Lagrangian Particle Dispersion Model Research Articles

Revealing the significant acceleration of hydrofluorocarbon (HFC) emissions in eastern Asia through long-term atmospheric observations

Abstract. Hydrofluorocarbons (HFCs) are powerful anthropogenic greenhouse gases (GHGs) with high global-warming potentials (GWPs). They have been widely used as refrigerants, insulation foam-blowing agents, aerosol propellants, and fire suppression agents. Since the mid-1990s, emissions of HFCs have been increasing rapidly as they are used in many applications to replace ozone-depleting chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons (HCFCs) whose consumption and production have been phased out under the Montreal Protocol (MP). Due to the high GWP of HFCs, the Kigali Amendment to the MP requires the phasedown of production and consumption of HFCs to gradually achieve an 80 %–85 % reduction by 2047, starting in 2019 for non-Article 5 (developed) countries with a 10 % reduction against each defined baseline and later schedules for Article 5 (developing) countries. In this study, we have examined long-term high-precision measurements of atmospheric abundances of five major HFCs (HFC-134a, HFC-143a, HFC-32, HFC-125, and HFC-152a) at Gosan station, Jeju Island, South Korea, from 2008 to 2020. Background abundances of HFCs gradually increased, and the inflow of polluted air masses with elevated abundances from surrounding source regions were detected over the entire period. From these pollution events, we inferred regional and country-specific HFC emission estimates using two independent Lagrangian particle dispersion models and Bayesian inversion frameworks (FLEXPART-FLEXINVERT+ and NAME-InTEM). The spatial distribution of the derived “top-down” (measurement based) emissions for all HFCs shows large fluxes from megacities and industrial areas in the region. Our most important finding is that HFC emissions in eastern China and Japan have sharply increased from 2016 to 2018. The contribution of East Asian HFC emissions to the global total increased from 9 % (2008–2014) to 13 % (2016–2020). In particular, HFC emissions in Japan (Annex I country) rose rapidly from 2016 onward, with accumulated total inferred HFC emissions being ∼ 114 Gg yr−1, which is ∼ 76 Gg yr−1 higher for 2016–2020 than the “bottom-up” (i.e., based on activity data and emission factors) emissions of ∼ 38 Gg yr−1 reported to the United Nations Framework Convention on Climate Change (UNFCCC). This is likely related to the increase in domestic demand in Japan for refrigerants and air-conditioning-system-related products and incomplete accounting. A downward trend of HFC emissions that started in 2019 reflects the effectiveness of the F-gas policy in Japan. Eastern China and South Korea, though not obligated to report to the UNFCCC, voluntarily reported emissions, which also show differences between top-down and bottom-up emission estimates, demonstrating the need for atmospheric measurements, comprehensive data analysis, and accurate reporting for precise emission management. Further, the proportional contribution of each country's CO2-equivalent HFC emissions has changed over time, with HFC-134a decreasing and HFC-125 increasing. This demonstrates the transition in the predominant HFC substances contributing to global warming in each country.

Observed decade-long improvement of combustion efficiency in the Yangtze River Delta region in China

The ΔCO/ΔCO2 ratio is a good indicator of the combustion efficiency of carbon-containing fuels, and can be useful to assess the combustion efficiency on a regional scale. In this study, we analyzed in-situ observations of CO2 and CO concentrations from 2011 to 2021 at the Station for Observing Regional Processes of the Earth System (SORPES), in the Yangtze River Delta (YRD) region of eastern China, and calculated the ΔCO/ΔCO2 ratio to investigate the combustion efficiency in the YRD region. Furthermore, we used a Lagrangian particle dispersion model WRF-FLEXPART to evaluate the contribution of each emission sources to the observed ΔCO/ΔCO2 ratio. We found that the observed ΔCO/ΔCO2 ratio showed a persistent decreasing trend of 1.0 ppb/ppm per year and decreased ∼47.9% during this period, illustrating an evident improvement in the combustion efficiency in the YRD region. The improvement of the combustion efficiency is a result of China’s Air Pollution Prevention and Control Action Plan announced in 2013. However, the decrease of ΔCO/ΔCO2 ratio slowed down from 1.3 ppb ppm−1 per year during 2011–2016 to 0.6 ppb ppm−1 per year during 2017–2021. The simulation results reveal that the slowdown of the decrease in the ΔCO/ΔCO2 ratios can be explained by the slowing improvement of combustion efficiency in steel source in the industry sector. Our results verify the effectiveness of emission reduction efforts in the YRD region and highlight the necessity of long-term observations of CO2 and CO.

A Numerical Study of Smoke Concentration Distribution near Low-Intensity Fires: Sensitivity to Vertical Canopy Structure and Fire Heat Source Strength

Abstract Broadly speaking, prediction of the negative impacts of prescribed fire on air quality is limited by gaps in our understanding of the underlying fire, fuels, and atmospheric processes. These knowledge gaps hinder our ability to accurately predict smoke concentration distributions, leading to unintended smoke intrusions into nearby communities and subsequent threats to public health and safety. In this study, numerical simulations are performed using the Flexible Particle Weather Research and Forecasting (FLEXPART-WRF) Model, a Lagrangian particle dispersion model, with particle motion driven by output from a full-physics atmospheric model with a forest canopy submodel and 10-m horizontal grid spacing [Advanced Regional Prediction System (ARPS)-CANOPY], to address two research questions. First, what is the relationship between near-fire (within ∼50–150 m of fire) smoke concentration distribution and (i) vertical canopy structure and (ii) fire heat source strength? Second, what roles do mean transport (i.e., transport by the mean wind) and turbulent diffusion play in shaping the near-fire smoke concentration distribution? To address these questions, simulations are run with 25 combinations of plant area density profile and fire sensible heat flux magnitude, and smoke is represented by particles with diameters ≤ 2.5 μm (PM2.5). Results show that near-fire PM2.5 concentration distribution is primarily controlled by vertical canopy structure, with fire heat source strength primarily controlling the PM2.5 concentration magnitude. Analysis of the underlying ARPS-CANOPY variables driving the FLEXPART-WRF particle dispersion helps elucidate the roles of mean transport and turbulent diffusion. In total, the study findings suggest that the vertical distribution of canopy vegetation and fire heat source strength are important factors influencing PM2.5 dispersion and concentration distribution near low-intensity fires.

Accelerating Lagrangian transport simulations on graphics processing units: performance optimizations of Massive-Parallel Trajectory Calculations (MPTRAC) v2.6

Abstract. Lagrangian particle dispersion models are indispensable tools for the study of atmospheric transport processes. However, Lagrangian transport simulations can become numerically expensive when large numbers of air parcels are involved. To accelerate these simulations, we made considerable efforts to port the Massive-Parallel Trajectory Calculations (MPTRAC) model to graphics processing units (GPUs). Here we discuss performance optimizations of the major bottleneck of the GPU code of MPTRAC, the advection kernel. Timeline, roofline, and memory analyses of the baseline GPU code revealed that the application is memory-bound, and performance suffers from near-random memory access patterns. By changing the data structure of the horizontal wind and vertical velocity fields of the global meteorological data driving the simulations from structure of arrays (SoAs) to array of structures (AoSs) and by introducing a sorting method for better memory alignment of the particle data, performance was greatly improved. We evaluated the performance on NVIDIA A100 GPUs of the Jülich Wizard for European Leadership Science (JUWELS) Booster module at the Jülich Supercomputing Center, Germany. For our largest test case, transport simulations with 108 particles driven by the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 reanalysis, we found that the runtime for the full set of physics computations was reduced by 75 %, including a reduction of 85 % for the advection kernel. In addition to demonstrating the benefits of code optimization for GPUs, we show that the runtime of central processing unit (CPU-)only simulations is also improved. For our largest test case, we found a runtime reduction of 34 % for the physics computations, including a reduction of 65 % for the advection kernel. The code optimizations discussed here bring the MPTRAC model closer to applications on upcoming exascale high-performance computing systems and will also be of interest for optimizing the performance of other models using particle methods.

Bayesian Inference for Multiple Datasets

Estimating parameters for multiple datasets can be time consuming, especially when the number of datasets is large. One solution is to sample from multiple datasets simultaneously using Bayesian methods such as adaptive multiple importance sampling (AMIS). Here, we use the AMIS approach to fit a von Mises distribution to multiple datasets for wind trajectories derived from a Lagrangian Particle Dispersion Model driven from 3D meteorological data. A posterior distribution of parameters can help to characterise the uncertainties in wind trajectories in a form that can be used as inputs for predictive models of wind-dispersed insect pests and the pathogens of agricultural crops for use in evaluating risk and in planning mitigation actions. The novelty of our study is in testing the performance of the method on a very large number of datasets (>11,000). Our results show that AMIS can significantly improve the efficiency of parameter inference for multiple datasets.

Lagrangian and Eulerian approach to predict the movement of radionuclides in selected potential sites in Malaysia during the monsoon period

In computing the movement of particles through the fluid, three types of dispersion models are used which are Gaussian plume models, Lagrangian puff models, and small-scale numerical models. Lagrangian particle dispersion models are increasingly used for nuclear applications. In this study, the usage of the Lagrangian model is implemented in a Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) which combines with the Eulerian model to predict the movement of radionuclides from the simulated nuclear power plant accident in selected locations in Malaysia. These include nuclear risk studies in these areas, emergency response systems, and source term analyses. This study aims to compare the risk from the dispersion of 137Cs from a simulated Nuclear Power Plant (NPP) at Mukah and Tasik Kenyir during the Northeast monsoon (8th October 2022) and Southeast monsoon (16th May 2022). The HYSPLIT model was set up based on the meteorological data during both monsoon periods and was simulated for 5 days. The outcomes revealed that the concentration of 137Cs is higher at Tasik Kenyir compared to Mukah for both during the Northeast monsoon period (3.5 x 107 Bq/m2) while for Southwest monsoon, the result is vice versa where the concentration of 137Cs at Mukah is (2.2 x 107Bq/m2). The activity concentration is then converted into rate effective dose measurement and compared with the limit suggested by ICRP which is 1 Sv per year. It is concluded that Tasik Kenyir is a more suitable location to build NPP compared to Mukah as the movement of radionuclides was dispersed to the location that contained less population.

Tracking Carbon Dioxide with Lagrangian Transport Simulations: Case Study of Canadian Forest Fires in May 2021

The large amounts of greenhouse gases, such as carbon dioxide, produced by severe forest fires not only seriously affect the ecosystems in the area where the fires occur but also cause a greenhouse effect that has a profound impact on the natural environment in other parts of the world. Numerical simulations of greenhouse gas transport processes are often affected by uncertainties in the location and timing of the emission sources and local meteorological conditions, and it is difficult to obtain accurate and credible predictions by combining remote sensing satellite data with given meteorological forecasts or reanalyses. To study the regional transport processes and impacts of greenhouse gases produced by sudden large-scale forest fires, this study applies the Lagrangian particle dispersion model Massive-Parallel Trajectory Calculations (MPTRAC) to conduct forward simulations of the CO2 transport process of greenhouse gases emitted from forest fires in the central region of Saskatchewan, Canada, during the period of 17 May to 25 May 2021. The simulation results are validated with the Orbiting Carbon Observatory-2 Goddard Earth Observing System (OCO-2 GEOS) Level 3 daily gridded CO2 product over the study area. In order to leverage the high computational costs of the numerical simulations of the model, we implement the forward simulations on the Tianhe-2 supercomputer platform and the JUWELS HPC system, which greatly improves the computational efficiency through parallel computation and makes near-real-time predictions of atmospheric transport processes feasible.

The Lagrangian Atmospheric Radionuclide Transport Model (ARTM) – sensitivity studies and evaluation using airborne measurements of power plant emissions

Abstract. The Atmospheric Radionuclide Transport Model (ARTM) operates at the meso-γ scale and simulates the dispersion of radionuclides originating from nuclear facilities under routine operation within the planetary boundary layer. This study presents the extension and validation of this Lagrangian particle dispersion model and consists of three parts: (i) a sensitivity study that aims to assess the impact of key input parameters on the simulation results, (ii) the evaluation of the mixing properties of five different turbulence models using the well-mixed criterion, and (iii) a comparison of model results to airborne observations of carbon dioxide (CO2) emissions from a power plant and the evaluation of related uncertainties. In the sensitivity study, we analyse the effects of the stability class, roughness length, zero-plane displacement factor, and source height on the three-dimensional plume extent as well as the distance between the source and maximum concentration at the ground. The results show that the stability class is the most sensitive input parameter as expected. The five turbulence models are the default turbulence models of ARTM 2.8.0 and ARTM 3.0.0, one alternative built-in turbulence model of ARTM, and two further turbulence models implemented for this study. The well-mixed condition tests showed that all five turbulence models are able to preserve an initially well-mixed atmospheric boundary layer reasonably well. The models deviate only 6 % from the expected uniform concentration below 80 % of the mixing layer height, except for the default turbulence model of ARTM 3.0.0 with deviations of up to 18 %. CO2 observations along a flight path in the vicinity of the lignite power plant Bełchatów, Poland, measured by the Deutsches Zentrum für Luft- und Raumfahrt (DLR) Cessna aircraft during the Carbon Dioxide and Methane Mission (CoMet) campaign in 2018 allowed for evaluation of model performance for the different turbulence models under unstable boundary layer conditions. All simulated mixing ratios are of the same order of magnitude as the airborne in situ data. An extensive uncertainty analysis using probability distribution functions, statistical tests, and direct spatio-temporal comparisons of measurements and model results help to quantify the model uncertainties. With the default turbulence setups of ARTM versions 2.8.0 and 3.0.0, the plume widths are underestimated by up to 50 %, resulting in a strong overestimation of the maximum plume CO2 mixing ratios. The comparison of the three alternative turbulence models shows good agreement of the peak plume CO2 concentrations, the CO2 distribution within the plumes, and the plume width, with a 30 % deviation in the peak CO2 concentration and a less than 25 % deviation in the measured CO2 plume width. Uncertainties in the simulations may arise from the different spatial and temporal resolutions of simulations and measurements in addition to the turbulence parametrisation and boundary conditions. The results of this work may help to improve the accurate representation of real plumes in very unstable atmospheric conditions through the selection of distinct turbulence models. Further comparisons at different stability regimes are required for a final assessment of model uncertainties.

Identifying the Moisture Sources in Different Seasons for Abaya-Chamo Basin of Southern Ethiopia Using Lagrangian Particle Dispersion Model

Understanding the sources of precipitation and their impacts is crucial for basin-wide water balance research. Previous research concentrated on the sources of moisture in Ethiopia. The southern part’s moisture sources, however, were not investigated. The primary objective of this study is to trace the source of atmospheric moisture in the Abaya-Chamo sub-basin of southern Ethiopia using numerical water vapor tracers like Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model. Exploring the possible regions of atmospheric vapor roots and the path of moist air initiating rainfall that reaches the basin was feasible for the year 2018–2020. The anticyclone from the Arabian High, which is positioned in the Arabian and Mediterranean seas, was the primary source of moisture supply in the study area during the Belg (March to May) season, according to the back trajectory cluster analysis results. Additionally, the Indian Ocean adds moisture resulting from Mascarene highs brought by equatorial easterlies. Furthermore, during Kiremt (June to September), air masses from the Congo basin were the potential moisture source region for the study areas in combination with air masses originating from the Mascarene highs, located in the South Indian Ocean, and the St. Helena high, centered in the subtropical southern Atlantic Ocean. This study primarily focuses on the complex dynamics of atmospheric moisture sources around Abaya-Chamo sub-basin of southern Ethiopia, offering insight into seasonal fluctuations and contributing various components. These findings contribute to basin-specific water balance research by filling gaps in the previous studies.

Reduction of simulation times by application of a kernel method in a high-resolution Lagrangian particle dispersion model

Reduction of simulation times by application of a kernel method in a high-resolution Lagrangian particle dispersion model

Impact of transport model resolution and a priori assumptions on inverse modeling of Swiss F-gas emissions

Abstract. Inverse modeling is a widely used top-down method to infer greenhouse gas (GHG) emissions and their spatial distribution based on atmospheric observations. The errors associated with inverse modeling have multiple sources, such as observations and a priori emission estimates, but they are often dominated by the transport model error. Here, we utilize the Lagrangian particle dispersion model (LPDM) FLEXPART (FLEXible PARTicle Dispersion Model), driven by the meteorological fields of the regional numerical weather prediction model COSMO. The main sources of errors in LPDMs are the turbulence diffusion parameterization and the meteorological fields. The latter are outputs of an Eulerian model. Recently, we introduced an improved parameterization scheme of the turbulence diffusion in FLEXPART, which significantly improves FLEXPART-COSMO simulations at 1 km resolution. We exploit F-gas measurements from two extended field campaigns on the Swiss Plateau (in Beromünster and Sottens), and we conduct both high-resolution (1 km) and low-resolution (7 km) FLEXPART transport simulations that are then used in a Bayesian analytical inversion to estimate spatial emission distributions. Our results for four F-gases (HFC-134a, HFC-125, HFC-32, SF6) indicate that both high-resolution inversions and a dense measurement network significantly improve the ability to estimate spatial distribution of the emissions. Furthermore, the total emission estimates from the high-resolution inversions (351 ± 44 Mg yr−1 for HFC-134a, 101 ± 21 Mg yr−1 for HFC-125, 50 ± 8 Mg yr−1 for HFC-32, 9.0 ± 1.1 Mg yr−1 for SF6) are significantly higher compared to the low-resolution inversions (20 %–40 % increase) and result in total a posteriori emission estimates that are closer to national inventory values as reported to the UNFCCC (10 %–20 % difference between high-resolution inversion estimates and inventory values compared to 30 %–40 % difference between the low-resolution inversion estimates and inventory values). Specifically, we attribute these improvements to a better representation of the atmospheric flow in complex terrain in the high-resolution model, partly induced by the more realistic topography. We further conduct numerous sensitivity inversions, varying different parameters and variables of our Bayesian inversion framework to explore the whole range of uncertainty in the inversion errors (e.g., inversion grid, spatial distribution of a priori emissions, covariance parameters like baseline uncertainty and spatial correlation length, temporal resolution of the assimilated observations, observation network, seasonality of emissions). From the abovementioned parameters, we find that the uncertainty of the mole fraction baseline and the spatial distribution of the a priori emissions have the largest impact on the a posteriori total emission estimates and their spatial distribution. This study is a step towards mitigating the errors associated with the transport models and better characterizing the uncertainty inherent in the inversion error. Improvements in the latter will facilitate the validation and standardization of national GHG emission inventories and support policymakers.

Modelling the impacts of emission changes on O3 sensitivity, atmospheric oxidation capacity, and pollution transport over the Catalonia region

Abstract. Tropospheric ozone (O3) is an important surface pollutant in urban areas, and it has complex formation mechanisms that depend on the atmospheric chemistry and on meteorological factors. The severe reductions observed in anthropogenic emissions during the COVID-19 pandemic can further our understanding of the photochemical mechanisms leading to O3 formation and provide guidance for policies aimed at reducing air pollution. In this study, we use the Weather Research and Forecasting model with Chemistry (WRF-Chem) coupled with the urban canopy building effect parameterization and building energy model (BEP + BEM) to investigate changes in the ozone chemistry over the metropolitan area of Barcelona (AMB) and its atmospheric plume moving northwards, which is responsible for the highest number of hourly O3 exceedances in Spain. The trajectories of the air masses from the AMB to the Pyrenees are studied with the Lagrangian FLEXible PARTicle dispersion model with WRF (FLEXPART-WRF). The aim is to investigate the response of ozone chemistry to reduction in precursor emissions (NOx – nitrogen oxides; VOCs – volatile organic compounds). The results show that, with the reduction in emissions, (1) the ozone chemistry tends to enter the NOx-limited or transition regimes, but highly polluted urban areas are still in the VOC-limited regime; (2) the reduced O3 production is overwhelmed by reduced nitric oxide (NO) titration, resulting in a net increase in the O3 concentration (up to 20 %) in the evening; (3) the increase in the maximum O3 level (up to 6 %) during the highest emission-reduction period could be attributed to an enhancement in the atmospheric oxidants hydroxyl and nitrate radical (OH and NO3) given their strong link with O3 loss or production chemistry; (4) the daily maximum levels of ozone and odd oxygen species (Ox) generally decreased (4 %) in May – a period with intense radiation which favours ozone production – with the reduced atmospheric OH and NO3 oxidants, indicating an improvement in the air quality; and (5) ozone precursor concentration changes in the urban plume of Barcelona contribute significantly to the level of pollution along the 150 km south-to-north valley in the Pyrenees. Our results indicate that O3 abatement strategies cannot rely only on NOx emission control but must include a significant reduction in anthropogenic sources of VOCs. In addition, our results show that mitigation strategies intended to reduce O3 should be designed according to the local meteorology, air transport, particular ozone regimes, and oxidation capacity of the atmosphere of the urban area.

Cloud condensation nuclei activity of internally mixed particle populations at a remote marine free troposphere site in the North Atlantic Ocean

This study reports results from research conducted at the Observatory of Mount Pico (OMP), 2225 m above mean sea level on Pico Island in the Azores archipelago in June and July 2017. We investigated the chemical composition, mixing state, and cloud condensation nuclei (CCN) activities of long-range transported free tropospheric (FT) particles. FLEXible PARTicle Lagrangian particle dispersion model (FLEXPART) simulations reveal that most air masses that arrived at the OMP during the sampling period originated in North America and were highly aged (average plume age > 10 days). We probed size-resolved chemical composition, mixing state, and hygroscopicity parameter (κ) of individual particles using computer-controlled scanning electron microscopy with an energy-dispersive X-ray spectrometer (CCSEM-EDX). Based on the estimated individual particle mass from elemental composition, we calculated the mixing state index, χ. During our study, FT particle populations were internally mixed (χ of samples are between 53 % and 87 %), owing to the long atmospheric aging time. We used data from a miniature Cloud Condensation Nucleus Counter (miniCCNC) to derive the hygroscopicity parameter, κCCNC. Combining κCCNC and FLEXPART, we found that air masses recirculated above the North Atlantic Ocean with lower mean altitude had higher κCCNC due to the higher contribution of sea salt particles. We used CCSEM-EDX and phase state measurements to predict single-particle κ (κCCSEM-EDX) values, which overlap with the lower range of κCCNC measured below 0.15 % SS. Therefore, CCSEM-EDX measurements can be useful in predicting the lower bound of κ, which can be used in climate models to predict CCN activities, especially in remote locations where online CCN measurements are unavailable.

Modelling the Potential Long-Range Dispersion of Atmospheric Microplastics Reaching a Remote Site

The study of the microplastic dispersion in the atmosphere is a relatively new field of research. In the studies available in literature, it was suggested that long-range transport may play a main role in the motion of microplastics towards remote sites, where no significant local sources are present. In this framework, here the Lagrangian particle dispersion model MILORD was used in the backward mode to investigate the potential long-range dispersion of microplastics in the atmosphere and to assess possible source areas that may determine microplastic pollution in a remote site. The chosen case study refers to the work of Allen et al. (2019), who analysed the observations of atmospheric microplastics deposition in a remote and pristine site in the French Pyrenees, also through numerical backward trajectories. Based on their dataset, they showed that microplastics could arrive at the site from distances up to about 100 km. Yet, they considered that it was unlikely to have only microplastic emissions from local sources, because of the low population density in the area. The results of MILORD simulations for different periods showed that for this case study the dominant scale of the microplastics transport process is regional up to mesoscale, while long-range transport may contribute when the tracer is mostly travelling in the free atmosphere before being deposited at the site. The relative contributions of dry and wet depositions and corresponding rates were also assessed. It was confirmed that plastic particles characterised by the highest settling velocities, reasonably associated to large particles, are unlikely to be subjected to long-range transport.

A new method for the quantification of ambient particulate-matter emission fluxes

Abstract. An inversion method has been developed in order to quantify the emission fluxes of certain aerosol pollution sources across a wide region in the Northern Hemisphere, mainly in Europe and western Asia. The data employed are the aerosol contribution factors deducted by positive matrix factorization (PMF) on a PM2.5 chemical composition dataset from 16 European and Asian cities for the period 2014 to 2016. The spatial resolution of the method corresponds to the geographic grid cell size of the Lagrangian particle dispersion model (Flexible Particle Dispersion Model, FLEXPART, 1∘ × 1∘) which was utilized for the air mass backward simulations. The area covered is also related to the location of the 16 cities under study. Species with an aerodynamic geometric mean diameter of 400 nm and 3.1 µm and a geometric standard deviation of 1.6 and 2.25, respectively, were used to model the secondary sulfate and dust aerosol transport. Potential source contribution function (PSCF) analysis and generalized Tikhonov regularization were applied so as to acquire potential source areas and quantify their emission fluxes. A significant source area for secondary sulfate on the east of the Caspian Sea is indicated, when data from all stations are used. The maximum emission flux in that area is as high as 10 × 10−12 kg m−2 s−1. When Vilnius, Dushanbe, and Kurchatov data were excluded, the areas with the highest emission fluxes were the western and central Balkans and southern Poland. The results display many similarities to the SO2 emission maps provided by the OMI-HTAP (Ozone Monitoring Instrument-Hemispheric Transport Air Pollution) and ECLIPSE (Evaluating the Climate and Air Quality Impacts of Short-Lived Pollutants) databases. For dust aerosol, measurements from Athens, Belgrade, Debrecen, Lisbon, Tirana, and Zagreb are utilized. The west Sahara region is indicated as the most important source area, and its contribution is quantified, with a maximum of 17.6 × 10−12 kg m−2 s−1. When we apply the emission fluxes from every geographic grid cell (1∘ × 1∘) for secondary sulfate aerosol deducted with the new method to air masses originating from Vilnius, a useful approximation to the measured values is achieved.

High-Resolution Bayesian Inversion of Carbon Dioxide Flux Over Peninsular India

Accurate estimation of the flux of the most important greenhouse gas, carbon dioxide (CO2), plays a decisive role in designing the policies to reduce the atmospheric CO2 burden to combat climate change. Here we integrate recent (2017–2020) in-situ measurements with a high-resolution particle dispersion model for optimising CO2 flux over peninsular India based on a top-down approach. Using weekly observations of atmospheric boundary layer CO2 from three distinct locations representing peninsular India, the total surface fluxes of CO2 are optimised based on the fluxes from fossil fuel, wildfire, terrestrial biosphere and ocean. Our Bayesian mode inversion using the Lagrangian Particle Dispersion Model FLEXPART suggests that the peninsular Indian region is a slightly stronger source of CO2 (3.34 TgC yr−1) than what is synthesised in the fossil fuel and ecosystem exchanges based on independent estimations. On seasonal scale, the flux corrections to the prior fluxes over peninsular India are 4.68, 6.53, −2.28 and 4.41 TgC yr−1during winter, pre-monsoon, monsoon and post-monsoon seasons, respectively. The CO2 measurements from the stations attempted in the present inversion experiments could capture the footprints of peninsular India, giving better constraints to fluxes in the inversion. However, the data gap, pristineness of the samples, and large variabilities in the observations call for sustained monitoring for a longer term to further reduce the uncertainty of the estimated flux.

Three-Dimensional Visualization of Long-Range Atmospheric Transport of Crop Pathogens and Insect Pests

Some of the most devastating crop diseases and insect pests can be transmitted by wind over extremely long distances. These low-probability but high-impact events can have severe consequences for crop production and food security by causing epidemic outbreaks or devastating insect infestations in previously uninfected geographic areas. Two prominent examples that have recently caused substantial damage to agricultural production are novel strains of wheat rusts and desert locust swarm infestations. Whilst quantitative estimates of long-range atmospheric transport events can be obtained using meteorological transport simulations, the exact characteristics of three-dimensional spatiotemporal dynamics of crop pathogen transport and insect flight on extremely large spatial scales, over entire regions and continents, remain largely unknown. Here, we investigate the feasibility and usefulness of new advanced geospatial data visualization methods for studying extremely long-distance airborne transmission of crop pathogens and insect pests. We combine field surveillance data and a Lagrangian Particle Dispersion Model with novel techniques from computer graphics to obtain, for the first time, detailed three-dimensional visual insights into airborne crop pathogen and insect pest transport on regional and continental scales. Visual insights into long-distance dispersal of pests and pathogens are presented as a series of short 3D movies. We use interactive three-dimensional visual data analysis for explorative examination of long-range atmospheric transport events from a selection of outbreak and infestation sites in East Africa and South East Asia. The practical usefulness of advanced 3D visualization methods for improving risk estimates and early warning is discussed in the context of two operational crop disease and insect pest management systems (for wheat rusts and desert locusts). The tools and methods introduced here can be applied to other pathogens, pests, and geographical areas and can improve understanding of risks posed to agricultural production by crop disease and insect pest transmission caused by meteorological extreme events.

Improving trajectory calculations by FLEXPART 10.4+ using single-image super-resolution

Abstract. Lagrangian trajectory or particle dispersion models as well as semi-Lagrangian advection schemes require meteorological data such as wind, temperature and geopotential at the exact spatiotemporal locations of the particles that move independently from a regular grid. Traditionally, these high-resolution data have been obtained by interpolating the meteorological parameters from the gridded data of a meteorological model or reanalysis, e.g., using linear interpolation in space and time. However, interpolation is a large source of error for these models. Reducing them requires meteorological input fields with high space and time resolution, which may not always be available and can cause severe data storage and transfer problems. Here, we interpret this problem as a single-image super-resolution task. That is, we interpret meteorological fields available at their native resolution as low-resolution images and train deep neural networks to upscale them to a higher resolution, thereby providing more accurate data for Lagrangian models. We train various versions of the state-of-the-art enhanced deep residual networks for super-resolution (EDSR) on low-resolution ERA5 reanalysis data with the goal to upscale these data to an arbitrary spatial resolution. We show that the resulting upscaled wind fields have root-mean-squared errors half the size of the winds obtained with linear spatial interpolation at acceptable computational inference costs. In a test setup using the Lagrangian particle dispersion model FLEXPART and reduced-resolution wind fields, we find that absolute horizontal transport deviations of calculated trajectories from “true” trajectories calculated with un-degraded 0.5∘ × 0.5∘ winds are reduced by at least 49.5 % (21.8 %) after 48 h relative to trajectories using linear interpolation of the wind data when training on 2∘ × 2∘ to 1∘ × 1∘ (4∘ × 4∘ to 2∘ × 2∘) resolution data.

A machine learning emulator for Lagrangian particle dispersion model footprints: a case study using NAME

Abstract. Lagrangian particle dispersion models (LPDMs) have been used extensively to calculate source-receptor relationships (“footprints”) for use in applications such as greenhouse gas (GHG) flux inversions. Because a single model simulation is required for each data point, LPDMs do not scale well to applications with large data sets such as flux inversions using satellite observations. Here, we develop a proof-of-concept machine learning emulator for LPDM footprints over a ∼ 350 km × 230 km region around an observation point, and test it for a range of in situ measurement sites from around the world. As opposed to previous approaches to footprint approximation, it does not require the interpolation or smoothing of footprints produced by the LPDM. Instead, the footprint is emulated entirely from meteorological inputs. This is achieved by independently emulating the footprint magnitude at each grid cell in the domain using gradient-boosted regression trees with a selection of meteorological variables as inputs. The emulator is trained based on footprints from the UK Met Office's Numerical Atmospheric-dispersion Modelling Environment (NAME) for 2014 and 2015, and the emulated footprints are evaluated against hourly NAME output from 2016 and 2020. When compared to CH4 concentration time series generated by NAME, we show that our emulator achieves a mean R-squared score of 0.69 across all sites investigated between 2016 and 2020. The emulator can predict a footprint in around 10 ms, compared to around 10 min for the 3D simulator. This simple and interpretable proof-of-concept emulator demonstrates the potential of machine learning for LPDM emulation.

Oceanic climate changes threaten the sustainability of Asia’s water tower

Water resources sustainability in High Mountain Asia (HMA) surrounding the Tibetan Plateau (TP)—known as Asia’s water tower—has triggered widespread concerns because HMA protects millions of people against water stress1,2. However, the mechanisms behind the heterogeneous trends observed in terrestrial water storage (TWS) over the TP remain poorly understood. Here we use a Lagrangian particle dispersion model and satellite observations to attribute about 1 Gt of monthly TWS decline in the southern TP during 2003–2016 to westerlies-carried deficit in precipitation minus evaporation (PME) from the southeast North Atlantic. We further show that HMA blocks the propagation of PME deficit into the central TP, causing a monthly TWS increase by about 0.5 Gt. Furthermore, warming-induced snow and glacial melt as well as drying-induced TWS depletion in HMA weaken the blocking of HMA’s mountains, causing persistent northward expansion of the TP’s TWS deficit since 2009. Future projections under two emissions scenarios verified by satellite observations during 2020–2021 indicate that, by the end of the twenty-first century, up to 84% (for scenario SSP245) and 97% (for scenario SSP585) of the TP could be afflicted by TWS deficits. Our findings indicate a trajectory towards unsustainable water systems in HMA that could exacerbate downstream water stress.