban heat mitigation with InVEST (v3.8.0)
S-SOM v1.0: a structural self-organizing map algorithm for weather typing
Using Shapley additive explanations to interpret extreme gradient boosting predictions of grassland degradation in Xilingol, China
Current status on the need for improved accessibility to climate models code
ClimateNet: an expert-labeled open dataset and deep learning architecture for enabling high-precision analyses of extreme weather
A spatiotemporal weighted regression model (STWR v1.0) for analyzing local nonstationarity in space and time
A new end-to-end workflow for the Community Earth System Model (version 2.0) for the Coupled Model Intercomparison Project Phase 6 (CMIP6)
HyLands 1.0: a hybrid landscape evolution model to simulate the impact of landslides and landslide-derived sediment on landscape evolution
Comparative analysis of atmospheric radiative transfer models using the Atmospheric Look-up table Generator (ALG) toolbox (version 2.0)
Fast domain-aware neural network emulation of a planetary boundary layer parameterization in a numerical weather forecast model
VISIR-1.b: ocean surface gravity waves and currents for energy-efficient navigation
Topological data analysis and machine learning for recognizing atmospheric river patterns in large climate datasets
Global hydro-climatic biomes identified via multitask learning
A run control framework to streamline profiling, porting, and tuning simulation runs and provenance tracking of geoscientific applications
An improved logistic regression model based on a spatially weighted technique (ILRBSWT v1.0) and its application to mineral prospectivity mapping
High-performance software framework for the calculation of satellite-to-satellite data matchups (MMS version 1.2)
A data model of the Climate and Forecast metadata conventions (CF-1.6) with a software implementation (cf-python v2.1)
Reverse engineering model structures for soil and ecosystem respiration: the potential of gene expression programming
A high-fidelity multiresolution digital elevation model for Earth systems
CPMIP: measurements of real computational performance of Earth system models in CMIP6
Automatic delineation of geomorphological slope units with r.slopeunits v1.0 and their optimization for landslide susceptibility modeling
Community Intercomparison Suite (CIS) v1.4.0: a tool for intercomparing models and observations
Asynchronous communication in spectral-element and discontinuous Galerkin methods for atmospheric dynamics – a case study using the High-Order Methods Modeling Environment (HOMME-homme_dg_branch)
GO2OGS 1.0: a versatile workflow to integrate complex geological information with fault data into numerical simulation models
An open and extensible framework for spatially explicit land use change modelling: the lulcc R package
Plant functional type classification for earth system models: results from the European Space Agency's Land Cover Climate Change Initiative
Non-singular spherical harmonic expressions of geomagnetic vector and gradient tensor fields in the local north-oriented reference frame
An approach to enhance pnetCDF performance in environmental modeling applications
A strategy for GIS-based 3-D slope stability modelling over large areas
An approach to computing direction relations between separated object groups
Improving computational efficiency in large linear inverse problems: an example from carbon dioxide flux estimation
Coupling technologies for Earth System Modelling
Quality assessment concept of the World Data Center for Climate and its application to CMIP5 data
Dominikus Heinzeller, Ligia Bernardet, Grant Firl, Man Zhang, Xia Sun, and Michael Ek
Geosci. Model Dev., 16, 2235–2259, https://doi.org/10.5194/gmd-16-2235-2023, https://doi.org/10.5194/gmd-16-2235-2023, 2023
Short summary
Short summary
The Common Community Physics Package is a collection of physical atmospheric parameterizations for use in Earth system models and a framework that couples the physics to a host model’s dynamical core. A primary goal for this effort is to facilitate research and development of physical parameterizations and physics–dynamics coupling methods while offering capabilities for numerical weather prediction operations, for example in the upcoming implementation of the Global Forecast System (GFS) v17.
Tobias Tesch, Stefan Kollet, and Jochen Garcke
Geosci. Model Dev., 16, 2149–2166, https://doi.org/10.5194/gmd-16-2149-2023, https://doi.org/10.5194/gmd-16-2149-2023, 2023
Short summary
Short summary
A recent statistical approach for studying relations in the Earth system is to train deep learning (DL) models to predict Earth system variables given one or several others and use interpretable DL to analyze the relations learned by the models. Here, we propose to combine the approach with a theorem from causality research to ensure that the deep learning model learns causal rather than spurious relations. As an example, we apply the method to study soil-moisture–precipitation coupling.
Yao Hu, Chirantan Ghosh, and Siamak Malakpour-Estalaki
Geosci. Model Dev., 16, 1925–1936, https://doi.org/10.5194/gmd-16-1925-2023, https://doi.org/10.5194/gmd-16-1925-2023, 2023
Short summary
Short summary
Data-driven models (DDMs) gain popularity in earth and environmental systems, thanks in large part to advancements in data collection techniques and artificial intelligence (AI). The performance of these models is determined by the underlying machine learning (ML) algorithms. In this study, we develop a framework to improve the model performance by optimizing ML algorithms and demonstrate the effectiveness of the framework using a DDM to predict edge-of-field runoff in the Maumee domain, USA.
Ruidong Li, Ting Sun, Fuqiang Tian, and Guang-Heng Ni
Geosci. Model Dev., 16, 751–778, https://doi.org/10.5194/gmd-16-751-2023, https://doi.org/10.5194/gmd-16-751-2023, 2023
Short summary
Short summary
We developed SHAFTS (Simultaneous building Height And FootprinT extraction from Sentinel imagery), a multi-task deep-learning-based Python package, to estimate average building height and footprint from Sentinel imagery. Evaluation in 46 cities worldwide shows that SHAFTS achieves significant improvement over existing machine-learning-based methods.
Feng Yin, Philip E. Lewis, and Jose L. Gómez-Dans
Geosci. Model Dev., 15, 7933–7976, https://doi.org/10.5194/gmd-15-7933-2022, https://doi.org/10.5194/gmd-15-7933-2022, 2022
Short summary
Short summary
The proposed SIAC atmospheric correction method provides consistent surface reflectance estimations from medium spatial-resolution satellites (Sentinel 2 and Landsat 8) with per-pixel uncertainty information. The outputs from SIAC have been validated against a wide range of ground measurements, and it shows that SIAC can provide accurate estimations of both surface reflectance and atmospheric parameters, with meaningful uncertainty information.
Junda Zhan, Sensen Wu, Jin Qi, Jindi Zeng, Mengjiao Qin, Yuanyuan Wang, and Zhenhong Du
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-198, https://doi.org/10.5194/gmd-2022-198, 2022
Revised manuscript accepted for GMD
Short summary
Short summary
We developed a generalized spatial autoregressive neural network used for three-dimensional spatial interpolation. Taking the different changing trend of geographic elements along various directions into consideration, the model defined spatial distance in a generalized way and integrated it into the process of spatial interpolation with the theories of spatial autoregression and neural network. Compared with traditional methods, the model achieved better performance and was more adaptable.
Martina Stockhause and Michael Lautenschlager
Geosci. Model Dev., 15, 6047–6058, https://doi.org/10.5194/gmd-15-6047-2022, https://doi.org/10.5194/gmd-15-6047-2022, 2022
Short summary
Short summary
The Data Distribution Centre (DDC) of the Intergovernmental Panel on Climate Change (IPCC) celebrates its 25th anniversary in 2022. DDC Partner DKRZ has supported the IPCC Assessments and preserved the quality-assured, citable climate model data underpinning the Assessment Reports over these years over the long term. With the introduction of the IPCC FAIR Guidelines into the current AR6, the value of DDC services has been recognized. However, DDC sustainability remains unresolved.
Daiane Iglesia Dolci, Felipe A. G. Silva, Pedro S. Peixoto, and Ernani V. Volpe
Geosci. Model Dev., 15, 5857–5881, https://doi.org/10.5194/gmd-15-5857-2022, https://doi.org/10.5194/gmd-15-5857-2022, 2022
Short summary
Short summary
We investigate and compare the theoretical and computational characteristics of several absorbing boundary conditions (ABCs) for the full-waveform inversion (FWI) problem. The different ABCs are implemented in an optimized computational framework called Devito. The computational efficiency and memory requirements of the ABC methods are evaluated in the forward and adjoint wave propagators, from simple to realistic velocity models.
Mauro Rossi, Txomin Bornaetxea, and Paola Reichenbach
Geosci. Model Dev., 15, 5651–5666, https://doi.org/10.5194/gmd-15-5651-2022, https://doi.org/10.5194/gmd-15-5651-2022, 2022
Short summary
Short summary
LAND-SUITE is a software package designed to support landslide susceptibility zonation. The software integrates, extends, and completes LAND-SE (Rossi et al., 2010; Rossi and Reichenbach, 2016). The software is implemented in R, a free software environment for statistical computing and graphics, and gives expert users the possibility to perform easier, more flexible, and more informed statistically based landslide susceptibility applications and zonations.
Paul F. Baumeister and Lars Hoffmann
Geosci. Model Dev., 15, 1855–1874, https://doi.org/10.5194/gmd-15-1855-2022, https://doi.org/10.5194/gmd-15-1855-2022, 2022
Short summary
Short summary
The efficiency of the numerical simulation of radiative transport is shown on modern server-class graphics cards (GPUs). The low-cost prefactor on GPUs compared to general-purpose processors (CPUs) enables future large retrieval campaigns for multi-channel data from infrared sounders aboard low-orbit satellites. The validated research software JURASSIC is available in the public domain.
Gregory E. Tucker, Eric W. H. Hutton, Mark D. Piper, Benjamin Campforts, Tian Gan, Katherine R. Barnhart, Albert J. Kettner, Irina Overeem, Scott D. Peckham, Lynn McCready, and Jaia Syvitski
Geosci. Model Dev., 15, 1413–1439, https://doi.org/10.5194/gmd-15-1413-2022, https://doi.org/10.5194/gmd-15-1413-2022, 2022
Short summary
Short summary
Scientists use computer simulation models to understand how Earth surface processes work, including floods, landslides, soil erosion, river channel migration, ocean sedimentation, and coastal change. Research benefits when the software for simulation modeling is open, shared, and coordinated. The Community Surface Dynamics Modeling System (CSDMS) is a US-based facility that supports research by providing community support, computing tools and guidelines, and educational resources.
Danilo César de Mello, Gustavo Vieira Veloso, Marcos Guedes de Lana, Fellipe Alcantara de Oliveira Mello, Raul Roberto Poppiel, Diego Ribeiro Oquendo Cabrero, Luis Augusto Di Loreto Di Raimo, Carlos Ernesto Gonçalves Reynaud Schaefer, Elpídio Inácio Fernandes Filho, Emilson Pereira Leite, and José Alexandre Melo Demattê
Geosci. Model Dev., 15, 1219–1246, https://doi.org/10.5194/gmd-15-1219-2022, https://doi.org/10.5194/gmd-15-1219-2022, 2022
Short summary
Short summary
We used soil parent material, terrain attributes, and geophysical data from the soil surface to test and compare different and unprecedented geophysical sensor combination, as well as different machine learning algorithms to model and predict several soil attributes. Also, we analyzed the importance of pedoenvironmental variables. The soil attributes were modeled throughout different machine learning algorithms and related to different geophysical sensor combinations.
Duncan Watson-Parris, Andrew Williams, Lucia Deaconu, and Philip Stier
Geosci. Model Dev., 14, 7659–7672, https://doi.org/10.5194/gmd-14-7659-2021, https://doi.org/10.5194/gmd-14-7659-2021, 2021
Short summary
Short summary
The Earth System Emulator (ESEm) provides a fast and flexible framework for emulating a wide variety of Earth science datasets and tools for constraining (or tuning) models of any complexity. Three distinct use cases are presented that demonstrate the utility of ESEm and provide some insight into the use of machine learning for emulation in these different settings. The open-source Python package is freely available so that it might become a valuable tool for the community.
Chongyang Wang, Li Wang, Danni Wang, Dan Li, Chenghu Zhou, Hao Jiang, Qiong Zheng, Shuisen Chen, Kai Jia, Yangxiaoyue Liu, Ji Yang, Xia Zhou, and Yong Li
Geosci. Model Dev., 14, 6833–6846, https://doi.org/10.5194/gmd-14-6833-2021, https://doi.org/10.5194/gmd-14-6833-2021, 2021
Short summary
Short summary
The turbidity maximum zone (TMZ) is a special phenomenon in estuaries worldwide. However, the extraction methods and criteria used to describe the TMZ vary significantly both spatially and temporally. This study proposes an new index, the turbidity maximum zone index, based on the corresponding relationship of total suspended solid concentration and Chl a concentration, which could better extract TMZs in different estuaries and on different dates.
Ranee Joshi, Kavitha Madaiah, Mark Jessell, Mark Lindsay, and Guillaume Pirot
Geosci. Model Dev., 14, 6711–6740, https://doi.org/10.5194/gmd-14-6711-2021, https://doi.org/10.5194/gmd-14-6711-2021, 2021
Short summary
Short summary
We have developed a software that allows the user to extract and standardize drill hole information from legacy datasets and/or different drilling campaigns. It also provides functionality to upscale the lithological information. These functionalities were possible by developing thesauri to identify and group geological terminologies together.
David Meyer, Thomas Nagler, and Robin J. Hogan
Geosci. Model Dev., 14, 5205–5215, https://doi.org/10.5194/gmd-14-5205-2021, https://doi.org/10.5194/gmd-14-5205-2021, 2021
Short summary
Short summary
A major limitation in training machine-learning emulators is often caused by the lack of data. This paper presents a cheap way to increase the size of training datasets using statistical techniques and thereby improve the performance of machine-learning emulators.
Mark Jessell, Vitaliy Ogarko, Yohan de Rose, Mark Lindsay, Ranee Joshi, Agnieszka Piechocka, Lachlan Grose, Miguel de la Varga, Laurent Ailleres, and Guillaume Pirot
Geosci. Model Dev., 14, 5063–5092, https://doi.org/10.5194/gmd-14-5063-2021, https://doi.org/10.5194/gmd-14-5063-2021, 2021
Short summary
Short summary
We have developed software that allows the user to extract sufficient information from unmodified digital maps and associated datasets that we are able to use to automatically build 3D geological models. By automating the process we are able to remove human bias from the procedure, which makes the workflow reproducible.
Martí Bosch, Maxence Locatelli, Perrine Hamel, Roy P. Remme, Jérôme Chenal, and Stéphane Joost
Geosci. Model Dev., 14, 3521–3537, https://doi.org/10.5194/gmd-14-3521-2021, https://doi.org/10.5194/gmd-14-3521-2021, 2021
Short summary
Short summary
The article presents a novel approach to simulate urban heat mitigation from land use/land cover data based on three biophysical mechanisms: tree shade, evapotranspiration and albedo. An automated procedure is proposed to calibrate the model parameters to best fit temperature observations from monitoring stations. A case study in Lausanne, Switzerland, shows that the approach outperforms regressions based on satellite data and provides valuable insights into design heat mitigation policies.
Quang-Van Doan, Hiroyuki Kusaka, Takuto Sato, and Fei Chen
Geosci. Model Dev., 14, 2097–2111, https://doi.org/10.5194/gmd-14-2097-2021, https://doi.org/10.5194/gmd-14-2097-2021, 2021
Short summary
Short summary
This study proposes a novel structural self-organizing map (S-SOM) algorithm. The superiority of S-SOM is that it can better recognize the difference (or similarity) among spatial (or temporal) data used for training and thus improve the clustering quality compared to traditional SOM algorithms.
Batunacun, Ralf Wieland, Tobia Lakes, and Claas Nendel
Geosci. Model Dev., 14, 1493–1510, https://doi.org/10.5194/gmd-14-1493-2021, https://doi.org/10.5194/gmd-14-1493-2021, 2021
Short summary
Short summary
Extreme gradient boosting (XGBoost) can provide alternative insights that conventional land-use models are unable to generate. Shapley additive explanations (SHAP) can interpret the results of the purely data-driven approach. XGBoost achieved similar and robust simulation results. SHAP values were useful for analysing the complex relationship between the different drivers of grassland degradation.
Juan A. Añel, Michael García-Rodríguez, and Javier Rodeiro
Geosci. Model Dev., 14, 923–934, https://doi.org/10.5194/gmd-14-923-2021, https://doi.org/10.5194/gmd-14-923-2021, 2021
Short summary
Short summary
This work shows that it continues to be hard, if not impossible, to obtain some of the most used climate models worldwide. We reach this conclusion through a systematic study and encourage all development teams and research centres to make public the models they use to produce scientific results.
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.
Xiang Que, Xiaogang Ma, Chao Ma, and Qiyu Chen
Geosci. Model Dev., 13, 6149–6164, https://doi.org/10.5194/gmd-13-6149-2020, https://doi.org/10.5194/gmd-13-6149-2020, 2020
Short summary
Short summary
This paper presents a spatiotemporal weighted regression (STWR) model for exploring nonstationary spatiotemporal processes in nature and socioeconomics. A value change rate is introduced in the temporal kernel, which presents significant model fitting and accuracy in both simulated and real-world data. STWR fully incorporates observed data in the past and outperforms geographic temporal weighted regression (GTWR) and geographic weighted regression (GWR) models in several experiments.
Sheri Mickelson, Alice Bertini, Gary Strand, Kevin Paul, Eric Nienhouse, John Dennis, and Mariana Vertenstein
Geosci. Model Dev., 13, 5567–5581, https://doi.org/10.5194/gmd-13-5567-2020, https://doi.org/10.5194/gmd-13-5567-2020, 2020
Short summary
Short summary
Every generation of MIP exercises introduces new layers of complexity and an exponential growth in the amount of data requested. CMIP6 required us to develop a new tool chain and forced us to change our methodologies. The new methods discussed in this paper provided us with an 18 times faster speedup over our existing methods. This allowed us to meet our deadlines and we were able to publish more than half a million data sets on the Earth System Grid Federation (ESGF) for the CMIP6 project.
Benjamin Campforts, Charles M. Shobe, Philippe Steer, Matthias Vanmaercke, Dimitri Lague, and Jean Braun
Geosci. Model Dev., 13, 3863–3886, https://doi.org/10.5194/gmd-13-3863-2020, https://doi.org/10.5194/gmd-13-3863-2020, 2020
Short summary
Short summary
Landslides shape the Earth’s surface and are a dominant source of terrestrial sediment. Rivers, then, act as conveyor belts evacuating landslide-produced sediment. Understanding the interaction among rivers and landslides is important to predict the Earth’s surface response to past and future environmental changes and for mitigating natural hazards. We develop HyLands, a new numerical model that provides a toolbox to explore how landslides and rivers interact over several timescales.
Jorge Vicent, Jochem Verrelst, Neus Sabater, Luis Alonso, Juan Pablo Rivera-Caicedo, Luca Martino, Jordi Muñoz-Marí, and José Moreno
Geosci. Model Dev., 13, 1945–1957, https://doi.org/10.5194/gmd-13-1945-2020, https://doi.org/10.5194/gmd-13-1945-2020, 2020
Short summary
Short summary
The modeling of light propagation through the atmosphere is key to process satellite images and to understand atmospheric processes. However, existing atmospheric models can be complex to use in practical applications. Here we aim at providing a new software tool to facilitate using advanced models and to generate large databases of simulated data. As a test case, we use this tool to analyze differences between several atmospheric models, showing the capabilities of this open-source tool.
Jiali Wang, Prasanna Balaprakash, and Rao Kotamarthi
Geosci. Model Dev., 12, 4261–4274, https://doi.org/10.5194/gmd-12-4261-2019, https://doi.org/10.5194/gmd-12-4261-2019, 2019
Short summary
Short summary
Parameterizations are frequently used in models representing physical phenomena and are often the computationally expensive portions of the code. Using model output from simulations performed using a weather model, we train deep neural networks to provide an accurate alternative to a physics-based parameterization. We demonstrate that a domain-aware deep neural network can successfully simulate the entire diurnal cycle of the boundary layer physics and the results are transferable.
Gianandrea Mannarini and Lorenzo Carelli
Geosci. Model Dev., 12, 3449–3480, https://doi.org/10.5194/gmd-12-3449-2019, https://doi.org/10.5194/gmd-12-3449-2019, 2019
Short summary
Short summary
The VISIR ship-routing model is updated in order to deal with ocean currents.
The optimal tracks we computed through VISIR in the Atlantic ocean show great seasonal and regional variability, following a variable influence of surface gravity waves and currents. We assess how these tracks contribute to voyage energy-efficiency gains through a standard indicator (EEOI) of the International Maritime Organization. Also, the new model features are validated against an exact analytical benchmark.
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.
Christina Papagiannopoulou, Diego G. Miralles, Matthias Demuzere, Niko E. C. Verhoest, and Willem Waegeman
Geosci. Model Dev., 11, 4139–4153, https://doi.org/10.5194/gmd-11-4139-2018, https://doi.org/10.5194/gmd-11-4139-2018, 2018
Short summary
Short summary
Common global land cover and climate classifications are based on vegetation–climatic characteristics derived from observational data, ignoring the interaction between the local climate and biome. Here, we model the interplay between vegetation and local climate by discovering spatial relationships among different locations. The resulting global
hydro-climatic biomescorrespond to regions of coherent climate–vegetation interactions that agree well with traditional global land cover maps.
Wendy Sharples, Ilya Zhukov, Markus Geimer, Klaus Goergen, Sebastian Luehrs, Thomas Breuer, Bibi Naz, Ketan Kulkarni, Slavko Brdar, and Stefan Kollet
Geosci. Model Dev., 11, 2875–2895, https://doi.org/10.5194/gmd-11-2875-2018, https://doi.org/10.5194/gmd-11-2875-2018, 2018
Short summary
Short summary
Next-generation geoscientific models are based on complex model implementations and workflows. Next-generation HPC systems require new programming paradigms and code optimization. In order to meet the challenge of running complex simulations on new massively parallel HPC systems, we developed a run control framework that facilitates code portability, code profiling, and provenance tracking to reduce both the duration and the cost of code migration and development, while ensuring reproducibility.
Daojun Zhang, Na Ren, and Xianhui Hou
Geosci. Model Dev., 11, 2525–2539, https://doi.org/10.5194/gmd-11-2525-2018, https://doi.org/10.5194/gmd-11-2525-2018, 2018
Short summary
Short summary
Geographically weighted regression is a widely used method to deal with spatial heterogeneity, which is common in geostatistics. However, most existing software does not support logistic regression and cannot deal with missing data, which exist extensively in mineral prospectivity mapping. This work generalized logistic regression to spatial statistics based on a spatially weighted technique. The new model also supports an anisotropic local window, which is another innovative point.
Thomas Block, Sabine Embacher, Christopher J. Merchant, and Craig Donlon
Geosci. Model Dev., 11, 2419–2427, https://doi.org/10.5194/gmd-11-2419-2018, https://doi.org/10.5194/gmd-11-2419-2018, 2018
Short summary
Short summary
For calibration and validation purposes it is necessary to detect simultaneous data acquisitions from different spaceborne platforms. We present an algorithm and a software system which implements a general approach to resolve this problem. The multisensor matchup system (MMS) can detect simultaneous acquisitions in a large dataset (> 100 TB) and extract data for matching locations for further analysis. The MMS implements a flexible software infrastructure and allows for high parallelization.
David Hassell, Jonathan Gregory, Jon Blower, Bryan N. Lawrence, and Karl E. Taylor
Geosci. Model Dev., 10, 4619–4646, https://doi.org/10.5194/gmd-10-4619-2017, https://doi.org/10.5194/gmd-10-4619-2017, 2017
Short summary
Short summary
We present a formal data model for version 1.6 of the CF (Climate and Forecast) metadata conventions that provide a description of the physical meaning of geoscientific data and their spatial and temporal properties. We describe the CF conventions and how they lead to our CF data model, and compare it other data models for storing data and metadata. We present cf-python version 2.1: a software implementation of the CF data model capable of manipulating any CF-compliant dataset.
Iulia Ilie, Peter Dittrich, Nuno Carvalhais, Martin Jung, Andreas Heinemeyer, Mirco Migliavacca, James I. L. Morison, Sebastian Sippel, Jens-Arne Subke, Matthew Wilkinson, and Miguel D. Mahecha
Geosci. Model Dev., 10, 3519–3545, https://doi.org/10.5194/gmd-10-3519-2017, https://doi.org/10.5194/gmd-10-3519-2017, 2017
Short summary
Short summary
Accurate representation of land-atmosphere carbon fluxes is essential for future climate projections, although some of the responses of CO2 fluxes to climate often remain uncertain. The increase in available data allows for new approaches in their modelling. We automatically developed models for ecosystem and soil carbon respiration using a machine learning approach. When compared with established respiration models, we found that they are better in prediction as well as offering new insights.
Xinqiao Duan, Lin Li, Haihong Zhu, and Shen Ying
Geosci. Model Dev., 10, 239–253, https://doi.org/10.5194/gmd-10-239-2017, https://doi.org/10.5194/gmd-10-239-2017, 2017
Short summary
Short summary
This article proposes an optimized transformation for topographic datasets. The resulting topographic grid exhibits good surface approximation and quasi-uniform high-quality. Both features of the processed topography build a concrete base from which improved endogenous or exogenous parameters can be derived, and makes it suitable for Earth and environmental simulations.
Venkatramani Balaji, Eric Maisonnave, Niki Zadeh, Bryan N. Lawrence, Joachim Biercamp, Uwe Fladrich, Giovanni Aloisio, Rusty Benson, Arnaud Caubel, Jeffrey Durachta, Marie-Alice Foujols, Grenville Lister, Silvia Mocavero, Seth Underwood, and Garrett Wright
Geosci. Model Dev., 10, 19–34, https://doi.org/10.5194/gmd-10-19-2017, https://doi.org/10.5194/gmd-10-19-2017, 2017
Short summary
Short summary
Climate models are among the most computationally expensive scientific applications in the world. We present a set of measures of computational performance that can be used to compare models that are independent of underlying hardware and the model formulation. They are easy to collect and reflect performance actually achieved in practice. We are preparing a systematic effort to collect these metrics for the world's climate models during CMIP6, the next Climate Model Intercomparison Project.
Massimiliano Alvioli, Ivan Marchesini, Paola Reichenbach, Mauro Rossi, Francesca Ardizzone, Federica Fiorucci, and Fausto Guzzetti
Geosci. Model Dev., 9, 3975–3991, https://doi.org/10.5194/gmd-9-3975-2016, https://doi.org/10.5194/gmd-9-3975-2016, 2016
Short summary
Short summary
Slope units are morphological mapping units bounded by drainage and divide lines that maximize within-unit homogeneity and between-unit heterogeneity. We use r.slopeunits, a software for the automatic delination of slope units. We outline an objective procedure to optimize the software input parameters for landslide susceptibility (LS) zonation. Optimization is achieved by maximizing an objective function that simultaneously evaluates terrain aspect segmentation quality and LS model performance.
Duncan Watson-Parris, Nick Schutgens, Nicholas Cook, Zak Kipling, Philip Kershaw, Edward Gryspeerdt, Bryan Lawrence, and Philip Stier
Geosci. Model Dev., 9, 3093–3110, https://doi.org/10.5194/gmd-9-3093-2016, https://doi.org/10.5194/gmd-9-3093-2016, 2016
Short summary
Short summary
In this paper we describe CIS, a new command line tool for the easy visualization, analysis and comparison of a wide variety of gridded and ungridded data sets used in Earth sciences. Users can now use a single tool to not only view plots of satellite, aircraft, station or model data, but also bring them onto the same spatio-temporal sampling. This allows robust, quantitative comparisons to be made easily. CIS is an open-source project and welcomes input from the community.
Benjamin F. Jamroz and Robert Klöfkorn
Geosci. Model Dev., 9, 2881–2892, https://doi.org/10.5194/gmd-9-2881-2016, https://doi.org/10.5194/gmd-9-2881-2016, 2016
Short summary
Short summary
The scalability of computational applications on current and next-generation supercomputers is increasingly limited by the cost of inter-process communication. We implement communication hiding data exchange in the High-Order Methods Modeling Environment (HOMME) for the time integration of the hydrostatic fluid equations using both the spectral-element and discontinuous Galerkin methods. The presented approach produces significant performance and scalability gains in large-scale simulations.
T. Fischer, D. Naumov, S. Sattler, O. Kolditz, and M. Walther
Geosci. Model Dev., 8, 3681–3694, https://doi.org/10.5194/gmd-8-3681-2015, https://doi.org/10.5194/gmd-8-3681-2015, 2015
Short summary
Short summary
We present a workflow to convert geological models into the open-source VTU format for usage in numerical simulation models. Tackling relevant scientific questions or engineering tasks often involves multidisciplinary approaches. Conversion workflows are needed between the diverse tools of the various disciplines. Our approach offers an open-source, platform-independent, robust, and comprehensible method that is potentially useful for a multitude of similar environmental studies.
S. Moulds, W. Buytaert, and A. Mijic
Geosci. Model Dev., 8, 3215–3229, https://doi.org/10.5194/gmd-8-3215-2015, https://doi.org/10.5194/gmd-8-3215-2015, 2015
Short summary
Short summary
The contribution of lulcc is to provide a free and open-source framework for land use change modelling. The software, which is provided as an R package, addresses problems associated with the current paradigm of closed-source, specialised land use change modelling software which disrupt the scientific process. It is an attempt to move the discipline towards open and transparent science and to ensure land use change models are accessible to scientists working across the geosciences.
B. Poulter, N. MacBean, A. Hartley, I. Khlystova, O. Arino, R. Betts, S. Bontemps, M. Boettcher, C. Brockmann, P. Defourny, S. Hagemann, M. Herold, G. Kirches, C. Lamarche, D. Lederer, C. Ottlé, M. Peters, and P. Peylin
Geosci. Model Dev., 8, 2315–2328, https://doi.org/10.5194/gmd-8-2315-2015, https://doi.org/10.5194/gmd-8-2315-2015, 2015
Short summary
Short summary
Land cover is an essential variable in earth system models and determines conditions driving biogeochemical, energy and water exchange between ecosystems and the atmosphere. A methodology is presented for mapping plant functional types used in global vegetation models from a updated land cover classification system and open-source conversion tool, resulting from a consultative process among map producers and modelers engaged in the European Space Agency’s Land Cover Climate Change Initiative.
J. Du, C. Chen, V. Lesur, and L. Wang
Geosci. Model Dev., 8, 1979–1990, https://doi.org/10.5194/gmd-8-1979-2015, https://doi.org/10.5194/gmd-8-1979-2015, 2015
D. C. Wong, C. E. Yang, J. S. Fu, K. Wong, and Y. Gao
Geosci. Model Dev., 8, 1033–1046, https://doi.org/10.5194/gmd-8-1033-2015, https://doi.org/10.5194/gmd-8-1033-2015, 2015
M. Mergili, I. Marchesini, M. Alvioli, M. Metz, B. Schneider-Muntau, M. Rossi, and F. Guzzetti
Geosci. Model Dev., 7, 2969–2982, https://doi.org/10.5194/gmd-7-2969-2014, https://doi.org/10.5194/gmd-7-2969-2014, 2014
Short summary
Short summary
The article deals with strategies to (i) reduce computation time and to (ii) appropriately account for uncertain input parameters when applying an open source GIS sliding surface model to estimate landslide susceptibility for a 90km² study area in central Italy. For (i), the area is split into a large number of tiles, enabling the exploitation of multi-processor computing environments. For (ii), the model is run with various parameter combinations to compute the slope failure probability.
H. Yan, Z. Wang, and J. Li
Geosci. Model Dev., 6, 1591–1599, https://doi.org/10.5194/gmd-6-1591-2013, https://doi.org/10.5194/gmd-6-1591-2013, 2013
V. Yadav and A. M. Michalak
Geosci. Model Dev., 6, 583–590, https://doi.org/10.5194/gmd-6-583-2013, https://doi.org/10.5194/gmd-6-583-2013, 2013
S. Valcke, V. Balaji, A. Craig, C. DeLuca, R. Dunlap, R. W. Ford, R. Jacob, J. Larson, R. O'Kuinghttons, G. D. Riley, and M. Vertenstein
Geosci. Model Dev., 5, 1589–1596, https://doi.org/10.5194/gmd-5-1589-2012, https://doi.org/10.5194/gmd-5-1589-2012, 2012
M. Stockhause, H. Höck, F. Toussaint, and M. Lautenschlager
Geosci. Model Dev., 5, 1023–1032, https://doi.org/10.5194/gmd-5-1023-2012, https://doi.org/10.5194/gmd-5-1023-2012, 2012
Cited articles
Abarbanel, H. D., Rozdeba, P. J., and Shirman, S.: Machine learning: Deepest learning as statistical data assimilation problems, Neural Comput., 30, 2025–2055, 2018. a
Arcomano, T., Szunyogh, I., Pathak, J., Wikner, A., Hunt, B. R., and Ott, E.: A Machine Learning-Based Global Atmospheric Forecast Model, Geophys. Res. Lett., 47, e2020GL087776, https://doi.org/10.1029/2020GL087776, 2020. a, b, c
Bach, E., Mote, S., Krishnamurthy, V., Sharma, A. S., Ghil, M., and Kalnay, E.: Ensemble Oscillation Correction (EnOC): Leveraging oscillatory modes to improve forecasts of chaotic systems, J. Climate, 34, 5673–5686, 2021. a
Balaji, V.: Climbing down Charney's ladder: machine learning and the post-Dennard era of computational climate science, Philos. T. Roy. Soc. A, 379, 20200085, https://doi.org/10.1098/rsta.2020.0085, 2021. a
Bannister, R.: A review of operational methods of variational and ensemble-variational data assimilation, Q. J. Roy. Meteor. Soc., 143, 607–633, 2017. a
Beucler, T., Rasp, S., Pritchard,
M., and Gentine, P.: Achieving conservation of energy in neural
network emulators for climate modeling, arXiv [preprint],
arXiv:1906.06622,
2019. a
Beucler, T., Pritchard, M., Rasp, S., Ott, J., Baldi, P., and Gentine, P.: Enforcing analytic constraints in neural networks emulating physical systems, Phys. Rev. Lett., 126, 098302, https://doi.org/10.1103/PhysRevLett.126.098302, 2021. a
Bihlo, A. and Popovych, R. O.: Physics-informed neural networks for the shallow-water equations on the sphere, arXiv [preprint], arXiv:2104.00615, 2021. a
Brajard, J., Carrassi, A., Bocquet, M., and Bertino, L.: Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: a case study with the Lorenz 96 model, J. Comput. Sci., 44, 101171, https://doi.org/10.1016/j.jocs.2020.101171, 2020. a, b
Brajard, J., Carrassi, A., Bocquet, M., and Bertino, L.: Combining data assimilation and machine learning to infer unresolved scale parametrization, Philos. T. Roy. Soc. A, 379, 20200086, https://doi.org/10.1098/rsta.2020.0086, 2021. a, b
Bronstein, M. M., Bruna, J., Cohen, T., and Veličković, P.: Geometric deep learning: Grids, groups, graphs, geodesics, and gauges, arXiv [preprint], arXiv:2104.13478, 2021. a, b
Carrassi, A., Bocquet, M., Bertino, L., and Evensen, G.: Data assimilation in the geosciences: An overview of methods, issues, and perspectives, WIRes Clim. Change, 9, e535, https://doi.org/10.1002/wcc.535, 2018. a
Chantry, M., Christensen, H., Dueben, P., and Palmer, T.: Opportunities and challenges for machine learning in weather and climate modelling: hard, medium and soft AI, Philos. T. Roy. Soc. A, 379, 20200083, https://doi.org/10.1098/rsta.2020.0083, 2021. a
Chattopadhyay, A.: Towards physically consistent data-driven weather forecasting: Integrating data assimilation with equivariance-preserving deep spatial transformers, Zenodo [code], https://doi.org/10.5281/zenodo.6112374, 2021. a
Chattopadhyay, A., Hassanzadeh, P., and Pasha, S.: Predicting clustered weather patterns: A test case for applications of convolutional neural networks to spatio-temporal climate data, Sci. Rep., 10, 1–13, 2020a. a
Chattopadhyay, A., Hassanzadeh, P., and Subramanian, D.: Data-driven predictions of a multiscale Lorenz 96 chaotic system using machine-learning methods: reservoir computing, artificial neural network, and long short-term memory network, Nonlin. Processes Geophys., 27, 373–389, https://doi.org/10.5194/npg-27-373-2020, 2020b. a, b
Chattopadhyay, A., Mustafa, M., Hassanzadeh, P., and Kashinath, K.: Deep spatial transformers for autoregressive data-driven forecasting of geophysical turbulence, in: Proceedings of the 10th International Conference on Climate Informatics, Oxford, UK, 106–112, https://doi.org/10.1145/3429309.3429325, 2020c. a, b
Chattopadhyay, A., Nabizadeh, E., and Hassanzadeh, P.: Analog forecasting of extreme-causing weather patterns using deep learning, J. Adv. Model. Earth Sy., 12, e2019MS001958, https://doi.org/10.1029/2019MS001958, 2020d. a, b, c, d
Chattopadhyay, A., Subel, A., and Hassanzadeh, P.: Data-driven super-parameterization using deep learning: Experimentation with multi-scale Lorenz 96 systems and transfer-learning, J. Adv. Model. Earth Sy., 12, e2020MS002084, https://doi.org/10.1029/2020MS002084, 2020e. a, b
Daw, A., Thomas, R. Q., Carey, C. C., Read, J. S., Appling, A. P., and Karpatne, A.: Physics-guided architecture (pga) of neural networks for quantifying uncertainty in lake temperature modeling, in: Proceedings of the 2020 Siam International Conference on Data Mining, SIAM, Cincinnati, Ohio, 532–540, https://doi.org/10.1137/1.9781611976236.60, 2020. a
de Haan, P., Weiler, M., Cohen, T., and Welling, M.: Gauge equivariant mesh CNNs: Anisotropic convolutions on geometric graphs, arXiv [preprint], arXiv:2003.05425, 2020. a
Dueben, P. D. and Bauer, P.: Challenges and design choices for global weather and climate models based on machine learning, Geosci. Model Dev., 11, 3999–4009, https://doi.org/10.5194/gmd-11-3999-2018, 2018. a, b
Edwards, C. A., Moore, A. M., Hoteit, I., and Cornuelle, B. D.: Regional ocean data assimilation, Annu. Rev. Mar. Sci., 7, 21–42, 2015. a
Geer, A.: Learning earth system models from observations: machine learning or data assimilation?, Philos. T. Roy. Soc. A, 379, 20200089, https://doi.org/10.1098/rsta.2020.0089, 2021. a
Grönquist, P., Yao, C., Ben-Nun, T., Dryden, N., Dueben, P., Li, S., and Hoefler, T.: Deep learning for post-processing ensemble weather forecasts, Philos. T. Roy. Soc. A, 379, 20200092, https://doi.org/10.1098/rsta.2020.0092, 2021. a
Grooms, I.: Analog ensemble data assimilation and a method for constructing analogs with variational autoencoders, Q. J. Roy. Meteor. Soc., 147, 139–149, 2021. a
Guan, Y., Chattopadhyay, A., Subel, A., and Hassanzadeh, P.: Stable a posteriori LES of 2D turbulence using convolutional neural networks: Backscattering analysis and generalization to higher Re via transfer learning,
J. Computat. Phys.,
458,
111090,
https://doi.org/10.1016/j.jcp.2022.111090,
2022. a
Hamill, T. M., Whitaker, J. S., Anderson, J. L., and Snyder, C.: Comments on “Sigma-point Kalman filter data assimilation methods for strongly nonlinear systems”, J. Atmos. Sci., 66, 3498–3500, 2009. a
Hamilton, F., Berry, T., and Sauer, T.: Ensemble Kalman Filtering without a Model, Phys. Rev. X, 6, 011021, https://doi.org/10.1103/PhysRevX.6.011021, 2016. a
Hanc, J., Tuleja, S., and Hancova, M.: Symmetries and conservation laws: Consequences of Noether's theorem, Am. J. Phys., 72, 428–435, 2004. a
Hatfield, S. E., Chantry, M., Dueben, P. D., Lopez, P., Geer, A. J., and Palmer, T. N.: Building tangent-linear and adjoint models for data assimilation with neural networks, Earth and Space Science Open Archive ESSOAr, https://doi.org/10.1002/essoar.10506310.1, 2021. a
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz-Sabater, J., Nicolas, J., Peubey, C., Radu, R., Schepers, D., Simmons, A., Soci, C., Abdalla, S., Abellan, X., Balsamo, G., Bechtold, P., Biavati, G., Bidlot, J., Bonavita, M., Chiara, G. D., Dahlgren, P., Dee, D., Diamantakis, M., Dragani, R., Flemming, J., Forbes, R., Fuentes, M., Geer, A., Haimberger, L., Healy, S., Hogan, R. J., Hólm, E., Janisková, M., Keeley, S., Laloyaux, P., Lopez, P., Lupu, C., Radnoti, G., de Rosnay, P., Rozum, I., Vamborg, F., Villaume, S., and Thépaut, J.-N.: The ERA5 global reanalysis, Q. J. Roy. Meteor. Soc., 146, 1999–2049, 2020. a
Houtekamer, P. L. and Zhang, F.: Review of the Ensemble Kalman Filter for Atmospheric Data Assimilation, Mon. Weather Rev., 144, 4489–4532, 2016. a
Irrgang, C., Boers, N., Sonnewald, M., Barnes, E. A., Kadow, C., Staneva, J., and Saynisch-Wagner, J.: Towards neural Earth system modelling by integrating artificial intelligence in Earth system science, Nature Machine Intelligence, 3, 667–674, 2021. a
Julier, S. J. and Uhlmann, J. K.: Unscented filtering and nonlinear estimation, P. IEEE, 92, 401–422, 2004. a
Kashinath, K., Mustafa, M., Albert, A., Wu, J., Jiang, C., Esmaeilzadeh, S., Azizzadenesheli, K., Wang, R., Chattopadhyay, A., Singh, A., Manepalli, A., Chirila, D., Yu, R., Walters, R., White, B., Xiao, H., Tchelepi, H. A., Marcus, P., Anandkumar, A., Hassanzadeh, P., and Prabhat: Physics-informed machine learning: case studies for weather and climate modelling, Philos. T. Roy. Soc. A, 379, 20200093, https://doi.org/10.1098/rsta.2020.0093, 2021. a
Kovachki, N. B. and Stuart, A. M.: Ensemble Kalman inversion: a derivative-free technique for machine learning tasks, Inverse Probl., 35, 095005, https://doi.org/10.1088/1361-6420/ab1c3a, 2019. a
Kumar, S., Peters-Lidard, C., Tian, Y., Reichle, R., Geiger, J., Alonge, C., Eylander, J., and Houser, P.: An integrated hydrologic modeling and data assimilation framework, Computer, 41, 52–59, 2008a. a
Kumar, S. V., Reichle, R. H., Peters-Lidard, C. D., Koster, R. D., Zhan, X., Crow, W. T., Eylander, J. B., and Houser, P. R.: A land surface data assimilation framework using the land information system: Description and applications, Adv. Water Resour., 31, 1419–1432, 2008b. a
Leutbecher, M.: Ensemble size: How suboptimal is less than infinity?, Q. J. Roy. Meteor. Soc., 145, 107–128, 2019. a
Lguensat, R., Tandeo, P., Ailliot, P., Pulido, M., and Fablet, R.: The analog data assimilation, Mon. Weather Rev., 145, 4093–4107, 2017. a
Lguensat, R., Viet, P. H., Sun, M., Chen, G., Fenglin, T., Chapron, B., and Fablet, R.: Data-driven interpolation of sea level anomalies using analog data assimilation, Remote Sens., 11, 858, https://doi.org/10.3390/rs11070858, 2019. a
Liang, X., Losch, M., Nerger, L., Mu, L., Yang, Q., and Liu, C.: Using sea surface temperature observations to constrain upper ocean properties in an Arctic sea ice-ocean data assimilation system, J. Geophys. Res.-Oceans, 124, 4727–4743, 2019. a
Liu, Y., Kutz, J. N., and Brunton, S. L.: Hierarchical Deep Learning of Multiscale Differential Equation Time-Steppers, arXiv [preprint], arXiv:2008.09768, 2020. a
Lütkepohl, H.: Vector autoregressive models, in: Handbook of research methods and applications in empirical macroeconomics, Edward Elgar Publishing, ISBN 978 1 78254 507 1, 2013. a
Lynch, E. M.: Data Driven Prediction Without a Model, Doctoral thesis, University of Maryland, College Park, https://doi.org/10.13016/quty-dayf, 2019. a
Maron, H., Ben-Hamu, H., Shamir, N., and Lipman, Y.: Invariant and equivariant graph networks, arXiv [preprint], arXiv:1812.09902, 2018. a
Maron, H., Fetaya, E., Segol, N., and Lipman, Y.: On the universality of invariant networks, in: International Conference on Machine Learning, Long beach, California, PMLR, 97, 4363–4371, 2019. a
Maulik, R., Egele, R., Lusch, B., and Balaprakash, P.: Recurrent neural network architecture search for geophysical emulation, in: SC20: International Conference for High Performance Computing, Networking, Storage and Analysis, Atlanta, Georgia, IEEE, 1–14, ISBN 978-1-7281-9998-6, 2020. a
Maulik, R., Lusch, B., and Balaprakash, P.: Reduced-order modeling of advection-dominated systems with recurrent neural networks and convolutional autoencoders, Phys. Fluids, 33, 037106, https://doi.org/10.1063/5.0039986, 2021. a
Mohan, A. T., Lubbers, N., Livescu, D., and Chertkov, M.: Embedding hard physical constraints in neural network coarse-graining of 3D turbulence, arXiv [preprint], arXiv:2002.00021, 2020. a
Nadiga, B.: Reservoir Computing as a Tool for Climate Predictability Studies, J. Adv. Model. Earth Sy., e2020MS002290, https://doi.org/10.1029/2020MS002290, 2020. a
Pathak, J., Hunt, B., Girvan, M., Lu, Z., and Ott, E.: Model-free prediction of large spatiotemporally chaotic systems from data: A reservoir computing approach, Phys. Rev. Lett., 120, 024102, https://doi.org/10.1103/PhysRevLett.120.024102, 2018. a
Pawar, S. and San, O.: Data assimilation empowered neural network parameterizations for subgrid processes in geophysical flows, arXiv [preprint], arXiv:2006.08901, 2020. a
Pawar, S., Ahmed, S. E., San, O., Rasheed, A., and Navon, I. M.: Long short-term memory embedded nudging schemes for nonlinear data assimilation of geophysical flows, Phys. Fluids, 32, 076606, https://doi.org/10.1063/5.0012853, 2020. a
Penny, S., Bach, E., Bhargava, K., Chang, C.-C., Da, C., Sun, L., and Yoshida, T.: Strongly coupled data assimilation in multiscale media: Experiments using a quasi-geostrophic coupled model, J. Adv. Model. Earth Sy., 11, 1803–1829, 2019. a
Raissi, M., Yazdani, A., and Karniadakis, G. E.: Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations, Science, 367, 1026–1030, 2020. a
Reichstein, M., Camps-Valls, G., Stevens, B., Jung, M., Denzler, J.,Carvalhais, N., and Prabhat: Deep learning and process understanding for data-driven Earth system science, Nature, 566, 195–204, https://doi.org/10.1038/s41586-019-0912-1, 2019. a, b
Ronneberger, O., Fischer, P., and Brox, T.: U-net: Convolutional networks for biomedical image segmentation, in: International Conference on Medical image computing and computer-assisted intervention, Munich, Germany, Springer, 234–241, 2015. a
Sabour, S., Frosst, N., and Hinton, G. E.: Dynamic routing between capsules, arXiv [preprint], arXiv:1710.09829, 2017. a
Scher, S.: Toward data-driven weather and climate forecasting: Approximating a simple general circulation model with deep learning, Geophys. Res. Lett., 45, 12–616, 2018. a
Scher, S. and Messori, G.: Predicting weather forecast uncertainty with machine learning, Q. J. Roy. Meteor. Soc., 144, 2830–2841, 2018. a
Scher, S. and Messori, G.: Weather and climate forecasting with neural networks: using general circulation models (GCMs) with different complexity as a study ground, Geosci. Model Dev., 12, 2797–2809, https://doi.org/10.5194/gmd-12-2797-2019, 2019. a
Scher, S. and Messori, G.: Ensemble methods for neural network-based weather forecasts, J. Adv. Model. Earth Sy., e2020MS002331, https://doi.org/10.1029/2020MS002331, 2021. a
Schubert, S., Neubert, P., Pöschmann, J., and Pretzel, P.: Circular convolutional neural networks for panoramic images and laser data, in: 2019 IEEE Intelligent Vehicles Symposium (IV), Paris, France, IEEE, 653–660, 2019. a
Schultz, M., Betancourt, C., Gong, B., Kleinert, F., Langguth, M., Leufen, L., Mozaffari, A., and Stadtler, S.: Can deep learning beat numerical weather prediction?, Philos. T. Roy. Soc. A, 379, 20200097, https://doi.org/10.1098/rsta.2020.0097, 2021. a, b, c, d
Subel, A., Chattopadhyay, A., Guan, Y., and Hassanzadeh, P.: Data-driven subgrid-scale modeling of forced Burgers turbulence using deep learning with generalization to higher Reynolds numbers via transfer learning, Phys. Fluids, 33, 031702, https://doi.org/10.1063/5.0040286, 2021. a
Tang, M., Liu, Y., and Durlofsky, L. J.: A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems, J. Comput. Phys., 109456, https://doi.org/10.1016/j.jcp.2020.109456, 2020. a
Thiagarajan, J. J., Venkatesh, B., Anirudh, R., Bremer, P.-T., Gaffney, J., Anderson, G., and Spears, B.: Designing accurate emulators for scientific processes using calibration-driven deep models, Nat. Commun., 11, 1–10, 2020. a
Vlachas, P. R., Byeon, W., Wan, Z. Y., Sapsis, T. P., and Koumoutsakos, P.: Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks, P. Roy. Soc. A-Math. Phy., 474, 20170844, https://doi.org/10.1098/rspa.2017.0844, 2018. a
Wan, E. A., Van Der Merwe, R., and Haykin, S.: The unscented Kalman filter, Kalman filtering and neural networks, Proceedings of the IEEE 2000 Adaptive Systems for Signal Processing, Communications, and Control Symposium, Lake Louise, AB, Canada, 5, 221–280, https://doi.org/10.1109/ASSPCC.2000, 2001. a, b
Wang, R., Walters, R., and Yu, R.: Incorporating Symmetry into Deep Dynamics Models for Improved Generalization, arXiv [preprint], arXiv:2002.03061, 2020. a, b, c
Watson-Parris, D.: Machine learning for weather and climate are worlds apart, Philos. T. Roy. Soc. A, 379, 20200098, https://doi.org/10.1098/rsta.2020.0098, 2021. a
Weyn, J. A., Durran, D. R., Caruana, R., and Cresswell-Clay, N.: Sub-seasonal forecasting with a large ensemble of deep-learning weather prediction models, arXiv [preprint], arXiv:2102.05107, 2021. a, b, c
Wikner, A., Pathak, J., Hunt, B. R., Szunyogh, I., Girvan, M., and Ott, E.: Using Data Assimilation to Train a Hybrid Forecast System that Combines Machine-Learning and Knowledge-Based Components, arXiv [preprint], arXiv:2102.07819, 2021. a
Xie, J., Xu, L., and Chen, E.: Image denoising and inpainting with deep neural networks, Adv. Neur. In., 25, 341–349, 2012. a
Yin, Y., Alves, O., and Oke, P. R.: An ensemble ocean data assimilation system for seasonal prediction, Mon. Weather Rev., 139, 786–808, 2011. a
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.
There is growing interest in data-driven weather forecasting, i.e., to predict the weather by...
