Articles | Volume 15, issue 8
https://doi.org/10.5194/gmd-15-3417-2022
© 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-15-3417-2022
© 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
| 
28 Apr 2022
Development and technical paper |
| 28 Apr 2022
| 28 Apr 2022
Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0)
Air Quality Research Center, University of California, Davis, Davis,
California 95616, USA
Anthony S. Wexler
CORRESPONDING AUTHOR
Air Quality Research Center, University of California, Davis, Davis,
California 95616, USA
Department of Mechanical and Aerospace Engineering, University of
California, Davis, Davis, California 95616, USA
Department of Civil and
Environmental Engineering, University of
California, Davis, Davis, California 95616, USA
Department of Land, Air and Water Resources, University of
California, Davis, Davis, California 95616, USA
Related authors
Patrick Obin Sturm and Anthony S. Wexler
Geosci. Model Dev., 13, 4435–4442, https://doi.org/10.5194/gmd-13-4435-2020, https://doi.org/10.5194/gmd-13-4435-2020, 2020
Short summary
Short summary
Large air quality and climate models calculate different physical and chemical phenomena in separate operators within the overall model, some of which are computationally intensive. Machine learning tools can memorize the behavior of these operators and replace them, but the replacements must still obey physical laws, like conservation principles. This work derives a mathematical framework for machine learning replacements that conserves properties, such as mass or energy, to machine precision.
Bruno Debus, Andrew T. Weakley, Satoshi Takahama, Kathryn M. George, Anahita Amiri-Farahani, Bret Schichtel, Scott Copeland, Anthony S. Wexler, and Ann M. Dillner
Atmos. Meas. Tech., 15, 2685–2702, https://doi.org/10.5194/amt-15-2685-2022, https://doi.org/10.5194/amt-15-2685-2022, 2022
Short summary
Short summary
In the US, routine particulate matter composition is measured on samples collected on three types of filter media and analyzed using several techniques. We propose an alternate approach that uses one analytical technique, Fourier transform-infrared spectroscopy (FT-IR), and one filter type to measure the chemical composition of particulate matter in a major US monitoring network. This method could be used to add low-cost sites to the network, fill-in missing data, or for quality control.
Christopher D. Wallis, Mason D. Leandro, Patrick Y. Chuang, and Anthony S. Wexler
Atmos. Meas. Tech., 15, 2547–2556, https://doi.org/10.5194/amt-15-2547-2022, https://doi.org/10.5194/amt-15-2547-2022, 2022
Short summary
Short summary
Measuring emissions from stacks requires techniques to address a broad range of conditions and measurement challenges. Here we describe an instrument package held by a crane above a stack to characterize both wet droplet and dried aerosol emissions from cooling tower spray drift in situ. The instrument package characterizes the velocity, size distribution, and concentration of the wet droplet emissions and the mass concentration and elemental composition of the dried PM2.5 and PM10 emissions.
Seyyed Ali Davari and Anthony S. Wexler
Atmos. Meas. Tech., 13, 5369–5377, https://doi.org/10.5194/amt-13-5369-2020, https://doi.org/10.5194/amt-13-5369-2020, 2020
Short summary
Short summary
Traditional instruments for detection and quantification of toxic metals in the atmosphere are expensive. In this study, we have designed, fabricated, and tested a low-cost instrument, which employs cheap components to detect and quantify toxic metals. Advanced machine learning (ML) techniques have been used to improve the instrument's performance. This study demonstrates how the combination of low-cost sensors with ML can address problems that traditionally have been too expensive to be solved.
Patrick Obin Sturm and Anthony S. Wexler
Geosci. Model Dev., 13, 4435–4442, https://doi.org/10.5194/gmd-13-4435-2020, https://doi.org/10.5194/gmd-13-4435-2020, 2020
Short summary
Short summary
Large air quality and climate models calculate different physical and chemical phenomena in separate operators within the overall model, some of which are computationally intensive. Machine learning tools can memorize the behavior of these operators and replace them, but the replacements must still obey physical laws, like conservation principles. This work derives a mathematical framework for machine learning replacements that conserves properties, such as mass or energy, to machine precision.
Satoshi Takahama, Ann M. Dillner, Andrew T. Weakley, Matteo Reggente, Charlotte Bürki, Mária Lbadaoui-Darvas, Bruno Debus, Adele Kuzmiakova, and Anthony S. Wexler
Atmos. Meas. Tech., 12, 525–567, https://doi.org/10.5194/amt-12-525-2019, https://doi.org/10.5194/amt-12-525-2019, 2019
Short summary
Short summary
Mid-infrared spectra of particulate matter (PM) samples are complex but chemically informative and present an opportunity for cost-effective measurement of PM provided that quantitative calibration models can be built. We review an emerging strategy for building statistical calibration models using collocated measurements, interpreting the physical bases for such models and evaluating the suitability of existing calibration models to new samples.
Christopher D. Cappa, Shantanu H. Jathar, Michael J. Kleeman, Kenneth S. Docherty, Jose L. Jimenez, John H. Seinfeld, and Anthony S. Wexler
Atmos. Chem. Phys., 16, 3041–3059, https://doi.org/10.5194/acp-16-3041-2016, https://doi.org/10.5194/acp-16-3041-2016, 2016
Short summary
Short summary
Losses of vapors to walls of chambers can negatively bias SOA formation measurements, consequently leading to low predicted SOA concentrations in air quality models. Here, we show that accounting for such vapor losses leads to substantial increases in the predicted amount of SOA formed from VOCs and to notable increases in the O : C atomic ratio in two US regions. Comparison with a variety of observational data suggests generally improved model performance when vapor wall losses are accounted for.
S. H. Jathar, C. D. Cappa, A. S. Wexler, J. H. Seinfeld, and M. J. Kleeman
Atmos. Chem. Phys., 16, 2309–2322, https://doi.org/10.5194/acp-16-2309-2016, https://doi.org/10.5194/acp-16-2309-2016, 2016
Short summary
Short summary
Multi-generational chemistry schemes applied in regional models do not increase secondary organic aerosol (SOA) mass production relative to traditional "two-product" schemes when both models are fitted to the same chamber data. The multi-generational chemistry schemes do change the predicted composition of SOA and the source attribution of SOA.
C. R. Hoyle, C. Fuchs, E. Järvinen, H. Saathoff, A. Dias, I. El Haddad, M. Gysel, S. C. Coburn, J. Tröstl, A.-K. Bernhammer, F. Bianchi, M. Breitenlechner, J. C. Corbin, J. Craven, N. M. Donahue, J. Duplissy, S. Ehrhart, C. Frege, H. Gordon, N. Höppel, M. Heinritzi, T. B. Kristensen, U. Molteni, L. Nichman, T. Pinterich, A. S. H. Prévôt, M. Simon, J. G. Slowik, G. Steiner, A. Tomé, A. L. Vogel, R. Volkamer, A. C. Wagner, R. Wagner, A. S. Wexler, C. Williamson, P. M. Winkler, C. Yan, A. Amorim, J. Dommen, J. Curtius, M. W. Gallagher, R. C. Flagan, A. Hansel, J. Kirkby, M. Kulmala, O. Möhler, F. Stratmann, D. R. Worsnop, and U. Baltensperger
Atmos. Chem. Phys., 16, 1693–1712, https://doi.org/10.5194/acp-16-1693-2016, https://doi.org/10.5194/acp-16-1693-2016, 2016
Short summary
Short summary
A significant portion of sulphate, an important constituent of atmospheric aerosols, is formed via the aqueous phase oxidation of sulphur dioxide by ozone. The rate of this reaction has previously only been measured over a relatively small temperature range. Here, we use the state of the art CLOUD chamber at CERN to perform the first measurements of this reaction rate in super-cooled droplets, confirming that the existing extrapolation of the reaction rate to sub-zero temperatures is accurate.
S. H. Jathar, C. D. Cappa, A. S. Wexler, J. H. Seinfeld, and M. J. Kleeman
Geosci. Model Dev., 8, 2553–2567, https://doi.org/10.5194/gmd-8-2553-2015, https://doi.org/10.5194/gmd-8-2553-2015, 2015
Short summary
Short summary
Multi-generational oxidation of organic vapors can significantly alter the mass, chemical composition and properties of secondary organic aerosol (SOA). Here, we implement a semi-explicit, constrained multi-generational oxidation model of Cappa and Wilson (2012) in a 3-D air quality model. When compared with results from a current-generation SOA model, we predict similar mass concentrations of SOA but a different chemical composition. O:C ratios of SOA are in line with those measured globally.
J. G. Charrier, N. K. Richards-Henderson, K. J. Bein, A. S. McFall, A. S. Wexler, and C. Anastasio
Atmos. Chem. Phys., 15, 2327–2340, https://doi.org/10.5194/acp-15-2327-2015, https://doi.org/10.5194/acp-15-2327-2015, 2015
Short summary
Short summary
We measured the oxidative potential of airborne particles – a property that has been linked to health problems caused by particles – from different emission source mixtures in Fresno, CA. Copper was responsible for the majority of the oxidative potential (as measured by the DTT assay), followed by unknown species (likely organics) and manganese. Sources of copper-rich particles, including vehicles, had higher oxidative potentials.
Related subject area
Atmospheric sciences
Convective-gust nowcasting based on radar reflectivity and a deep learning algorithm
Self-nested large-eddy simulations in PALM model system v21.10 for offshore wind prediction under different atmospheric stability conditions
How does cloud-radiative heating over the North Atlantic change with grid spacing, convective parameterization, and microphysics scheme in ICON version 2.1.00?
Updated isoprene and terpene emission factors for the Interactive BVOC (iBVOC) emission scheme in the United Kingdom Earth System Model (UKESM1.0)
Technical descriptions of the experimental dynamical downscaling simulations over North America by the CAM–MPAS variable-resolution model
Intercomparison of the weather and climate physics suites of a unified forecast–climate model system (GRIST-A22.7.28) based on single-column modeling
Halogen chemistry in volcanic plumes: a 1D framework based on MOCAGE 1D (version R1.18.1) preparing 3D global chemistry modelling
PyFLEXTRKR: a flexible feature tracking Python software for convective cloud analysis
CLGAN: a generative adversarial network (GAN)-based video prediction model for precipitation nowcasting
Long-term evaluation of surface air pollution in CAMSRA and MERRA-2 global reanalyses over Europe (2003–2020)
Emulating aerosol optics with randomly generated neural networks
Development of an ecophysiology module in the GEOS-Chem chemical transport model version 12.2.0 to represent biosphere–atmosphere fluxes relevant for ozone air quality
Comparison of ozone formation attribution techniques in the northeastern United States
Improving trajectory calculations by FLEXPART 10.4+ using single-image super-resolution
Data fusion uncertainty-enabled methods to map street-scale hourly NO2 in Barcelona: a case study with CALIOPE-Urban v1.0
Forecasting tropical cyclone tracks in the northwestern Pacific based on a deep-learning model
Accelerating models for multiphase chemical kinetics through machine learning with polynomial chaos expansion and neural networks
A machine learning emulator for Lagrangian particle dispersion model footprints: a case study using NAME
Improving the representation of shallow cumulus convection with the simplified-higher-order-closure–mass-flux (SHOC+MF v1.0) approach
ISAT v2.0: an integrated tool for nested-domain configurations and model-ready emission inventories for WRF-AQM
GPU-HADVPPM V1.0: high-efficient parallel GPU design of the Piecewise Parabolic Method (PPM) for horizontal advection in air quality model (CAMx V6.10)
Estimation of CH4 emission based on an advanced 4D-LETKF assimilation system
Accelerated estimation of sea-spray-mediated heat flux using Gaussian quadrature: case studies with a coupled CFSv2.0-WW3 system
AMORE-Isoprene v1.0: a new reduced mechanism for gas-phase isoprene oxidation
A method for generating a quasi-linear convective system suitable for observing system simulation experiments
The second Met Office Unified Model–JULES Regional Atmosphere and Land configuration, RAL2
A dynamic ammonia emission model and the online coupling with WRF–Chem (WRF–SoilN–Chem v1.0): development and regional evaluation in China
SCIATRAN software package (V4.6): update and further development of aerosol, clouds, surface reflectance databases and models
Deep learning models for generation of precipitation maps based on numerical weather prediction
An inconsistency in aviation emissions between CMIP5 and CMIP6 and the implications for short-lived species and their radiative forcing
On the use of Infrared Atmospheric Sounding Interferometer (IASI) spectrally resolved radiances to test the EC-Earth climate model (v3.3.3) in clear-sky conditions
Incorporation of aerosol into the COSPv2 satellite lidar simulator for climate model evaluation
The Fire Inventory from NCAR version 2.5: an updated global fire emissions model for climate and chemistry applications
An approach to refining the ground meteorological observation stations for improving PM2.5 forecasts in Beijing-Tianjin-Hebei region
The impact of altering emission data precision on compression efficiency and accuracy of simulations of the community multiscale air quality model
AerSett v1.0: a simple and straightforward model for the settling speed of big spherical atmospheric aerosols
The development and validation of an Inhomogeneous Wind Scheme for Urban Street
Optimization of weather forecasting for cloud cover over the European domain using the meteorological component of the Ensemble for Stochastic Integration of Atmospheric Simulations version 1.0
Bayesian transdimensional inverse reconstruction of the Fukushima Daiichi caesium 137 release
Implementation of HONO into the chemistry–climate model CHASER (V4.0): roles in tropospheric chemistry
Isoprene and monoterpene simulations using the chemistry–climate model EMAC (v2.55) with interactive vegetation from LPJ-GUESS (v4.0)
A modern-day Mars climate in the Met Office Unified Model: dry simulations
The AirGAM 2022r1 air quality trend and prediction model
Evaluation of a cloudy cold-air pool in the Columbia River basin in different versions of the High-Resolution Rapid Refresh (HRRR) model
Assessment of WRF (v 4.2.1) dynamically downscaled precipitation on subdaily and daily timescales over CONUS
Comparing Sentinel-5P TROPOMI NO2 column observations with the CAMS regional air quality ensemble
Dynamic Meteorology-Induced Emissions Coupler (MetEmis) development in the Community Multiscale Air Quality (CMAQ): CMAQ-MetEmis
Cross-evaluating WRF-Chem v4.1.2, TROPOMI, APEX, and in situ NO2 measurements over Antwerp, Belgium
Variability and combination as ensemble of mineral dust forecast during the 2021 CADDIWA experiment
Adapting a deep convolutional RNN model with imbalanced regression loss for improved spatio-temporal forecasting of extreme wind speed events in the short to medium range
Haixia Xiao, Yaqiang Wang, Yu Zheng, Yuanyuan Zheng, Xiaoran Zhuang, Hongyan Wang, and Mei Gao
Geosci. Model Dev., 16, 3611–3628, https://doi.org/10.5194/gmd-16-3611-2023, https://doi.org/10.5194/gmd-16-3611-2023, 2023
Short summary
Short summary
Due to the small-scale and nonstationary nature of convective wind gusts (CGs), reliable CG nowcasting has remained unattainable. Here, we developed a deep learning model — namely CGsNet — for 0—2 h of quantitative CG nowcasting, first achieving minute—kilometer-level forecasts. Based on the CGsNet model, the average surface wind speed (ASWS) and peak wind gust speed (PWGS) predictions are obtained. Experiments indicate that CGsNet exhibits higher accuracy than the traditional method.
Maria Krutova, Mostafa Bakhoday-Paskyabi, Joachim Reuder, and Finn Gunnar Nielsen
Geosci. Model Dev., 16, 3553–3564, https://doi.org/10.5194/gmd-16-3553-2023, https://doi.org/10.5194/gmd-16-3553-2023, 2023
Short summary
Short summary
Local refinement of the grid is a powerful method allowing us to reduce the computational time while preserving the accuracy in the area of interest. Depending on the implementation, the local refinement may introduce unwanted numerical effects into the results. We study the wind speed common to the wind turbine operational speeds and confirm strong alteration of the result when the heat fluxes are present, except for the specific refinement scheme used.
Sylvia Sullivan, Behrooz Keshtgar, Nicole Albern, Elzina Bala, Christoph Braun, Anubhav Choudhary, Johannes Hörner, Hilke Lentink, Georgios Papavasileiou, and Aiko Voigt
Geosci. Model Dev., 16, 3535–3551, https://doi.org/10.5194/gmd-16-3535-2023, https://doi.org/10.5194/gmd-16-3535-2023, 2023
Short summary
Short summary
Clouds absorb and re-emit infrared radiation from Earth's surface and absorb and reflect incoming solar radiation. As a result, they change atmospheric temperature gradients that drive large-scale circulation. To better simulate this circulation, we study how the radiative heating and cooling from clouds depends on model settings like grid spacing; whether we describe convection approximately or exactly; and the level of detail used to describe small-scale processes, or microphysics, in clouds.
James Weber, James A. King, Katerina Sindelarova, and Maria Val Martin
Geosci. Model Dev., 16, 3083–3101, https://doi.org/10.5194/gmd-16-3083-2023, https://doi.org/10.5194/gmd-16-3083-2023, 2023
Short summary
Short summary
The emissions of volatile organic compounds from vegetation (BVOCs) influence atmospheric composition and contribute to certain gases and aerosols (tiny airborne particles) which play a role in climate change. BVOC emissions are likely to change in the future due to changes in climate and land use. Therefore, accurate simulation of BVOC emission is important, and this study describes an update to the simulation of BVOC emissions in the United Kingdom Earth System Model (UKESM).
Koichi Sakaguchi, L. Ruby Leung, Colin M. Zarzycki, Jihyeon Jang, Seth McGinnis, Bryce E. Harrop, William C. Skamarock, Andrew Gettelman, Chun Zhao, William J. Gutowski, Stephen Leak, and Linda Mearns
Geosci. Model Dev., 16, 3029–3081, https://doi.org/10.5194/gmd-16-3029-2023, https://doi.org/10.5194/gmd-16-3029-2023, 2023
Short summary
Short summary
We document details of the regional climate downscaling dataset produced by a global variable-resolution model. The experiment is unique in that it follows a standard protocol designed for coordinated experiments of regional models. We found negligible influence of post-processing on statistical analysis, importance of simulation quality outside of the target region, and computational challenges that our model code faced due to rapidly changing super computer systems.
Xiaohan Li, Yi Zhang, Xindong Peng, Baiquan Zhou, Jian Li, and Yiming Wang
Geosci. Model Dev., 16, 2975–2993, https://doi.org/10.5194/gmd-16-2975-2023, https://doi.org/10.5194/gmd-16-2975-2023, 2023
Short summary
Short summary
The weather and climate physics suites used in GRIST-A22.7.28 are compared using single-column modeling. The source of their discrepancies in terms of modeling cloud and precipitation is explored. Convective parameterization is found to be a key factor responsible for the differences. The two suites also have intrinsic differences in the interaction between microphysics and other processes, resulting in different cloud features and time step sensitivities.
Virginie Marécal, Ronan Voisin-Plessis, Tjarda Jane Roberts, Alessandro Aiuppa, Herizo Narivelo, Paul David Hamer, Béatrice Josse, Jonathan Guth, Luke Surl, and Lisa Grellier
Geosci. Model Dev., 16, 2873–2898, https://doi.org/10.5194/gmd-16-2873-2023, https://doi.org/10.5194/gmd-16-2873-2023, 2023
Short summary
Short summary
We implemented a halogen volcanic chemistry scheme in a one-dimensional modelling framework preparing for further use in a three-dimensional global chemistry-transport model. The results of the simulations for an eruption of Mt Etna in 2008, including various sensitivity tests, show a good consistency with previous modelling studies.
Zhe Feng, Joseph Hardin, Hannah C. Barnes, Jianfeng Li, L. Ruby Leung, Adam Varble, and Zhixiao Zhang
Geosci. Model Dev., 16, 2753–2776, https://doi.org/10.5194/gmd-16-2753-2023, https://doi.org/10.5194/gmd-16-2753-2023, 2023
Short summary
Short summary
PyFLEXTRKR is a flexible atmospheric feature tracking framework with specific capabilities to track convective clouds from a variety of observations and model simulations. The package has a collection of multi-object identification algorithms and has been optimized for large datasets. This paper describes the algorithms and demonstrates applications for tracking deep convective cells and mesoscale convective systems from observations and model simulations at a wide range of scales.
Yan Ji, Bing Gong, Michael Langguth, Amirpasha Mozaffari, and Xiefei Zhi
Geosci. Model Dev., 16, 2737–2752, https://doi.org/10.5194/gmd-16-2737-2023, https://doi.org/10.5194/gmd-16-2737-2023, 2023
Short summary
Short summary
Formulating short-term precipitation forecasting as a video prediction task, a novel deep learning architecture (convolutional long short-term memory generative adversarial network, CLGAN) is proposed. A benchmark dataset is built on minute-level precipitation measurements. Results show that with the GAN component the model generates predictions sharing statistical properties with observations, resulting in it outperforming the baseline in dichotomous and spatial scores for heavy precipitation.
Aleksander Lacima, Hervé Petetin, Albert Soret, Dene Bowdalo, Oriol Jorba, Zhaoyue Chen, Raúl F. Méndez Turrubiates, Hicham Achebak, Joan Ballester, and Carlos Pérez García-Pando
Geosci. Model Dev., 16, 2689–2718, https://doi.org/10.5194/gmd-16-2689-2023, https://doi.org/10.5194/gmd-16-2689-2023, 2023
Short summary
Short summary
Understanding how air pollution varies across space and time is of key importance for the safeguarding of human health. This work arose in the context of the project EARLY-ADAPT, for which the Barcelona Supercomputing Center developed an air pollution database covering all of Europe. Through different statistical methods, we compared two global pollution models against measurements from ground stations and found significant discrepancies between the observed and the modeled surface pollution.
Andrew Geiss, Po-Lun Ma, Balwinder Singh, and Joseph C. Hardin
Geosci. Model Dev., 16, 2355–2370, https://doi.org/10.5194/gmd-16-2355-2023, https://doi.org/10.5194/gmd-16-2355-2023, 2023
Short summary
Short summary
Atmospheric aerosols play a critical role in Earth's climate, but it is too computationally expensive to directly model their interaction with radiation in climate simulations. This work develops a new neural-network-based parameterization of aerosol optical properties for use in the Energy Exascale Earth System Model that is much more accurate than the current one; it also introduces a unique model optimization method that involves randomly generating neural network architectures.
Joey C. Y. Lam, Amos P. K. Tai, Jason A. Ducker, and Christopher D. Holmes
Geosci. Model Dev., 16, 2323–2342, https://doi.org/10.5194/gmd-16-2323-2023, https://doi.org/10.5194/gmd-16-2323-2023, 2023
Short summary
Short summary
We developed a new component within an atmospheric chemistry model to better simulate plant ecophysiological processes relevant for ozone air quality. We showed that it reduces simulated biases in plant uptake of ozone in prior models. The new model enables us to explore how future climatic changes affect air quality via affecting plants, examine ozone–vegetation interactions and feedbacks, and evaluate the impacts of changing atmospheric chemistry and climate on vegetation productivity.
Qian Shu, Sergey L. Napelenok, William T. Hutzell, Kirk R. Baker, Barron H. Henderson, Benjamin N. Murphy, and Christian Hogrefe
Geosci. Model Dev., 16, 2303–2322, https://doi.org/10.5194/gmd-16-2303-2023, https://doi.org/10.5194/gmd-16-2303-2023, 2023
Short summary
Short summary
Source attribution methods are generally used to determine culpability of precursor emission sources to ambient pollutant concentrations. However, source attribution of secondarily formed pollutants such as ozone and its precursors cannot be explicitly measured, making evaluation of source apportionment methods challenging. In this study, multiple apportionment approach comparisons show common features but still reveal wide variations in predicted sector contribution and species dependency.
Rüdiger Brecht, Lucie Bakels, Alex Bihlo, and Andreas Stohl
Geosci. Model Dev., 16, 2181–2192, https://doi.org/10.5194/gmd-16-2181-2023, https://doi.org/10.5194/gmd-16-2181-2023, 2023
Short summary
Short summary
We use neural-network-based single-image super-resolution to improve the upscaling of meteorological wind fields to be used for particle dispersion models. This deep-learning-based methodology improves the standard linear interpolation typically used in particle dispersion models. The improvement of wind fields leads to substantial improvement in the computed trajectories of the particles.
Alvaro Criado, Jan Mateu Armengol, Hervé Petetin, Daniel Rodriguez-Rey, Jaime Benavides, Marc Guevara, Carlos Pérez García-Pando, Albert Soret, and Oriol Jorba
Geosci. Model Dev., 16, 2193–2213, https://doi.org/10.5194/gmd-16-2193-2023, https://doi.org/10.5194/gmd-16-2193-2023, 2023
Short summary
Short summary
This work aims to derive and evaluate a general statistical post-processing tool specifically designed for the street scale that can be applied to any urban air quality system. Our data fusion methodology corrects NO2 fields based on continuous hourly observations and experimental campaigns. This study enables us to obtain exceedance probability maps of air quality standards. In 2019, 13 % of the Barcelona area had a 70 % or higher probability of exceeding the annual legal NO2 limit of 40 µg/m3.
Liang Wang, Bingcheng Wan, Shaohui Zhou, Haofei Sun, and Zhiqiu Gao
Geosci. Model Dev., 16, 2167–2179, https://doi.org/10.5194/gmd-16-2167-2023, https://doi.org/10.5194/gmd-16-2167-2023, 2023
Short summary
Short summary
The past 24 h TC trajectories and meteorological field data were used to forecast TC tracks in the northwestern Pacific from hours 6–72 based on GRU_CNN, which we proposed in this paper and which has better prediction results than traditional single deep-learning methods. The historical steering flow of cyclones has a significant effect on improving the accuracy of short-term forecasting, while, in long-term forecasting, the SST and geopotential height will have a particular impact.
Thomas Berkemeier, Matteo Krüger, Aryeh Feinberg, Marcel Müller, Ulrich Pöschl, and Ulrich K. Krieger
Geosci. Model Dev., 16, 2037–2054, https://doi.org/10.5194/gmd-16-2037-2023, https://doi.org/10.5194/gmd-16-2037-2023, 2023
Short summary
Short summary
Kinetic multi-layer models (KMs) successfully describe heterogeneous and multiphase atmospheric chemistry. In applications requiring repeated execution, however, these models can be too expensive. We trained machine learning surrogate models on output of the model KM-SUB and achieved high correlations. The surrogate models run orders of magnitude faster, which suggests potential applicability in global optimization tasks and as sub-modules in large-scale atmospheric models.
Elena Fillola, Raul Santos-Rodriguez, Alistair Manning, Simon O'Doherty, and Matt Rigby
Geosci. Model Dev., 16, 1997–2009, https://doi.org/10.5194/gmd-16-1997-2023, https://doi.org/10.5194/gmd-16-1997-2023, 2023
Short summary
Short summary
Lagrangian particle dispersion models are used extensively for the estimation of greenhouse gas (GHG) fluxes using atmospheric observations. However, these models do not scale well as data volumes increase. Here, we develop a proof-of-concept machine learning emulator that can produce outputs similar to those of the dispersion model, but 50 000 times faster, using only meteorological inputs. This works demonstrates the potential of machine learning to accelerate GHG estimations across the globe.
Maria J. Chinita, Mikael Witte, Marcin J. Kurowski, Joao Teixeira, Kay Suselj, Georgios Matheou, and Peter Bogenschutz
Geosci. Model Dev., 16, 1909–1924, https://doi.org/10.5194/gmd-16-1909-2023, https://doi.org/10.5194/gmd-16-1909-2023, 2023
Short summary
Short summary
Low clouds are one of the largest sources of uncertainty in climate prediction. In this paper, we introduce the first version of the unified turbulence and shallow convection parameterization named SHOC+MF developed to improve the representation of shallow cumulus clouds in the Simple Cloud-Resolving E3SM Atmosphere Model (SCREAM). Here, we also show promising preliminary results in a single-column model framework for two benchmark cases of shallow cumulus convection.
Kun Wang, Chao Gao, Kai Wu, Kaiyun Liu, Haofan Wang, Mo Dan, Xiaohui Ji, and Qingqing Tong
Geosci. Model Dev., 16, 1961–1973, https://doi.org/10.5194/gmd-16-1961-2023, https://doi.org/10.5194/gmd-16-1961-2023, 2023
Short summary
Short summary
This study establishes an easy-to-use and integrated framework for a model-ready emission inventory for the Weather Research and Forecasting (WRF)–Air Quality Numerical Model (AQM). A free tool called the ISAT (Inventory Spatial Allocation Tool) was developed based on this framework. ISAT helps users complete the workflow from the WRF nested-domain configuration to a model-ready emission inventory for AQM with a regional emission inventory and a shapefile for the target region.
Kai Cao, Qizhong Wu, Lingling Wang, Nan Wang, Huaqiong Cheng, Xiao Tang, Dongqing Li, and Lanning Wang
EGUsphere, https://doi.org/10.5194/egusphere-2023-410, https://doi.org/10.5194/egusphere-2023-410, 2023
Short summary
Short summary
Offline performance experiments results show that the GPU-HADVPPM on V100 GPU can achieve up to more than one thousand (1113.6x) speedup to its original version on E5-2682 v4 CPU. A series of optimization measures are taken, the CAMx-CUDA model improves the computing efficiency by 128.4x on a single V100 GPU card. A parallel architecture with an MPI+CUDA hybrid paradigm is presented, and it can achieve up to 4.5x speedup when launch 8 CPU cores and 8 GPU cards.
Jagat S. H. Bisht, Prabir K. Patra, Masayuki Takigawa, Takashi Sekiya, Yugo Kanaya, Naoko Saitoh, and Kazuyuki Miyazaki
Geosci. Model Dev., 16, 1823–1838, https://doi.org/10.5194/gmd-16-1823-2023, https://doi.org/10.5194/gmd-16-1823-2023, 2023
Short summary
Short summary
In this study, we estimated CH4 fluxes using an advanced 4D-LETKF method. The system was tested and optimized using observation system simulation experiments (OSSEs), where a known surface emission distribution is retrieved from synthetic observations. The availability of satellite measurements has increased, and there are still many missions focused on greenhouse gas observations that have not yet launched. The technique being referred to has the potential to improve estimates of CH4 fluxes.
Ruizi Shi and Fanghua Xu
Geosci. Model Dev., 16, 1839–1856, https://doi.org/10.5194/gmd-16-1839-2023, https://doi.org/10.5194/gmd-16-1839-2023, 2023
Short summary
Short summary
Based on the Gaussian quadrature method, a fast algorithm of sea-spray-mediated heat flux is developed. Compared with the widely used single-radius algorithm, the new fast algorithm shows a better agreement with the full spectrum integral of spray flux. The new fast algorithm is evaluated in a coupled modeling system, and the simulations of sea surface temperature, wind speed and wave height are improved. Thereby, the new fast algorithm has great potential to be used in coupled modeling systems.
Forwood Wiser, Bryan K. Place, Siddhartha Sen, Havala O. T. Pye, Benjamin Yang, Daniel M. Westervelt, Daven K. Henze, Arlene M. Fiore, and V. Faye McNeill
Geosci. Model Dev., 16, 1801–1821, https://doi.org/10.5194/gmd-16-1801-2023, https://doi.org/10.5194/gmd-16-1801-2023, 2023
Short summary
Short summary
We developed a reduced model of atmospheric isoprene oxidation, AMORE-Isoprene 1.0. It was created using a new Automated Model Reduction (AMORE) method designed to simplify complex chemical mechanisms with minimal manual adjustments to the output. AMORE-Isoprene 1.0 has improved accuracy and similar size to other reduced isoprene mechanisms. When included in the CRACMM mechanism, it improved the accuracy of EPA’s CMAQ model predictions for the northeastern USA compared to observations.
Jonathan D. Labriola, Jeremy A. Gibbs, and Louis J. Wicker
Geosci. Model Dev., 16, 1779–1799, https://doi.org/10.5194/gmd-16-1779-2023, https://doi.org/10.5194/gmd-16-1779-2023, 2023
Short summary
Short summary
Observing system simulation experiments (OSSEs) are simulated case studies used to understand how different assimilated weather observations impact forecast skill. This study introduces the methods used to create an OSSE for a tornadic quasi-linear convective system event. These steps provide an opportunity to simulate a realistic high-impact weather event and can be used to encourage a more diverse set of OSSEs.
Mike Bush, Ian Boutle, John Edwards, Anke Finnenkoetter, Charmaine Franklin, Kirsty Hanley, Aravindakshan Jayakumar, Huw Lewis, Adrian Lock, Marion Mittermaier, Saji Mohandas, Rachel North, Aurore Porson, Belinda Roux, Stuart Webster, and Mark Weeks
Geosci. Model Dev., 16, 1713–1734, https://doi.org/10.5194/gmd-16-1713-2023, https://doi.org/10.5194/gmd-16-1713-2023, 2023
Short summary
Short summary
Building on the baseline of RAL1, the RAL2 science configuration is used for regional modelling around the UM partnership and in operations at the Met Office. RAL2 has been tested in different parts of the world including Australia, India and the UK. RAL2 increases medium and low cloud amounts in the mid-latitudes compared to RAL1, leading to improved cloud forecasts and a reduced diurnal cycle of screen temperature. There is also a reduction in the frequency of heavier precipitation rates.
Chuanhua Ren, Xin Huang, Tengyu Liu, Yu Song, Zhang Wen, Xuejun Liu, Aijun Ding, and Tong Zhu
Geosci. Model Dev., 16, 1641–1659, https://doi.org/10.5194/gmd-16-1641-2023, https://doi.org/10.5194/gmd-16-1641-2023, 2023
Short summary
Short summary
Ammonia in the atmosphere has wide impacts on the ecological environment and air quality, and its emission from soil volatilization is highly sensitive to meteorology, making it challenging to be well captured in models. We developed a dynamic emission model capable of calculating ammonia emission interactively with meteorological and soil conditions. Such a coupling of soil emission with meteorology provides a better understanding of ammonia emission and its contribution to atmospheric aerosol.
Linlu Mei, Vladimir Rozanov, Alexei Rozanov, and John P. Burrows
Geosci. Model Dev., 16, 1511–1536, https://doi.org/10.5194/gmd-16-1511-2023, https://doi.org/10.5194/gmd-16-1511-2023, 2023
Short summary
Short summary
This paper summarizes recent developments of aerosol, cloud and surface reflectance databases and models in the framework of the software package SCIATRAN. These updates and developments extend the capabilities of the radiative transfer modeling, especially by accounting for different kinds of vertical inhomogeneties. Vertically inhomogeneous clouds and different aerosol types can be easily accounted for within SCIATRAN (V4.6). The widely used surface models and databases are now available.
Adrian Rojas-Campos, Michael Langguth, Martin Wittenbrink, and Gordon Pipa
Geosci. Model Dev., 16, 1467–1480, https://doi.org/10.5194/gmd-16-1467-2023, https://doi.org/10.5194/gmd-16-1467-2023, 2023
Short summary
Short summary
Our paper presents an alternative approach for generating high-resolution precipitation maps based on the nonlinear combination of the complete set of variables of the numerical weather predictions. This process combines the super-resolution task with the bias correction in a single step, generating high-resolution corrected precipitation maps with a lead time of 3 h. We used using deep learning algorithms to combine the input information and increase the accuracy of the precipitation maps.
Robin N. Thor, Mariano Mertens, Sigrun Matthes, Mattia Righi, Johannes Hendricks, Sabine Brinkop, Phoebe Graf, Volker Grewe, Patrick Jöckel, and Steven Smith
Geosci. Model Dev., 16, 1459–1466, https://doi.org/10.5194/gmd-16-1459-2023, https://doi.org/10.5194/gmd-16-1459-2023, 2023
Short summary
Short summary
We report on an inconsistency in the latitudinal distribution of aviation emissions between two versions of a data product which is widely used by researchers. From the available documentation, we do not expect such an inconsistency. We run a chemistry–climate model to compute the effect of the inconsistency in emissions on atmospheric chemistry and radiation and find that the radiative forcing associated with aviation ozone is 7.6 % higher when using the less recent version of the data.
Stefano Della Fera, Federico Fabiano, Piera Raspollini, Marco Ridolfi, Ugo Cortesi, Flavio Barbara, and Jost von Hardenberg
Geosci. Model Dev., 16, 1379–1394, https://doi.org/10.5194/gmd-16-1379-2023, https://doi.org/10.5194/gmd-16-1379-2023, 2023
Short summary
Short summary
The long-term comparison between observed and simulated outgoing longwave radiances represents a strict test to evaluate climate model performance. In this work, 9 years of synthetic spectrally resolved radiances, simulated online on the basis of the atmospheric fields predicted by the EC-Earth global climate model (v3.3.3) in clear-sky conditions, are compared to IASI spectral radiance climatology in order to detect model biases in temperature and humidity at different atmospheric levels.
Marine Bonazzola, Hélène Chepfer, Po-Lun Ma, Johannes Quaas, David M. Winker, Artem Feofilov, and Nick Schutgens
Geosci. Model Dev., 16, 1359–1377, https://doi.org/10.5194/gmd-16-1359-2023, https://doi.org/10.5194/gmd-16-1359-2023, 2023
Short summary
Short summary
Aerosol has a large impact on climate. Using a lidar aerosol simulator ensures consistent comparisons between modeled and observed aerosol. We present a lidar aerosol simulator that applies a cloud masking and an aerosol detection threshold. We estimate the lidar signals that would be observed at 532 nm by the Cloud-Aerosol Lidar with Orthogonal Polarization overflying the atmosphere predicted by a climate model. Our comparison at the seasonal timescale shows a discrepancy in the Southern Ocean.
Christine Wiedinmyer, Yosuke Kimura, Elena C. McDonald-Buller, Louisa K. Emmons, Rebecca R. Buchholz, Wenfu Tang, Keenan Seto, Maxwell B. Joseph, Kelley C. Barsanti, Annmarie G. Carlton, and Robert Yokelson
EGUsphere, https://doi.org/10.5194/egusphere-2023-124, https://doi.org/10.5194/egusphere-2023-124, 2023
Short summary
Short summary
The Fire INventory from NCAR (FINN) provides daily, global estimates of emissions from open fires based on satellite detections of hot spots. This version has been updated to apply MODIS and VIIRS satellite fire detections, and better represents both large and small fires.. FINNv2.5 generates more emissions than FINNv1, in general agreement with other fire emissions inventories. The new estimates are consistent with satellite observations, but uncertainties remain regionally and by pollutant.
Lichao Yang, Wansuo Duan, and Zifa Wang
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2023-10, https://doi.org/10.5194/gmd-2023-10, 2023
Revised manuscript accepted for GMD
Short summary
Short summary
We refine the ground meteorological stations by a nonlinear approach for improving the regional PM2.5 forecasts. The refined observation network (about 60 % of the current stations) can achieve almost the same improvements in PM2.5 forecasts as all the current station observations. The study will provide a scientific guidance to optimize the ground meteorological stations relative to PM2.5 forecasts and suggests an idea of cost-effective data assimilation for enhancing the PM2.5 forecast skills.
Michael S. Walters and David C. Wong
Geosci. Model Dev., 16, 1179–1190, https://doi.org/10.5194/gmd-16-1179-2023, https://doi.org/10.5194/gmd-16-1179-2023, 2023
Short summary
Short summary
A typical numerical simulation that associates with a large amount of input and output data, applying popular compression software, gzip or bzip2, on data is one good way to mitigate data storage burden. This article proposes a simple technique to alter input, output, or input and output by keeping a specific number of significant digits in data and demonstrates an enhancement in compression efficiency on the altered data but maintains similar statistical performance of the numerical simulation.
Sylvain Mailler, Laurent Menut, Arineh Cholakian, and Romain Pennel
Geosci. Model Dev., 16, 1119–1127, https://doi.org/10.5194/gmd-16-1119-2023, https://doi.org/10.5194/gmd-16-1119-2023, 2023
Short summary
Short summary
Large or even
giantparticles of mineral dust exist in the atmosphere but, so far, solving an non-linear equation was needed to calculate the speed at which they fall in the atmosphere. The model we present, AerSett v1.0 (AERosol SETTling version 1.0), provides a new and simple way of calculating their free-fall velocity in the atmosphere, which will be useful to anyone trying to understand and represent adequately the transport of giant dust particles by the wind.
Yuanhao Chen, Zhenxin Liu, Yuhang Wang, Cheng Liu, Shuhua Liu, and Hong Liao
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-267, https://doi.org/10.5194/gmd-2022-267, 2023
Revised manuscript accepted for GMD
Short summary
Short summary
The heterogenous layout of urban buildings leads to the complex wind field in and over the urban. Large discrepancies exist between the observation and the current simulations without considering the character. The Inhomogeneous Wind Scheme in Urban Street (IWSUS) is developed to simulation the complex wind speed in a typical street. It can better simulate the wind field in and over the inhomogenous street canyon then improve the simulated energy budget in lower atmosphere layer over the urban.
Yen-Sen Lu, Garrett H. Good, and Hendrik Elbern
Geosci. Model Dev., 16, 1083–1104, https://doi.org/10.5194/gmd-16-1083-2023, https://doi.org/10.5194/gmd-16-1083-2023, 2023
Short summary
Short summary
The Weather Forecasting and Research (WRF) model consists of many parameters and options that can be adapted to different conditions. This expansive sensitivity study uses a large-scale simulation system to determine the most suitable options for predicting cloud cover in Europe for deterministic and probabilistic weather predictions for day-ahead forecasting simulations.
Joffrey Dumont Le Brazidec, Marc Bocquet, Olivier Saunier, and Yelva Roustan
Geosci. Model Dev., 16, 1039–1052, https://doi.org/10.5194/gmd-16-1039-2023, https://doi.org/10.5194/gmd-16-1039-2023, 2023
Short summary
Short summary
When radionuclides are released into the atmosphere, the assessment of the consequences depends on the evaluation of the magnitude and temporal evolution of the release, which can be highly variable as in the case of Fukushima Daiichi.
Here, we propose Bayesian inverse modelling methods and the reversible-jump Markov chain Monte Carlo technique, which allows one to evaluate the temporal variability of the release and to integrate different types of information in the source reconstruction.
Phuc Thi Minh Ha, Yugo Kanaya, Fumikazu Taketani, Maria Dolores Andrés Hernández, Benjamin Schreiner, Klaus Pfeilsticker, and Kengo Sudo
Geosci. Model Dev., 16, 927–960, https://doi.org/10.5194/gmd-16-927-2023, https://doi.org/10.5194/gmd-16-927-2023, 2023
Short summary
Short summary
HONO affects tropospheric oxidizing capacity; thus, it is implemented into the chemistry–climate model CHASER. The model substantially underpredicts daytime HONO, while nitrate photolysis on surfaces can supplement the daytime HONO budget. Current HONO chemistry predicts reductions of 20.4 % for global tropospheric NOx, 40–67 % for OH, and 30–45 % for O3 in the summer North Pacific. In contrast, OH and O3 winter levels in China are greatly enhanced.
Ryan Vella, Matthew Forrest, Jos Lelieveld, and Holger Tost
Geosci. Model Dev., 16, 885–906, https://doi.org/10.5194/gmd-16-885-2023, https://doi.org/10.5194/gmd-16-885-2023, 2023
Short summary
Short summary
Biogenic volatile organic compounds (BVOCs) are released by vegetation and have a major impact on atmospheric chemistry and aerosol formation. Non-interacting vegetation constrains the majority of numerical models used to estimate global BVOC emissions, and thus, the effects of changing vegetation on emissions are not addressed. In this work, we replace the offline vegetation with dynamic vegetation states by linking a chemistry–climate model with a global dynamic vegetation model.
Danny McCulloch, Denis E. Sergeev, Nathan Mayne, Matthew Bate, James Manners, Ian Boutle, Benjamin Drummond, and Kristzian Kohary
Geosci. Model Dev., 16, 621–657, https://doi.org/10.5194/gmd-16-621-2023, https://doi.org/10.5194/gmd-16-621-2023, 2023
Short summary
Short summary
We present results from the Met Office Unified Model (UM) to study the dry Martian climate. We describe our model set-up conditions and run two scenarios, with radiatively active/inactive dust. We compare both scenarios to results from an existing Mars climate model, the planetary climate model. We find good agreement in winds and air temperatures, but dust amounts differ between models. This study highlights the importance of using the UM for future Mars research.
Sam-Erik Walker, Sverre Solberg, Philipp Schneider, and Cristina Guerreiro
Geosci. Model Dev., 16, 573–595, https://doi.org/10.5194/gmd-16-573-2023, https://doi.org/10.5194/gmd-16-573-2023, 2023
Short summary
Short summary
We have developed a statistical model for estimating trends in the daily air quality observations of NO2, O3, PM10 and PM2.5, adjusting for trends and short-term variations in meteorology. The model is general and may also be used for prediction purposes, including forecasting. It has been applied in a recent comprehensive study in Europe. Significant declines are shown for the pollutants from 2005 to 2019, mainly due to reductions in emissions not attributable to changes in meteorology.
Bianca Adler, James M. Wilczak, Jaymes Kenyon, Laura Bianco, Irina V. Djalalova, Joseph B. Olson, and David D. Turner
Geosci. Model Dev., 16, 597–619, https://doi.org/10.5194/gmd-16-597-2023, https://doi.org/10.5194/gmd-16-597-2023, 2023
Short summary
Short summary
Rapid changes in wind speed make the integration of wind energy produced during persistent orographic cold-air pools difficult to integrate into the electrical grid. By evaluating three versions of NOAA’s High-Resolution Rapid Refresh model, we demonstrate how model developments targeted during the second Wind Forecast Improvement Project improve the forecast of a persistent cold-air pool event.
Abhishekh Kumar Srivastava, Paul Aaron Ullrich, Deeksha Rastogi, Pouya Vahmani, Andrew Jones, and Richard Grotjahn
EGUsphere, https://doi.org/10.5194/egusphere-2022-1382, https://doi.org/10.5194/egusphere-2022-1382, 2023
Short summary
Short summary
Stakeholders need high-resolution regional climate data for applications such as assessing water availability, mountain snowpack, etc. This study examines 3- and 24-hr historical precipitation over the contiguous United States in the 12-km WRF version 4.2.1-based dynamical downscaling of the ERA5 reanalysis. WRF improves precipitation characteristics such as the annual cycle and distribution of the precipitation maxima, but it also displays regionally and seasonally varying precipitation biases.
John Douros, Henk Eskes, Jos van Geffen, K. Folkert Boersma, Steven Compernolle, Gaia Pinardi, Anne-Marlene Blechschmidt, Vincent-Henri Peuch, Augustin Colette, and Pepijn Veefkind
Geosci. Model Dev., 16, 509–534, https://doi.org/10.5194/gmd-16-509-2023, https://doi.org/10.5194/gmd-16-509-2023, 2023
Short summary
Short summary
We focus on the challenges associated with comparing atmospheric composition models with satellite products such as tropospheric NO2 columns. The aim is to highlight the methodological difficulties and propose sound ways of doing such comparisons. Building on the comparisons, a new satellite product is proposed and made available, which takes advantage of higher-resolution, regional atmospheric modelling to improve estimates of troposheric NO2 columns over Europe.
Bok H. Baek, Carlie Coats, Siqi Ma, Chi-Tsan Wang, Jia Xing, Daniel Tong, Soontae Kim, and Jung-Hun Woo
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-253, https://doi.org/10.5194/gmd-2022-253, 2023
Revised manuscript accepted for GMD
Short summary
Short summary
To enable the direct feedback effects of aerosols and local meteorology in an air quality modeling system without any computational bottleneck, we have developed an “inline” meteorology-induce emissions coupler module within the US EPA’s CMAQ modeling system, called “Meteorologically-induced anthropogenic Emissions: CMAQ-MetEmis”, to dynamically model the complex MOVES onroad mobile emissions inline without a separate dedicated emissions processing model like SMOKE.
Catalina Poraicu, Jean-François Müller, Trissevgeni Stavrakou, Dominique Fonteyn, Frederik Tack, Felix Deutsch, Quentin Laffineur, Roeland Van Malderen, and Nele Veldeman
Geosci. Model Dev., 16, 479–508, https://doi.org/10.5194/gmd-16-479-2023, https://doi.org/10.5194/gmd-16-479-2023, 2023
Short summary
Short summary
High-resolution WRF-Chem simulations are conducted over Antwerp, Belgium, in June 2019 and evaluated using meteorological data and in situ, airborne, and spaceborne NO2 measurements. An intercomparison of model, aircraft, and TROPOMI NO2 columns is conducted to characterize biases in versions 1.3.1 and 2.3.1 of the satellite product. A mass balance method is implemented to provide improved emissions for simulating NO2 distribution over the study area.
Laurent Menut
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2022-306, https://doi.org/10.5194/gmd-2022-306, 2023
Revised manuscript accepted for GMD
Short summary
Short summary
This study analyzes the forecast that were made in 2021 to help trigger measurements during the CADDIWA experiment. The WRF-CHIMERE models were run each day and the first goal is to quantify the variability of the forecast as a function of forecast leads and forecast location. The possibility of using the different leads as an ensemble is also tested. For some locations, the correlation scores are better with this approach. This could be tested on operational forecast chains in the future.
Daan R. Scheepens, Irene Schicker, Kateřina Hlaváčková-Schindler, and Claudia Plant
Geosci. Model Dev., 16, 251–270, https://doi.org/10.5194/gmd-16-251-2023, https://doi.org/10.5194/gmd-16-251-2023, 2023
Short summary
Short summary
The production of wind energy is increasing rapidly and relies heavily on atmospheric conditions. To ensure power grid stability, accurate predictions of wind speed are needed, especially in the short range and for extreme wind speed ranges. In this work, we demonstrate the forecasting skills of a data-driven deep learning model with model adaptations to suit higher wind speed ranges. The resulting model can be applied to other data and parameters, too, to improve nowcasting predictions.
Cited articles
Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C.,
Corrado, G. S., Davis, A., Dean, J., Devin, M. Ghemawat, S., Goodfellow, I.,
Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, L.,
Kudlur, M., Levenberg, J., Mané, D., Monga, R., Moore, S., Murray, D.,
Olah, C., Schuster, M., Shlens, J., Steiner, B., Sutskever, I., Talwar, K.,
Tucker, P., Vanhoucke, V., Vasudevan, V., Viégas, F., Vinyals, O.,
Warden, P., Wattenberg, M., Wicke, M., Yu, Y., and Zheng, X.: TensorFlow:
Large-Scale Machine Learning on Heterogeneous Systems, https://www.tensorflow.org/ (last access: 8 June 2021),
2015.
Beucler, T., Rasp, S., Pritchard, M., and Gentine, P.: Achieving
Conservation of Energy in Neural Network Emulators for Climate Modeling,
arXiv, https://arxiv.org/abs/1906.06622 (last
access: 17 June 2020), 2019.
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.
Brenowitz, N. D. and Bretherton, C. S.: Prognostic Validation of a Neural
Network Unified Physics Parameterization, Geophys. Res. Lett., 45,
6289–6298, https://doi.org/10.1029/2018GL078510, 2018.
Carter, W. P.: A detailed mechanism for the gas-phase atmospheric reactions
of organic compounds, Atmos. Environ., 24, 481–518,
https://doi.org/10.1016/0960-1686(90)90005-8, 1990.
Carter, W. P. L. and Heo, G.: Development of revised SAPRC aromatics
mechanisms, Atmos. Environ., 77, 404–414,
https://doi.org/10.1016/j.atmosenv.2013.05.021, 2013.
Chollet, F.: Keras, https://keras.io (last access: 8 June 2021), 2015.
Gardner, M. W. and Dorling, S. R.: Artificial neural networks (the
multilayer perceptron) – a review of applications in the atmospheric
sciences, Atmos. Environ., 32, 2627–2636, 1998.
Gery, M. W., Whitten, G. Z., Killus, J. P., and Dodge, M. C.: A
photochemical kinetics mechanism for urban and regional scale computer
modeling, J. Geophys. Res.-Atmos., 94, 12925–12956, 1989.
Harder, P., Watson-Parris, D., Strassel, D., Gauger, N., Stier, P., and
Keuper, J.: Emulating Aerosol Microphysics with Machine Learning, arXiv preprint arXiv:2109.10593, 2021.
Huang, L. and Topping, D.: JlBox v1.1: a Julia-based multi-phase atmospheric chemistry box model, Geosci. Model Dev., 14, 2187–2203, https://doi.org/10.5194/gmd-14-2187-2021, 2021.
Kashinath, K., Mustafa, M., Albert, A., Wu J. L., 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. A. Math. Phys Eng Sci., 379, 20200093, https://doi.org/10.1098/rsta.2020.0093, 2021.
Keller, C. A. and Evans, M. J.: Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10, Geosci. Model Dev., 12, 1209–1225, https://doi.org/10.5194/gmd-12-1209-2019, 2019.
Kelp, M. M., Tessum, C. W., and Marshall, J. D.: Orders-of-magnitude speedup
in atmospheric chemistry modeling through neural network-based emulation,
arXiv preprint arXiv:1808.03874, 2018.
Kelp, M. M., Jacob, D. J., Kutz, J. N., Marshall, J. D., and Tessum, C. W.:
Toward Stable, General Machine-Learned Models of the Atmospheric Chemical
System, J. Geophys. Res.-Atmos., 125, e2020JD032759, https://doi.org/10.1029/2020JD032759, 2020.
Krasnopolsky, V. M., Rabinovitz, M. S., Hou, Y. T., Lord, S. J.,
and Belochitski, A. A.: Accurate and Fast Neural Network Emulations of Model Radiation for the NCEP
Coupled Climate Forecast System: Climate Simulations and Seasonal Predictions, Mon. Weather Rev., 138,
1822–1842, https://doi.org/10.1175/2009MWR3149.1, 2010.
Manders, A. M. M., Builtjes, P. J. H., Curier, L., Denier van der Gon, H. A. C., Hendriks, C., Jonkers, S., Kranenburg, R., Kuenen, J. J. P., Segers, A. J., Timmermans, R. M. A., Visschedijk, A. J. H., Wichink Kruit, R. J., van Pul, W. A. J., Sauter, F. J., van der Swaluw, E., Swart, D. P. J., Douros, J., Eskes, H., van Meijgaard, E., van Ulft, B., van Velthoven, P., Banzhaf, S., Mues, A. C., Stern, R., Fu, G., Lu, S., Heemink, A., van Velzen, N., and Schaap, M.: Curriculum vitae of the LOTOS–EUROS (v2.0) chemistry transport model, Geosci. Model Dev., 10, 4145–4173, https://doi.org/10.5194/gmd-10-4145-2017, 2017.
McGovern, A., Lagerquist, R., Gagne, D. J., 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.
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel,
O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., and Vanderplas,
J.: Scikit-learn: Machine learning in Python, J. Mach. Learn. Res., 12,
2825–2830, 2011.
Potukuchi, S. and Wexler, A. S.: Predicting vapor pressures using neural
networks, Atmos. Environ., 31, 741–753, https://doi.org/10.1016/S1352-2310(96)00203-8, 1997.
Raissi, M., Perdikaris, P., and Karniadakis, G. E.: Physics-informed neural
networks: A deep learning framework for solving forward and inverse problems
involving nonlinear partial differential equations, J. Computat. Phys., 378,
686–707, https://doi.org/10.1016/j.jcp.2018.10.045, 2019.
Silva, S. J., Burrows, S. M., Evans, M. J., and Halappanavar, M.: A Graph
Theoretical Intercomparison of Atmospheric Chemical Mechanisms, Geophys.
Res. Lett., 48, e2020GL090481, https://doi.org/10.1029/2020GL090481, 2021a.
Silva, S. J., Ma, P.-L., Hardin, J. C., and Rothenberg, D.: Physically regularized machine learning emulators of aerosol activation, Geosci. Model Dev., 14, 3067–3077, https://doi.org/10.5194/gmd-14-3067-2021, 2021b.
Sturm, P. O.: Photochemical Box Model in Julia (0.2.0), Zenodo [data set], https://doi.org/10.5281/zenodo.5736487, 2021.
Sturm, P. O.: Python code for Sturm and Wexler (2022): Conservation laws in a neural network architecture (0.2.0), Zenodo [code], https://doi.org/10.5281/zenodo.6363763, 2022.
Sturm, P. O. and Wexler, A. S.: A mass- and energy-conserving framework for using machine learning to speed computations: a photochemistry example, Geosci. Model Dev., 13, 4435–4442, https://doi.org/10.5194/gmd-13-4435-2020, 2020.
Wexler, A. S. and Seinfeld, J. H.: Second-generation inorganic aerosol model,
Atmos. Environ., 25A, 2731–2748, 1991.
Yuval, J., O'Gorman, P. A., and Hill, C. N.: Use of neural networks for
stable, accurate and physically consistent parameterization of subgrid
atmospheric processes with good performance at reduced
precision, Geophys. Res. Lett., 48, e2020GL091363, https://doi.org/10.1029/2020gl091363, 2021.
Zaveri, R. A. and Peters, L. K.: A new lumped structure photochemical
mechanism for large-scale applications, J. Geophys. Res.-Atmos, 104,
30387–30415, 1999.
Zaveri, R. A., Easter, R. C., Fast, J. D., and Peters, L. K.: Model for
simulating aerosol interactions and chemistry (MOSAIC), J. Geophys.
Res.-Atmos., 113, D13204, 2008.
Zhao, W. L., Gentine, P., Reichstein, M., Zhang, Y., Zhou, S., Wen, Y., Lin,
C., Li, X., and Qiu, G. Y.: Physics-constrained machine learning of
evapotranspiration, Geophys. Res. Lett., 46, 14496–14507, 2019.
Short summary
Large air quality and climate models require vast amounts of computational power. Machine learning tools like neural networks can be used to make these models more efficient, with the downside that their results might not make physical sense or be easy to interpret. This work develops a physically interpretable neural network that obeys scientific laws like conservation of mass and models atmospheric composition more accurately than a traditional neural network.
Large air quality and climate models require vast amounts of computational power. Machine...
