Articles | Volume 15, issue 22
Model description paper
21 Nov 2022
Model description paper | 21 Nov 2022
Development of an LSTM broadcasting deep-learning framework for regional air pollution forecast improvement
Haochen Sun et al.
No articles found.
Yun Fat Lam, Chi Chiu Cheung, Xuguo Zhang, Joshua S. Fu, and Jimmy Chi Hung Fung
Atmos. Chem. Phys., 21, 12895–12908,Short summary
In recent years, air pollution forecasting has become an important municipal service of the government. In this study, a new spatial allocation method based on satellite remote sensing and GIS techniques was developed to address the spatial deficiency of industrial source emissions in China, providing a substantial improvement on NO2 and PM2.5 forecast for the Pearl River Delta/Greater Bay Area.
Lin Su and Jimmy C. H. Fung
Atmos. Chem. Phys., 18, 11529–11545,Short summary
By applying an updated version of WRF-Chem, we aim to investigate the full effects of dust, including direct, semi-direct, and indirect effects, on the regional weather system over East Asia during a dust-intensive period. This is the first study to document the full effects of dust during a typical dust-intensive period over East Asia by applying an online coupled regional numerical model.
Lin Su and Jimmy C. H. Fung
Atmos. Chem. Phys., 18, 8707–8725,Short summary
A new treatment was implemented into a microphysics scheme in WRF-Chem to quantify and evaluate the effect of dust on the ice nucleation process. The simulation of atmospheric IWC during a typical dust-intensive period over East Asia was substantially improved with the effect of dust considered in the model. In addition, sensitivity experiments were carried out to optimize the parameters used in the implemented treatment.
Related subject area
Atmospheric sciencesA local particle filter and its Gaussian mixture extension implemented with minor modifications to the LETKFA comprehensive evaluation of the use of Lagrangian particle dispersion models for inverse modeling of greenhouse gas emissionsImportance of different parameterization changes for the updated dust cycle modeling in the Community Atmosphere Model (version 6.1)Evaluation of the NAQFC driven by the NOAA Global Forecast System (version 16): comparison with the WRF-CMAQ during the summer 2019 FIREX-AQ campaignData assimilation for the Model for Prediction Across Scales – Atmosphere with the Joint Effort for Data assimilation Integration (JEDI-MPAS 1.0.0): EnVar implementation and evaluationDevelopment of a regional feature selection-based machine learning system (RFSML v1.0) for air pollution forecasting over ChinaA lumped species approach for the simulation of secondary organic aerosol production from intermediate-volatility organic compounds (IVOCs): application to road transport in PMCAMx-iv (v1.0)TrackMatcher – a tool for finding intercepts in tracks of geographical positionsRecovery of sparse urban greenhouse gas emissionsTropospheric transport and unresolved convection: numerical experiments with CLaMS 2.0/MESSyMUNICH v2.0: a street-network model coupled with SSH-aerosol (v1.2) for multi-pollutant modellingA preliminary evaluation of FY-4A visible radiance data assimilation by the WRF (ARW v4.1.1)/DART (Manhattan release v9.8.0)-RTTOV (v12.3) system for a tropical storm caseRepeatable high-resolution statistical downscaling through deep learningAtmospherically Relevant Chemistry and Aerosol box model – ARCA box (version 1.2)MultilayerPy (v1.0): a Python-based framework for building, running and optimising kinetic multi-layer models of aerosols and filmsIntroduction of the DISAMAR radiative transfer model: determining instrument specifications and analysing methods for atmospheric retrieval (version 4.1.5)Assessment of the data assimilation framework for the Rapid Refresh Forecast System v0.1 and impacts on forecasts of a convective storm case studyDownscaling atmospheric chemistry simulations with physically consistent deep learningA machine learning methodology for the generation of a parameterization of the hydroxyl radicalLarge-eddy simulations with ClimateMachine v0.2.0: a new open-source code for atmospheric simulations on GPUs and CPUsHybrid ensemble-variational data assimilation in ABC-DA within a tropical frameworkOpenIFS/AC: atmospheric chemistry and aerosol in OpenIFS 43r3Simulations of aerosol pH in China using WRF-Chem (v4.0): sensitivities of aerosol pH and its temporal variations during haze episodesA daily highest air temperature estimation method and spatial–temporal changes analysis of high temperature in China from 1979 to 2018TransClim (v1.0): a chemistry–climate response model for assessing the effect of mitigation strategies for road traffic on ozoneA description of the first open-source community release of MISTRA-v9.0: a 0D/1D atmospheric boundary layer chemistry modelIntegrated Methane Inversion (IMI 1.0): a user-friendly, cloud-based facility for inferring high-resolution methane emissions from TROPOMI satellite observationsComputationally efficient methods for large-scale atmospheric inverse modelingImproving the joint estimation of CO2 and surface carbon fluxes using a constrained ensemble Kalman filter in COLA (v1.0)Evaluation of high-resolution predictions of fine particulate matter and its composition in an urban area using PMCAMx-v2.0RAP-Net: Region Attention Predictive Network for precipitation nowcastingEffects of point source emission heights in WRF–STILT: a step towards exploiting nocturnal observations in modelsuDALES 1.0: a large-eddy simulation model for urban environmentsDevelopment and evaluation of the Aerosol Forecast Member in the National Center for Environment Prediction (NCEP)'s Global Ensemble Forecast System (GEFS-Aerosols v1)Assimilation of GPM-retrieved ocean surface meteorology data for two snowstorm events during ICE-POP 2018A multi-pollutant and multi-sectorial approach to screening the consistency of emission inventoriesEvaluation of a forest parameterization to improve boundary layer flow simulations over complex terrain. A case study using WRF-LES V4.0.1Transporting CRM Variance in a Multiscale Modelling FrameworkStratospheric Nudging And Predictable Surface Impacts (SNAPSI): a protocol for investigating the role of stratospheric polar vortex disturbances in subseasonal to seasonal forecastsLightning Assimilation in the Weather Research and Forecasting (WRF) Model Version 4.1.1: Technique Updates and Assessment of the Applications from Regional to Hemispheric ScalesThe Mission Support System (MSS v7.0.0) and its use in planning for the SouthTRAC aircraft campaignVariational inverse modeling within the Community Inversion Framework v1.1 to assimilate δ13C(CH4) and CH4: a case study with model LMDz-SACSThe Comprehensive Automobile Research System (CARS) – a Python-based automobile emissions inventory modelValidation of turbulent heat transfer models against eddy covariance flux measurements over a seasonally ice-covered lakeRegional evaluation of the performance of the global CAMS chemical modeling system over the United States (IFS cycle 47r1)Order of magnitude wall time improvement of variational methane inversions by physical parallelization: a demonstration using TM5-4DVARSimulated microphysical properties of winter storms from bulk-type microphysics schemes and their evaluation in the Weather Research and Forecasting (v4.1.3) model during the ICE-POP 2018 field campaignGENerator of reduced Organic Aerosol mechanism (GENOA v1.0): An automatic generation tool of semi-explicit mechanismsA novel method for objective identification of 3-D potential vorticity anomaliesMultiple same-level and telescoping nesting in GFDL's dynamical core
Shunji Kotsuki, Takemasa Miyoshi, Keiichi Kondo, and Roland Potthast
Geosci. Model Dev., 15, 8325–8348,Short summary
Data assimilation plays an important part in numerical weather prediction (NWP) in terms of combining forecasted states and observations. While data assimilation methods in NWP usually assume the Gaussian error distribution, some variables in the atmosphere, such as precipitation, are known to have non-Gaussian error statistics. This study extended a widely used ensemble data assimilation algorithm to enable the assimilation of more non-Gaussian observations.
Martin Vojta, Andreas Plach, Rona L. Thompson, and Andreas Stohl
Geosci. Model Dev., 15, 8295–8323,Short summary
In light of recent global warming, we aim to improve methods for modeling greenhouse gas emissions in order to support the successful implementation of the Paris Agreement. In this study, we investigate certain aspects of a Bayesian inversion method that uses computer simulations and atmospheric observations to improve estimates of greenhouse gas emissions. We explore method limitations, discuss problems, and suggest improvements.
Longlei Li, Natalie M. Mahowald, Jasper F. Kok, Xiaohong Liu, Mingxuan Wu, Danny M. Leung, Douglas S. Hamilton, Louisa K. Emmons, Yue Huang, Neil Sexton, Jun Meng, and Jessica Wan
Geosci. Model Dev., 15, 8181–8219,Short summary
This study advances mineral dust parameterizations in the Community Atmospheric Model (CAM; version 6.1). Efforts include 1) incorporating a more physically based dust emission scheme; 2) updating the dry deposition scheme; and 3) revising the gravitational settling velocity to account for dust asphericity. Substantial improvements achieved with these updates can help accurately quantify dust–climate interactions using CAM, such as the dust-radiation and dust–cloud interactions.
Youhua Tang, Patrick C. Campbell, Pius Lee, Rick Saylor, Fanglin Yang, Barry Baker, Daniel Tong, Ariel Stein, Jianping Huang, Ho-Chun Huang, Li Pan, Jeff McQueen, Ivanka Stajner, Jose Tirado-Delgado, Youngsun Jung, Melissa Yang, Ilann Bourgeois, Jeff Peischl, Tom Ryerson, Donald Blake, Joshua Schwarz, Jose-Luis Jimenez, James Crawford, Glenn Diskin, Richard Moore, Johnathan Hair, Greg Huey, Andrew Rollins, Jack Dibb, and Xiaoyang Zhang
Geosci. Model Dev., 15, 7977–7999,Short summary
This paper compares two meteorological datasets for driving a regional air quality model: a regional meteorological model using WRF (WRF-CMAQ) and direct interpolation from an operational global model (GFS-CMAQ). In the comparison with surface measurements and aircraft data in summer 2019, these two methods show mixed performance depending on the corresponding meteorological settings. Direct interpolation is found to be a viable method to drive air quality models.
Zhiquan Liu, Chris Snyder, Jonathan J. Guerrette, Byoung-Joo Jung, Junmei Ban, Steven Vahl, Yali Wu, Yannick Trémolet, Thomas Auligné, Benjamin Ménétrier, Anna Shlyaeva, Stephen Herbener, Emily Liu, Daniel Holdaway, and Benjamin T. Johnson
Geosci. Model Dev., 15, 7859–7878,Short summary
JEDI-MPAS 1.0.0, a new data assimilation (DA) system for the MPAS model, was publicly released for community use. This article describes JEDI-MPAS's implementation of the ensemble–variational DA technique and demonstrates its robustness and credible performance by incrementally adding three types of microwave radiances (clear-sky AMSU-A, all-sky AMSU-A, clear-sky MHS) to a non-radiance DA experiment. We intend to periodically release new and improved versions of JEDI-MPAS in upcoming years.
Li Fang, Jianbing Jin, Arjo Segers, Hai Xiang Lin, Mijie Pang, Cong Xiao, Tuo Deng, and Hong Liao
Geosci. Model Dev., 15, 7791–7807,Short summary
This study proposes a regional feature selection-based machine learning system to predict short-term air quality in China. The system has a tool that can figure out the importance of input data for better prediction. It provides large-scale air quality prediction that exhibits improved interpretability, fewer training costs, and higher accuracy compared with a standard machine learning system. It can act as an early warning for citizens and reduce exposure to PM2.5 and other air pollutants.
Stella E. I. Manavi and Spyros N. Pandis
Geosci. Model Dev., 15, 7731–7749,Short summary
The paper describes the first step towards the development of a simulation framework for the chemistry and secondary organic aerosol production of intermediate-volatility organic compounds (IVOCs). These compounds can be a significant source of organic particulate matter. Our approach treats IVOCs as lumped compounds that retain their chemical characteristics. Estimated IVOC emissions from road transport were a factor of 8 higher than emissions used in previous applications.
Peter Bräuer and Matthias Tesche
Geosci. Model Dev., 15, 7557–7572,Short summary
This paper presents a tool for (i) finding temporally and spatially resolved intersections between two- or three-dimensional geographical tracks (trajectories) and (ii) extracting of data in the vicinity of intersections to achieve the optimal combination of various data sets.
Benjamin Zanger, Jia Chen, Man Sun, and Florian Dietrich
Geosci. Model Dev., 15, 7533–7556,Short summary
Gaussian priors (GPs) used in least squares inversion do not reflect the true distributions of greenhouse gas emissions well. A method that does not rely on GPs is sparse reconstruction (SR). We show that necessary conditions for SR are satisfied for cities and that the application of a wavelet transform can further enhance sparsity. We apply the theory of compressed sensing to SR. Our results show that SR needs fewer measurements and is superior for assessing unknown emitters compared to GPs.
Paul Konopka, Mengchu Tao, Marc von Hobe, Lars Hoffmann, Corinna Kloss, Fabrizio Ravegnani, C. Michael Volk, Valentin Lauther, Andreas Zahn, Peter Hoor, and Felix Ploeger
Geosci. Model Dev., 15, 7471–7487,Short summary
Pure trajectory-based transport models driven by meteorology derived from reanalysis products (ERA5) take into account only the resolved, advective part of transport. That means neither mixing processes nor unresolved subgrid-scale advective processes like convection are included. The Chemical Lagrangian Model of the Stratosphere (CLaMS) includes these processes. We show that isentropic mixing dominates unresolved transport. The second most important transport process is unresolved convection.
Youngseob Kim, Lya Lugon, Alice Maison, Thibaud Sarica, Yelva Roustan, Myrto Valari, Yang Zhang, Michel André, and Karine Sartelet
Geosci. Model Dev., 15, 7371–7396,Short summary
This paper presents the latest version of the street-network model MUNICH, v2.0. The description of MUNICH v1.0, which models gas-phase pollutants in a street network, was published in GMD in 2018. Since then, major modifications have been made to MUNICH. The comprehensive aerosol model SSH-aerosol is now coupled to MUNICH to simulate primary and secondary aerosol concentrations. New parameterisations have also been introduced. Test cases are defined to illustrate the new model functionalities.
Yongbo Zhou, Yubao Liu, Zhaoyang Huo, and Yang Li
Geosci. Model Dev., 15, 7397–7420,Short summary
The study evaluates the performance of the Data Assimilation Research Testbed (DART), equipped with the recently added forward operator Radiative Transfer for TOVS (RTTOV), in assimilating FY-4A visible images into the Weather Research and Forecasting (WRF) model. The ability of the WRF-DART/RTTOV system to improve the forecasting skills for a tropical storm over East Asia and the Western Pacific is demonstrated in an Observing System Simulation Experiment framework.
Dánnell Quesada-Chacón, Klemens Barfus, and Christian Bernhofer
Geosci. Model Dev., 15, 7353–7370,Short summary
We improved the performance of past perfect prognosis statistical downscaling methods while achieving full model repeatability with GPU-calculated deep learning models using the TensorFlow, climate4R, and VALUE frameworks. We employed the ERA5 reanalysis as predictors and ReKIS (eastern Ore Mountains, Germany, 1 km resolution) as precipitation predictand, while incorporating modern deep learning architectures. The achieved repeatability is key to accomplish further milestones with deep learning.
Petri Clusius, Carlton Xavier, Lukas Pichelstorfer, Putian Zhou, Tinja Olenius, Pontus Roldin, and Michael Boy
Geosci. Model Dev., 15, 7257–7286,Short summary
Atmospheric chemistry and aerosol processes form a dynamic and sensitively balanced system, and solving problems regarding air quality or climate requires detailed modelling and coupling of the processes. The models involved are often very complex to use. We have addressed this problem with the new ARCA box model. It puts much of the current knowledge of the nano- and microscale aerosol dynamics and chemistry into usable software and has the potential to become a valuable tool in the community.
Adam Milsom, Amy Lees, Adam M. Squires, and Christian Pfrang
Geosci. Model Dev., 15, 7139–7151,Short summary
MultilayerPy is a Python-based framework facilitating the creation, running and optimisation of state-of-the-art kinetic multi-layer models of aerosol and film processes. Models can be fit to data with local and global optimisation algorithms along with a statistical sampling algorithm, which quantifies the uncertainty in optimised model parameters. This “modelling study in a box” enables more reproducible and reliable results, with model code and outputs produced in a human-readable way.
Johan F. de Haan, Ping Wang, Maarten Sneep, J. Pepijn Veefkind, and Piet Stammes
Geosci. Model Dev., 15, 7031–7050,Short summary
We present an overview of the DISAMAR radiative transfer code, highlighting the novel semi-analytical derivatives for the doubling–adding formulae and the new DISMAS technique for weak absorbers. DISAMAR includes forward simulations and retrievals for satellite spectral measurements from 270 to 2400 nm to determine instrument specifications for passive remote sensing. It has been used in various Sentinel-4/5P/5 projects and in the TROPOMI aerosol layer height and ozone profile products.
Ivette H. Banos, Will D. Mayfield, Guoqing Ge, Luiz F. Sapucci, Jacob R. Carley, and Louisa Nance
Geosci. Model Dev., 15, 6891–6917,Short summary
A prototype data assimilation system for NOAA’s next-generation rapidly updated, convection-allowing forecast system, or Rapid Refresh Forecast System (RRFS) v0.1, is tested and evaluated. The impact of using data assimilation with a convective storm case study is examined. Although the convection in RRFS tends to be overestimated in intensity and underestimated in extent, the use of data assimilation proves to be crucial to improve short-term forecasts of storms and precipitation.
Andrew Geiss, Sam J. Silva, and Joseph C. Hardin
Geosci. Model Dev., 15, 6677–6694,Short summary
This work demonstrates the use of modern machine learning techniques to enhance the resolution of atmospheric chemistry simulations. We evaluate the schemes for an 8 x 10 increase in resolution and find that they perform substantially better than conventional methods. Methods are introduced to target machine learning methods towards this type of problem, most notably by ensuring they do not break known physical constraints.
Daniel C. Anderson, Melanie B. Follette-Cook, Sarah A. Strode, Julie M. Nicely, Junhua Liu, Peter D. Ivatt, and Bryan N. Duncan
Geosci. Model Dev., 15, 6341–6358,Short summary
The hydroxyl radical (OH) is the most important chemical in the atmosphere for removing certain pollutants, including methane, the second-most-important greenhouse gas. We present a methodology to create an easily modifiable parameterization that can calculate OH concentrations in a computationally efficient way. The parameterization, which predicts OH within 5 %, can be integrated into larger climate models to allow for calculation of the interactions between OH, methane, and other chemicals.
Akshay Sridhar, Yassine Tissaoui, Simone Marras, Zhaoyi Shen, Charles Kawczynski, Simon Byrne, Kiran Pamnany, Maciej Waruszewski, Thomas H. Gibson, Jeremy E. Kozdon, Valentin Churavy, Lucas C. Wilcox, Francis X. Giraldo, and Tapio Schneider
Geosci. Model Dev., 15, 6259–6284,Short summary
ClimateMachine is a new open-source Julia-language atmospheric modeling code. We describe its limited-area configuration and the model equations, and we demonstrate applicability through benchmark problems, including atmospheric flow in the shallow cumulus regime. We show that the discontinuous Galerkin numerics and model equations allow global conservation of key variables (up to sources and sinks). We assess CPU strong scaling and GPU weak scaling to show its suitability for large simulations.
Joshua Chun Kwang Lee, Javier Amezcua, and Ross Noel Bannister
Geosci. Model Dev., 15, 6197–6219,Short summary
In this article, we implement a novel data assimilation method for the ABC–DA system which combines traditional data assimilation approaches in a hybrid approach. We document the technical development and test the hybrid approach in idealised experiments within a tropical framework of the ABC–DA system. Our findings indicate that the hybrid approach outperforms individual traditional approaches. Its potential benefits have been highlighted and should be explored further within this framework.
Vincent Huijnen, Philippe Le Sager, Marcus O. Köhler, Glenn Carver, Samuel Rémy, Johannes Flemming, Simon Chabrillat, Quentin Errera, and Twan van Noije
Geosci. Model Dev., 15, 6221–6241,Short summary
We report on the first implementation of atmospheric chemistry and aerosol as part of the OpenIFS model, based on the CAMS global model. We give an overview of the model and evaluate two reference model configurations, with and without the stratospheric chemistry extension, against a variety of observational datasets. This OpenIFS version with atmospheric composition components is open to the scientific user community under a standard OpenIFS license.
Xueyin Ruan, Chun Zhao, Rahul A. Zaveri, Pengzhen He, Xinming Wang, Jingyuan Shao, and Lei Geng
Geosci. Model Dev., 15, 6143–6164,Short summary
Accurate prediction of aerosol pH in chemical transport models is essential to aerosol modeling. This study examines the performance of the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) on aerosol pH predictions and the sensitivities to emissions of nonvolatile cations and NH3, aerosol-phase state assumption, and heterogeneous sulfate production. Temporal evolution of aerosol pH during haze cycles in Beijing and the driving factors are also presented and discussed.
Ping Wang, Kebiao Mao, Fei Meng, Zhihao Qin, Shu Fang, and Sayed M. Bateni
Geosci. Model Dev., 15, 6059–6083,Short summary
In order to obtain the key parameters of high-temperature spatial–temporal variation analysis, this study proposed a daily highest air temperature (Tmax) estimation frame to build a Tmax dataset in China from 1979 to 2018. We found that the annual and seasonal mean Tmax in most areas of China showed an increasing trend. The abnormal temperature changes mainly occurred in El Nin~o years or La Nin~a years. IOBW had a stronger influence on China's warming events than other factors.
Vanessa Simone Rieger and Volker Grewe
Geosci. Model Dev., 15, 5883–5903,Short summary
Road traffic emissions of nitrogen oxides, volatile organic compounds and carbon monoxide produce ozone in the troposphere and thus influence Earth's climate. To assess the ozone response to a broad range of mitigation strategies for road traffic, we developed a new chemistry–climate response model called TransClim. It is based on lookup tables containing climate–response relations and thus is able to quickly determine the climate response of a mitigation option.
Josué Bock, Jan Kaiser, Max Thomas, Andreas Bott, and Roland von Glasow
Geosci. Model Dev., 15, 5807–5828,Short summary
MISTRA-v9.0 is an atmospheric boundary layer chemistry model. The model includes a detailed particle description with regards to the microphysics, gas–particle interactions, and liquid phase chemistry within particles. Version 9.0 is the first release of MISTRA as an open-source community model. This paper presents a thorough description of the model characteristics and components. We show some examples of simulations reproducing previous studies with MISTRA with good consistency.
Daniel J. Varon, Daniel J. Jacob, Melissa Sulprizio, Lucas A. Estrada, William B. Downs, Lu Shen, Sarah E. Hancock, Hannah Nesser, Zhen Qu, Elise Penn, Zichong Chen, Xiao Lu, Alba Lorente, Ashutosh Tewari, and Cynthia A. Randles
Geosci. Model Dev., 15, 5787–5805,Short summary
Reducing atmospheric methane emissions is critical to slow near-term climate change. Globally surveying satellite instruments like the TROPOspheric Monitoring Instrument (TROPOMI) have unique capabilities for monitoring atmospheric methane around the world. Here we present a user-friendly cloud-computing tool that enables researchers and stakeholders to quantify methane emissions across user-selected regions of interest using TROPOMI satellite observations.
Taewon Cho, Julianne Chung, Scot M. Miller, and Arvind K. Saibaba
Geosci. Model Dev., 15, 5547–5565,Short summary
Atmospheric inverse modeling describes the process of estimating greenhouse gas fluxes or air pollution emissions at the Earth's surface using observations of these gases collected in the atmosphere. The launch of new satellites, the expansion of surface observation networks, and a desire for more detailed maps of surface fluxes have yielded numerous computational and statistical challenges. This article describes computationally efficient methods for large-scale atmospheric inverse modeling.
Zhiqiang Liu, Ning Zeng, Yun Liu, Eugenia Kalnay, Ghassem Asrar, Bo Wu, Qixiang Cai, Di Liu, and Pengfei Han
Geosci. Model Dev., 15, 5511–5528,Short summary
We described the application of a constrained ensemble Kalman filter (CEnKF) in a joint CO2 and surface carbon fluxes estimation study. By assimilating the pseudo-surface and OCO-2 observations, the annual global flux estimation is significantly biased without mass conservation. With the additional CEnKF process, the CO2 mass is strictly constrained, and the estimation of annual fluxes is significantly improved.
Brian T. Dinkelacker, Pablo Garcia Rivera, Ioannis Kioutsioukis, Peter J. Adams, and Spyros N. Pandis
Geosci. Model Dev. Discuss.,
Preprint under review for GMDShort summary
The performance of a chemical transport model in reproducing PM2.5 concentrations and composition was evaluated at the finest scale using measurements from regulatory sites as well as a network of low-cost monitors. Total PM2.5 mass is reproduced well by the model during the winter when compared to regulatory measurements, but in the summer PM2.5 is underpredicted mainly due to difficulties in reproducing the regional secondary organic aerosol levels.
Zheng Zhang, Chuyao Luo, Shanshan Feng, Rui Ye, Yunming Ye, and Xutao Li
Geosci. Model Dev., 15, 5407–5419,Short summary
In this paper, we develop a model to predict radar echo sequences and apply it in the precipitation nowcasting field. Different from existing models, we propose two new attention modules. By introducing them, the performance of RAP-Net outperforms other models, especially in those regions with moderate and heavy rainfall. Considering that these regions cause more threats to human activities, the research in our work is significant for preventing natural disasters caused by heavy rainfall.
Fabian Maier, Christoph Gerbig, Ingeborg Levin, Ingrid Super, Julia Marshall, and Samuel Hammer
Geosci. Model Dev., 15, 5391–5406,Short summary
We show that the default representation of point source emissions in WRF–STILT leads to large overestimations when modelling fossil fuel CO2 concentrations for a 30 m high observation site during stable atmospheric conditions. We therefore introduce a novel point source modelling approach in WRF-STILT that takes into account their effective emission heights and results in a much better agreement with observations.
Ivo Suter, Tom Grylls, Birgit S. Sützl, Sam O. Owens, Chris E. Wilson, and Maarten van Reeuwijk
Geosci. Model Dev., 15, 5309–5335,Short summary
Cities are increasingly moving to the fore of climate and air quality research due to their central role in the population’s health and well-being, while suitable models remain scarce. This article describes the development of a new urban LES model, which allows examining the effects of various processes, infrastructure and vegetation on the local climate and air quality. Possible applications are demonstrated and a comparison to an experiment is shown.
Li Zhang, Raffaele Montuoro, Stuart A. McKeen, Barry Baker, Partha S. Bhattacharjee, Georg A. Grell, Judy Henderson, Li Pan, Gregory J. Frost, Jeff McQueen, Rick Saylor, Haiqin Li, Ravan Ahmadov, Jun Wang, Ivanka Stajner, Shobha Kondragunta, Xiaoyang Zhang, and Fangjun Li
Geosci. Model Dev., 15, 5337–5369,Short summary
The NOAA’s air quality predictions contribute to protecting lives and health in the US, which requires sustainable development and improvement of forecast systems. GEFS-Aerosols v1 has been developed in a collaboration between the NOAA research laboratories for operational forecast since September 2020 in the NCEP. The predictions demonstrate substantial improvements for both composition and variability of aerosol distributions over those from the former operational system.
Xuanli Li, Jason B. Roberts, Jayanthi Srikishen, Jonathan L. Case, Walter A. Petersen, Gyuwon Lee, and Christopher R. Hain
Geosci. Model Dev., 15, 5287–5308,Short summary
This research assimilated the Global Precipitation Measurement (GPM) satellite-retrieved ocean surface meteorology data into the Weather Research and Forecasting (WRF) model with the Gridpoint Statistical Interpolation (GSI) system. This was for two snowstorms during the International Collaborative Experiments for PyeongChang 2018 Olympic and Paralympic Winter Games' (ICE-POP 2018) field experiments. The results indicated a positive impact of the data for short-term forecasts for heavy snowfall.
Philippe Thunis, Alain Clappier, Enrico Pisoni, Bertrand Bessagnet, Jeroen Kuenen, Marc Guevara, and Susana Lopez-Aparicio
Geosci. Model Dev., 15, 5271–5286,Short summary
In this work, we propose a screening method to improve the quality of emission inventories, which are responsible for large uncertainties in air-quality modeling. The first step of screening consists of keeping only emission contributions that are relevant enough. In a second step, the method identifies large differences that provide evidence of methodological divergence or errors. We used the approach to compare two versions of the CAMS-REG European-scale inventory over 150 European cities.
Julian Quimbayo-Duarte, Johannes Wagner, Norman Wildmann, Thomas Gerz, and Juerg Schmidli
Geosci. Model Dev., 15, 5195–5209,Short summary
The ultimate objective of this model evaluation is to improve boundary layer flow representation over complex terrain. The numerical model is tested against observations retrieved during the Perdigão 2017 field campaign (moderate complex terrain). We observed that the inclusion of a forest parameterization in the numerical model significantly improves the representation of the wind field in the atmospheric boundary layer.
Walter Hannah and Kyle Pressel
A multiscale modelling framework couples two models of the atmosphere that each cover different scale ranges. Traditionally, fluctuations in the small-scale model are not transported by the flow on the large-scale model grid, but this is hypothesized to be responsible for a persistent, unphysical checkerboard pattern. A method is presented to facilitate the transport of these small-scale fluctuations, analogous to how small-scale clouds and turbulence are transported in the real atmosphere.
Peter Hitchcock, Amy Butler, Andrew Charlton-Perez, Chaim I. Garfinkel, Tim Stockdale, James Anstey, Dann Mitchell, Daniela I. V. Domeisen, Tongwen Wu, Yixiong Lu, Daniele Mastrangelo, Piero Malguzzi, Hai Lin, Ryan Muncaster, Bill Merryfield, Michael Sigmond, Baoqiang Xiang, Liwei Jia, Yu-Kyung Hyun, Jiyoung Oh, Damien Specq, Isla R. Simpson, Jadwiga H. Richter, Cory Barton, Jeff Knight, Eun-Pa Lim, and Harry Hendon
Geosci. Model Dev., 15, 5073–5092,Short summary
This paper describes an experimental protocol focused on sudden stratospheric warmings to be carried out by subseasonal forecast modeling centers. These will allow for inter-model comparisons of these major disruptions to the stratospheric polar vortex and their impacts on the near-surface flow. The protocol will lead to new insights into the contribution of the stratosphere to subseasonal forecast skill and new approaches to the dynamical attribution of extreme events.
Daiwen Kang, Nicholas Heath, Robert Gilliam, Tanya Spero, and Jonathan Pleim
Lightning assimilation implemented in the WRF model's Kain-Fritsch convective scheme is updated and applied to simulations from regional to hemispheric scales using observed lightning flashes from ground-based lightning detection networks. Different user-toggled options associated with the Kain-Fritsch scheme on simulations with and without lightning assimilation are assessed. The model's performance is improved significantly by lightning assimilation, but it is sensitive to various factors.
Reimar Bauer, Jens-Uwe Grooß, Jörn Ungermann, May Bär, Markus Geldenhuys, and Lars Hoffmann
Geosci. Model Dev. Discuss.,
Revised manuscript accepted for GMDShort summary
The Mission Support System (MSS) is an open source software package that has been used for planning flight tracks of scientific aircraft in multiple measurements campaigns during the last decade. Here, we describe the MSS software and its use during the SouthTRAC measurement campaign in 2019. As example for how the MSS software is used in conjunction with many data sets, we describe the planning of a single flight probing orographic gravity waves propagating up into the lower mesosphere.
Joël Thanwerdas, Marielle Saunois, Antoine Berchet, Isabelle Pison, Bruce H. Vaughn, Sylvia Englund Michel, and Philippe Bousquet
Geosci. Model Dev., 15, 4831–4851,Short summary
Estimating CH4 sources by exploiting observations within an inverse modeling framework is a powerful approach. Here, a new system designed to assimilate δ13C(CH4) observations together with CH4 observations is presented. By optimizing both the emissions and associated source signatures of multiple emission categories, this new system can efficiently differentiate the co-located emission categories and provide estimates of CH4 sources that are consistent with isotopic data.
Bok H. Baek, Rizzieri Pedruzzi, Minwoo Park, Chi-Tsan Wang, Younha Kim, Chul-Han Song, and Jung-Hun Woo
Geosci. Model Dev., 15, 4757–4781,Short summary
The Comprehensive Automobile Research System (CARS) is an open-source Python-based automobile emissions inventory model designed to efficiently estimate high-quality emissions. The CARS is designed to utilize the local vehicle activity database, such as vehicle travel distance, road-link-level network information, and vehicle-specific average speed by road type, to generate a temporally and spatially enhanced inventory for policymakers, stakeholders, and the air quality modeling community.
Joonatan Ala-Könni, Kukka-Maaria Kohonen, Matti Leppäranta, and Ivan Mammarella
Geosci. Model Dev., 15, 4739–4755,Short summary
Properties of seasonally ice-covered lakes are not currently sufficiently included in global climate models. To fill this gap, this study evaluates three models that could be used to quantify the amount of heat that moves from and into the lake by the air above it and through evaporation of the ice cover. The results show that the complex nature of the surrounding environment as well as difficulties in accurately measuring the surface temperature of ice introduce errors to these models.
Jason E. Williams, Vincent Huijnen, Idir Bouarar, Mehdi Meziane, Timo Schreurs, Sophie Pelletier, Virginie Marécal, Beatrice Josse, and Johannes Flemming
Geosci. Model Dev., 15, 4657–4687,Short summary
The global CAMS air quality model is used for providing tropospheric ozone information to end users. This paper updates the chemical mechanism employed (CBA) and compares it against two other mechanisms (MOCAGE, MOZART) and a multi-decadal dataset based on a previous version of CBA. We perform extensive validation for the US using multiple surface and aircraft datasets, providing an assessment of biases and the extent of correlation across different seasons during 2014.
Sudhanshu Pandey, Sander Houweling, and Arjo Segers
Geosci. Model Dev., 15, 4555–4567,Short summary
Inversions are used to calculate methane emissions using atmospheric mole-fraction measurements. Multidecadal inversions are needed to extract information from the long measurement records of methane. However, multidecadal inversion computations can take months to finish. Here, we demonstrate an order of magnitude improvement in wall clock time for an iterative multidecadal inversion by physical parallelization of chemical transport model.
Jeong-Su Ko, Kyo-Sun Sunny Lim, Kwonil Kim, Gyuwon Lee, Gregory Thompson, and Alexis Berne
Geosci. Model Dev., 15, 4529–4553,Short summary
This study evaluates the performance of the four microphysics parameterizations, the WDM6, WDM7, Thompson, and Morrison schemes, in simulating snowfall events during the ICE-POP 2018 field campaign. Eight snowfall events are selected and classified into three categories (cold-low, warm-low, and air–sea interaction cases). The evaluation focuses on the simulated hydrometeors, microphysics budgets, wind fields, and precipitation using the measurement data.
Zhizhao Wang, Florian Couvidat, and Karine Sartelet
3D air-quality models need to represent the complexity of SOA formation with a manageable computational cost. Therefore, we developed GENOA v1.0, an algorithm to generate semi-explicit SOA mechanisms preserving the accuracy of the explicit VOC chemical mechanisms on SOA formation. When applying to the MCM degradation scheme of beta-caryophyllene, GENOA managed to reduce the mechanism from 1626 to 23 reactions and from 579 to 15 species with an average error lower than 3 %.
Christoph Fischer, Andreas H. Fink, Elmar Schömer, Roderick van der Linden, Michael Maier-Gerber, Marc Rautenhaus, and Michael Riemer
Geosci. Model Dev., 15, 4447–4468,Short summary
Potential vorticity (PV) analysis plays a central role in studying atmospheric dynamics. For example, anomalies in the PV field near the tropopause are linked to extreme weather events. In this study, an objective strategy to identify these anomalies is presented and evaluated. As a novel concept, it can be applied to three-dimensional (3-D) data sets. Supported by 3-D visualizations, we illustrate advantages of this new analysis over existing studies along a case study.
Joseph Mouallem, Lucas Harris, and Rusty Benson
Geosci. Model Dev., 15, 4355–4371,Short summary
The single-nest capability in GFDL's dynamical core, FV3, is upgraded to support multiple same-level and telescoping nests. Grid nesting adds a refined grid over an area of interest to better resolve small-scale flow features necessary to accurately predict special weather events such as severe storms and hurricanes. This work allows concurrent execution of multiple same-level and telescoping multi-level nested grids in both global and regional setups.
Ayturan, Y. A., Ayturan, Z. C., and Altun, H. O.: Air pollution modelling with deep learning: a review, International Journal of Environmental Pollution and Environmental Modelling, 1, 58–62, 2018.
Bi, J., Knowland, K. E., Keller, C. A., and Liu, Y.: Combining Machine Learning and Numerical Simulation for High-Resolution PM2.5 Concentration Forecast, Environ. Sci. Technol., 56, 1544–1556, 2022.
Bui, T.-C., Le, V.-D., and Cha, S.-K.: A deep learning approach for forecasting air pollution in South Korea using LSTM, arXiv [preprint], https://doi.org/10.48550/arXiv.1804.07891, 2018.
Fan, C., Li, Y., Guang, J., Li, Z., Elnashar, A., Allam, M., and de Leeuw, G.: The impact of the control measures during the COVID-19 outbreak on air pollution in China, Remote Sensing, 12, 1613, https://doi.org/10.3390/rs12101613, 2020.
Gilliam, R. C., Hogrefe, C., Godowitch, J. M., Napelenok, S., Mathur, R., and Rao, S. T.: Impact of inherent meteorology uncertainty on air quality model predictions, J. Geophys. Res.-Atmos., 120, 12259–12280, 2015.
Greff, K., Srivastava, R., Koutník, J., Steunebrink, B., and Schmidhuber, J.: LSTM: A search space odyssey, IEEE Transactions on Neural Networks Learning Systems, https://doi.org/10.1109/TNNLS.2016.2582924, 2017.
Hähnel, P., Mareček, J., Monteil, J., and O'Donncha, F.: Using deep learning to extend the range of air pollution monitoring and forecasting, J. Comput. Phys., 408, 109278, 2020.
Han, J., Liu, H., Zhu, H., Xiong, H., and Dou, D.: Joint air quality and weather prediction based on multi-adversarial spatiotemporal networks, Proceedings of the AAAI Conference on Artificial Intelligence, 2–9 February 2021, virtual conference, 4081–4089, https://doi.org/10.48550/arXiv.2012.15037, 2021.
Hochreiter, S. and Schmidhuber, J.: Long short-term memory, Neural Comput., 9, 1735–1780, 1997.
Holnicki, P. and Nahorski, Z.: Emission data uncertainty in urban air quality modeling–case study, Environ. Model. Assess., 20, 583–597, 2015.
Huang, C.-J. and Kuo, P.-H.: A deep CNN-LSTM model for particulate matter (PM2.5) forecasting in smart cities, Sensors, 18, 2220, https://doi.org/10.3390/s18072220, 2018.
Ioffe, S. and Szegedy, C.: Batch normalization: Accelerating deep network training by reducing internal covariate shift, International conference on machine learning, 6–11 July 2015, Lille, France, 448–456, https://doi.org/10.48550/arXiv.1502.03167, 2015.
Janarthanan, R., Partheeban, P., Somasundaram, K., and Elamparithi, P. N.: A deep learning approach for prediction of air quality index in a metropolitan city, Sustain. Cities Soc., 67, 102720, 2021.
Karim, F., Majumdar, S., Darabi, H., and Chen, S.: LSTM fully convolutional networks for time series classification, IEEE access, 6, 1662–1669, 2017.
Kim, H. S., Park, I., Song, C. H., Lee, K., Yun, J. W., Kim, H. K., Jeon, M., Lee, J., and Han, K. M.: Development of a daily PM10 and PM2.5 prediction system using a deep long short-term memory neural network model, Atmos. Chem. Phys., 19, 12935–12951, https://doi.org/10.5194/acp-19-12935-2019, 2019.
Kingma, D. P. and Ba, J.: Adam: A method for stochastic optimization, arXiv [preprint],https://doi.org/10.48550/arXiv.1412.6980, 2014.
Lee, K., Yu, J., Lee, S., Park, M., Hong, H., Park, S. Y., Choi, M., Kim, J., Kim, Y., Woo, J.-H., Kim, S.-W., and Song, C. H.: Development of Korean Air Quality Prediction System version 1 (KAQPS v1) with focuses on practical issues, Geosci. Model Dev., 13, 1055–1073, https://doi.org/10.5194/gmd-13-1055-2020, 2020.
Li, T., Hua, M., and Wu, X.: A hybrid CNN-LSTM model for forecasting particulate matter (PM2.5), Ieee Access, 8, 26933–26940, 2020.
Lu, H., Xie, M., Liu, X., Liu, B., Jiang, M., Gao, Y., and Zhao, X.: Adjusting prediction of ozone concentration based on CMAQ model and machine learning methods in Sichuan-Chongqing region, China, Atmos. Pollut. Res., 12, 101066, https://doi.org/10.1016/j.apr.2021.101066, 2021.
Lu, X., Fung, J. C. H., and Wu, D.: Modeling wet deposition of acid substances over the PRD region in China, Atmos. Environ., 122, 819–828, 2015.
Lu, X., Wang, Y., Li, J., Shen, L., and Fung, J. C.: Evidence of heterogeneous HONO formation from aerosols and the regional photochemical impact of this HONO source, Environ. Res. Lett., 13, 114002, https://doi.org/10.1088/1748-9326/aae492, 2018.
Lu, X., Zhang, S., Xing, J., Wang, Y., Chen, W., Ding, D., Wu, Y., Wang, S., Duan, L., and Hao, J.: Progress of air pollution control in China and its challenges and opportunities in the ecological civilization era, Engineering, 6, 1423–1431, 2020.
Lu, X., Sha, Y. H., Li, Z., Huang, Y., Chen, W., Chen, D., Shen, J., Chen, Y., and Fung, J. C.: Development and application of a hybrid long-short term memory–three dimensional variational technique for the improvement of PM2.5 forecasting, Sci. Total Environ., 770, 144221, https://doi.org/10.1016/j.scitotenv.2020.144221, 2021.
Lyu, B., Hu, Y., Zhang, W., Du, Y., Luo, B., Sun, X., Sun, Z., Deng, Z., Wang, X., and Liu, J.: Fusion method combining ground-level observations with chemical transport model predictions using an ensemble deep learning framework: application in China to estimate spatiotemporally-resolved PM2.5 exposure fields in 2014–2017, Environ. Sci. Technol., 53, 7306–7315, 2019.
Ma, J., Ding, Y., Cheng, J. C., Jiang, F., and Wan, Z.: A temporal-spatial interpolation and extrapolation method based on geographic Long Short-Term Memory neural network for PM2.5, J. Clean. Prod., 237, 117729, https://doi.org/10.1016/j.jclepro.2019.117729, 2019.
Mao, W., Wang, W., Jiao, L., Zhao, S., and Liu, A.: Modeling air quality prediction using a deep learning approach: Method optimization and evaluation, Sustain. Cities Soc., 65, 102567, https://doi.org/10.1016/j.scs.2020.102567, 2021.
Pak, U., Ma, J., Ryu, U., Ryom, K., Juhyok, U., Pak, K., and Pak, C.: Deep learning-based PM2.5 prediction considering the spatiotemporal correlations: A case study of Beijing, China, Sci. Total Environ., 699, 133561, https://doi.org/10.1016/j.scitotenv.2019.07.367, 2020.
Qi, Y., Li, Q., Karimian, H., and Liu, D.: A hybrid model for spatiotemporal forecasting of PM2.5 based on graph convolutional neural network and long short-term memory, Sci. Total Environ., 664, 1–10, 2019.
Qin, D., Yu, J., Zou, G., Yong, R., Zhao, Q., and Zhang, B.: A novel combined prediction scheme based on CNN and LSTM for urban PM2.5 concentration, IEEE Access, 7, 20050–20059, 2019.
Samal, K. K. R., Panda, A. K., Babu, K. S., and Das, S. K.: An improved pollution forecasting model with meteorological impact using multiple imputation and fine-tuning approach, Sustain. Cities Soc., 70, 102923, https://doi.org/10.1016/j.scs.2021.102923, 2021.
Sayeed, A., Choi, Y., Eslami, E., Jung, J., Lops, Y., Salman, A. K., Lee, J.-B., Park, H.-J., and Choi, M.-H.: A novel CMAQ-CNN hybrid model to forecast hourly surface-ozone concentrations 14 days in advance, Sci. Rep.-UK, 11, 1–8, 2021a.
Sayeed, A., Lops, Y., Choi, Y., Jung, J., and Salman, A. K.: Bias correcting and extending the PM forecast by CMAQ up to 7 days using deep convolutional neural networks, Atmos. Environ., 253, 118376, https://doi.org/10.1016/j.atmosenv.2021.118376, 2021b.
Schuster, M. and Paliwal Kuldip, K.: Bidirectional recurrent neural networks, IEEE T. Signal Proces., 45, 2673–2681, 1997.
Siami-Namini, S., Tavakoli, N., and Namin, A. S.: A comparison of ARIMA and LSTM in forecasting time series, 2018 17th IEEE international conference on machine learning and applications (ICMLA), 17–20 December 2018, Orlando, Florida, USA, 1394–1401, https://doi.org/10.1109/ICMLA.2018.00227, 2018.
Skamarock, W. C., Klemp, J. B., Dudhia, J., Gill, D. O., Barker, D., Duda, M. G., Huang, X. Y., Wang W., and Powers, J. G.: A description of the Advanced Research WRF version 3, NCAR Technical note-475+ STR, https://doi.org/10.5065/D68S4MVH, 2008 (data available at https://www2.mmm.ucar.edu/wrf/users/download/get_source.html, last access: 14 November 2022).
Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., and Salakhutdinov, R.: Dropout: a simple way to prevent neural networks from overfitting, J. Mach. Learn. Res., 15, 1929–1958, 2014.
Sun, H., Fung, J. C., Chen, Y., Chen, W., Li, Z., Huang, Y., Lin, C., Hu, M., and Lu, X.: Improvement of PM2.5 and O3 forecasting by integration of 3D numerical simulation with deep learning techniques, Sustain. Cities Soc., 75, 103372, https://doi.org/10.1016/j.scs.2021.103372, 2021.
Sun, H., Fung, J. C. H., Chen, Y., Li, Z., Yuan, D., Chen, W., and Lu, X.: Ground obeservation data (meteorological factors and air pollution) in Greater Bay Area, 2015–2021, Zenodo [data set], https://doi.org/10.5281/zenodo.6598377, 2022a.
Sun, H., Fung, J. C. H., Chen, Y., Li, Z., Yuan, D., Chen, W., and Lu, X.: Prediction of the broadcasting model and various baselines, Zenodo [data set], https://doi.org/10.5281/zenodo.6833673, 2022b.
Sun, H., Fung, J. C. H., Chen, Y., Li, Z., Yuan, D., Chen, W., and Lu, X.: Deep learning models in the study “Development of an LSTM-Broadcasting deep-learning framework for regional air pollution forecast improvement”, Zenodo [data set], https://doi.org/10.5281/zenodo.6827585, 2022c.
Sun, H., Fung, J. C. H., Chen, Y., Li, Z., Yuan, D., Chen, W., and Lu, X.: Processed ground observation and WRF-CAMQ data for Greater Bay Area, 2015–2021, Zenodo [data set], https://doi.org/10.5281/zenodo.6601173, 2022d.
Sun, J. H.: jvhs0706/regional-forecast-new: GMD paper code, Zenodo [code], https://doi.org/10.5281/zenodo.7019243, 2022.
Sutskever, I., Vinyals, O., and Le, Q. V.: Sequence to sequence learning with neural networks, Adv. Nur. In., 27, 3104–3112, https://doi.org/10.48550/arXiv.1409.3215, 2014.
Tang, Y., Lee, P., Tsidulko, M., Huang, H.-C., McQueen, J. T., DiMego, G. J., Emmons, L. K., Pierce, R. B., Thompson, A. M., and Lin, H.-M.: The impact of chemical lateral boundary conditions on CMAQ predictions of tropospheric ozone over the continental United States, Environ. Fluid Mech., 9, 43–58, 2009.
Tsai, Y.-T., Zeng, Y.-R., and Chang, Y.-S.: Air pollution forecasting using RNN with LSTM, 2018 IEEE 16th Intl Conf on Dependable, Autonomic and Secure Computing, 16th Intl Conf on Pervasive Intelligence and Computing, 4th Intl Conf on Big Data Intelligence and Computing and Cyber Science and Technology Congress (DASC/PiCom/DataCom/CyberSciTech), 12–15 August 2018, Athens, Greece, 1074–1079, https://doi.org/10.1109/DASC/PiCom/DataCom/CyberSciTec.2018.00178, 2018.
United States Environmental Protection Agency: CMAQ (Version 5.0.2), Zenodo [software], https://doi.org/10.5281/zenodo.1079898, 2014.
Wu, Q. and Lin, H.: Daily urban air quality index forecasting based on variational mode decomposition, sample entropy and LSTM neural network, Sustain. Cities Soc., 50, 101657, https://doi.org/10.1016/j.scs.2019.101657, 2019.
Zhang, Y., Bocquet, M., Mallet, V., Seigneur, C., and Baklanov, A.: Real-time air quality forecasting, part II: State of the science, current research needs, and future prospects, Atmos. Environ., 60, 656–676, 2012.
Zhao, J., Deng, F., Cai, Y., and Chen, J.: Long short-term memory-Fully connected (LSTM-FC) neural network for PM2.5 concentration prediction, Chemosphere, 220, 486–492, 2019.
Zhou, X., Tong, W., and Li, L.: Deep learning spatiotemporal air pollution data in China using data fusion, Earth Sci. Inform., 13, 859–868, 2020.
This study developed a novel deep-learning layer, the broadcasting layer, to build an end-to-end LSTM-based deep-learning model for regional air pollution forecast. By combining the ground observation, WRF-CMAQ simulation, and the broadcasting LSTM deep-learning model, forecast accuracy has been significantly improved when compared to other methods. The broadcasting layer and its variants can also be applied in other research areas to supersede the traditional numerical interpolation methods.
This study developed a novel deep-learning layer, the broadcasting layer, to build an end-to-end...