MERRA-2 Reanalysis Research Articles

Spatiotemporal Heterogeneity of Precipitable Water Diurnal Variation over the Tibetan Plateau Based on a Refined 2D Water Vapor Reconstruction

Atmospheric water vapor is a critical factor influencing rainfall, snowfall, and avalanche occurrences, with its diurnal variations providing key insights into such phenomena. However, in high-altitude mountainous areas, the spatial and temporal resolution of water vapor products is insufficient to depict their diurnal variation characteristics. This study aimed to refine the understanding of diurnal Precipitable Water Vapor (PWV) variations using the ANUSPLIN to obtain a two-dimensional water vapor field with a resolution of 0.1° × 0.1° during the monsoon period from 2007 to 2015, which was verified based on ground-based GPS water vapor observations. The findings reveal that incorporating China Meteorological Forcing Data (CMFD) greatly enhances the diurnal variation patterns of PWV, resulting in the appearance of extreme values of PWV daily variation about three hours later than the MERRA reanalysis. In the southern Tibetan Plateau (TP), the diurnal variation of PWV is closer to GPS PWV in terms of amplitude and phase. Spatially, PWV variations are weaker in the central TP but stronger in other regions, particularly during the mid-monsoon period (July–August). Temporally, the central TP experiences maximum PWV from nighttime to early morning (18:00–4:00 UTC), while the surrounding southern and northern areas peak from afternoon to evening (8:00–12:00 UTC), approximately three hours earlier than precipitation peaks. Temporally, the central Tibetan Plateau experiences maximum PWV from nighttime to early morning (18:00–4:00 UTC), while the surrounding southern and northern areas peak from afternoon to evening (8:00–12:00 UTC), approximately three hours earlier than precipitation peaks.

The Use of Atmospheric Reanalysis Data for the Estimation of Solar Irradiation Considering the Effect of Atmospheric Aerosols over Brazil

In recent years, several studies have evaluated the potential of renewable energy sources in response to climate change and high energy demand. Due to its equatorial location and significant solar and wind potential, Brazil has incorporated alternative sources into its energy matrix, driven by more efficient and economical technologies for solar energy. However, the availability of observed data is still limited, and many studies rely on satellite estimates or extrapolations of in situ observations from other regions, compromising the efficiency of new technologies. This study uses NASA MERRA-2 reanalysis data to evaluate the influence of aerosols and cloudiness on the estimate of solar irradiance in Brazil. INMET stations were chosen in regions representative of the Brazilian climate and geography, with more than 12 years of observational data. MERRA-2 includes aerosol fields that interact with the model’s radiation fields, with a spatial resolution of 0.5° and hourly temporal resolution. Variables used include shortwave radiation fluxes and aerosol optical depth. Statistical indices used in performance analysis include mean bias, mean squared error, and Pearson correlation coefficient. The stations’ diurnal solar irradiance cycles were compared with MERRA-2 reanalysis data, considering different scenarios of aerosol and cloudiness effects. The reanalysis data represented the Bauru and Santa Maria stations well, while others, such as Barreiras and Goiânia, showed underestimation. Monthly cycling was also analyzed, highlighting seasonality, with greater amplitude in Santa Maria and lower in Caicó. In some locations, such as Campo Grande, the influence of aerosols is more significant, especially during the dry months, when forest fires, mainly in the Amazon region, increase the aerosol optical depth. The results show that reanalysis estimates can be used to evaluate the temporal variability of solar irradiation in regions without observational data. In conclusion, the study was able to evaluate the temporal variability of solar irradiation in Brazil using MERRA-2 atmospheric reanalysis data, demonstrating that, although there are differences with observational data, reanalysis estimates are useful in areas without observed data, with values correlation values above 0.8 and reaching values close to 0.95. However, although small, the differences observed between measured and estimated solar irradiation are generally caused by the inability of models to adequately represent the fraction of clouds and aerosols in the atmosphere.

Potential of long-term satellite observations and reanalysis products for characterising soil drying: trends and drought events

Abstract. Soil drying has multiple adverse impacts on the environment, society, and economy. Thus, it is crucial to monitor and characterise related drought events and to understand how underlying geophysical trends may affect them. Here, we compare the ability of long-term satellite observations and state-of-the-art reanalysis products to characterise soil drying. We consider the European Space Agency Climate Change Initiative (ESA CCI) remote-sensing surface soil moisture products (encompassing an ACTIVE, a PASSIVE, and a COMBINED product) as well as surface and root zone soil moisture from the ERA5, ERA5-Land, and MERRA-2 reanalysis products. In addition, we use a new root zone soil moisture dataset derived from the ESA CCI COMBINED product. We analyse global surface and root zone soil moisture trends in these products over the 2000–2022 period. Furthermore, we investigate the impact of the products' trend representation on their ability to capture major seasonal soil moisture (or agroecological) drought events as a use case. The latter is based on the analysis of 17 selected drought events documented in the scientific literature; these events are characterised by their severity (the time-accumulated standardised soil moisture anomalies), magnitude (the minimum of the standardised anomalies over time), duration, and spatial extent. The soil moisture trends are globally diverse and partly contradictory between products. ERA5, ERA5-Land, and ESA CCI COMBINED show larger fractions of drying trends, whereas ESA CCI ACTIVE and MERRA-2 display more widespread wetting trends. The differences between reanalysis products are related to a positive mean bias in the precipitation trends and regionally negative biases in surface air temperature trends in MERRA-2 compared with ground observational products, suggesting that this reanalysis underestimates drying trends. Given these biases in the MERRA-2 precipitation and temperature trends and considering available validation studies, the ESA CCI COMBINED-based products and ERA5-Land are considered more reliable and are consecutively used for a synthesis of global surface and root zone soil moisture trends. This synthesis suggests a consistent tendency towards soil drying during the last 2 decades in these products in 49.3 % of the surface and 44.5 % of the root zone layers of the covered global land area. The respective fractions of wetting trends amount to 21.1 % and 20.6 % for the surface and root zone, respectively, while areas with no trend direction consensus amount to 29.6 % and 35.0 %, respectively, reflecting the considerable uncertainties associated with global soil moisture trends. Geographically, drying is localised to parts of Europe and the Mediterranean; the Black Sea–Caspian Sea and Central Asian region; Siberia; parts of the western USA and the Canadian Prairies; and larger parts of South America, parts of southern and northern Africa, and parts of northwestern Australia. All investigated products mostly capture the considered drought events. Overall, the events tend to be least pronounced in the ACTIVE remote-sensing product across all drought metrics, particularly with respect to the magnitudes. Furthermore, MERRA-2 shows lower drought magnitudes than the other products, in both the surface layer and the root zone. The COMBINED remote-sensing products (surface and root zone soil moisture dataset) display partly stronger drought severities than the other products. In the root zone, the droughts are dampened with respect to the magnitude and smaller with respect to the spatial extent than in the surface layer, but they show a tendency toward prolonged durations and stronger severities. The product differences in the magnitude and severity of the drought events are consistent with the differences in soil moisture trends, which demonstrates that the representation of soil moisture trends plays a fundamental role in the drought-detection capacity of the different products.

Association of spring thermal forcing anomalies in the Tibetan Plateau with dust aerosol changes over the Taklamakan Desert.

The Tibetan Plateau (TP) is significantly influencing the climate and environmental evolution regionally and globally. Adjacent to the northwestern TP, the Taklimakan Desert (TD) experiences the unique pattern of dust aerosol variations due to the deep basin terrain. However, systematic studies on how TP climate change affects TD dust aerosol variations are lacking. This study employs MERRA-2 and ERA5 reanalysis data (1991-2020) to investigate the impact of springtime TP thermal forcing on TD dust aerosol. Results indicate that the interannual variation of dust column mass densities over the TD showed an increasing trend over 1991-2020, with a significant increase in the northern TD, where the maximum value exceeded 9.51mgm-2 per year. Thermal forcing strengthening on the northwestern TP are positively correlated with dust column mass density variations in the northern TD, while negatively correlated in the southern TD. Further comprehensive analyses indicate that strong northwestern TP thermal forcing has increased dust aerosol concentrations in northern TD, with a maximum 14mgm-3 increase at 700hPa, while decreasing them in the southern part. During strong SNTP-Q1 years, anticyclonic anomalies, strong updrafts, and the blocking effect of the Tian Shan Mountains, combined with high temperatures, result in dust aerosols suspended in the atmosphere over the northern TD. In contrast, during weak SNTP-Q1 years, cyclonic anomalies, strong downdrafts, and the large topography of the TP, along with inversion temperatures, contribute to dust aerosol accumulation in the southern TD. This study elucidates the mechanism of TP thermal forcing on TD dust aerosol variations, enhancing understanding of its effects on environmental and climatic changes in the TD and Central Asia.

An Investigation of the SOCOLv4 Model’s Suitability for Predicting the Future Evolution of the Total Column Ozone

The anthropogenic impact on the ozone layer is expressed in anomalies in the total ozone content (TOC) on a global scale, with periodic enhancements observed in high-latitude areas. In addition, there are significant variations in TOC time trends at different latitudes and seasons. The reliability of the TOC future trends projections using climate chemistry models must be constantly monitored and improved, exploiting comparisons against available measurements. In this study, the ability of the Earth’s system model SOCOLv4.0 to predict TOC is evaluated by using more than 40 years of satellite measurements and meteorological reanalysis data. In general, the model overpredicts TOC in the Northern Hemisphere (by up to 16 DU) and significantly underpredicts it in the South Pole region (by up to 28 DU). The worst agreement was found in both polar regions, while the best was in the tropics (the mean difference constitutes 4.2 DU). The correlation between monthly means is in the range of 0.75–0.92. The SOCOLv4 model significantly overestimates air temperature above 1 hPa relative to MERRA2 and ERA5 reanalysis (by 10–20 K), particularly during polar nights, which may be one of the reasons for the inaccuracies in the simulation of polar ozone anomalies by the model. It is proposed that the SOCOLv4 model can be used for future projections of TOC under the changing scenarios of human activities.

Role of QBO and MJO in Sudden Stratospheric Warmings: A Case Study

The impact of the quasi-biennial oscillation (QBO) and Madden–Julian oscillation (MJO) on the dynamics of major sudden stratospheric warmings (SSWs) observed in the winters of 2018, 2019, and 2021 is investigated. Using data from the MERRA-2 reanalysis, we analyze the daily mean variability of critical atmospheric parameters at the 10 hPa level, including zonal mean polar cap temperature, zonal mean zonal wind, and the amplitudes of planetary waves 1 and 2. The results reveal dramatic increases in polar cap temperature and significant wind reversals during the SSW events, particularly in 2018. The analysis of planetary wave (PW) amplitudes demonstrates intensified wave activity coinciding with the onset of SSWs, underscoring the pivotal role of PWs in these stratospheric disruptions. Further examination of outgoing long-wave radiation (OLR) anomalies highlights the influence of QBO phases on tropical convection patterns. During westerly QBO (w-QBO) phases, enhanced convective activity is observed in the western Pacific, whereas the easterly QBO (e-QBO) phase shifts convection patterns to the maritime continent and central Pacific. This modulation by QBO phases influences the MJO’s role during SSWs, affecting tropical and extra-tropical weather patterns. The day-altitude variability of upward heat flux reveals distinct spatiotemporal patterns, with pronounced warming in the polar regions and mixed heat flux patterns in low latitudes. The differences observed between the SSWs of 2017–2018 and 2018–2019 are likely related to the varying QBO phases, emphasizing the complexity of heat flux dynamics during these events. The northern annular mode (NAM) index analysis shows varied responses to SSWs, with stronger negative anomalies observed during the e-QBO phase compared to the w-QBO phases. This variability highlights the significant role of the QBO in shaping the stratospheric and tropospheric responses to SSWs, impacting surface weather patterns and the persistence of stratospheric anomalies. Overall, the study demonstrates the intricate interactions between stratospheric dynamics, QBO, and MJO during major SSW events, providing insights into the broader implications of these atmospheric phenomena on global weather patterns.

Long-term distribution and evolution trends of absorption aerosol optical depth with different chemical components in global and typical regions

Different types of atmospheric aerosols have different climatic effects. In this study, MERRA-2 reanalysis data of absorption aerosol optical depth (AAOD) products at 550 nm from 1980 to 2018 were used to analyse the long-term distribution characteristics and evolution trends of the AAOD of different chemical components globally and in 12 typical study areas. We also analysed the seasonal and interannual monthly variations of the different chemical components of AAOD. In the 40-year study period from 1980 to 2018, the maximum value of total AAOD (TAAOD) appears in the southern regions of SD (Sahara Desert), CSA (Central Southern Africa), NC (Northern China), SC (Southern China), and SEA (Southeastern Asia) (> 0.040). The highest value of dust AAOD (DUAAOD) is in SD (0.030–0.040), and the contribution rate reaches 80 %; while in SC, SEA, and AMZ, black carbon AAOD (BCAAOD) contributes 80 %–90 %. The high-value area of DUAAOD in SD-ME-NWC expands in spring, and the dust belt formed in summer results in a larger DUAAOD (> 0.050). The proportion of BCAAOD in autumn and winter is larger in the dust belt, which is another major contributor to AAOD in this region. The monthly distributions of TAAOD in SEA, CSA, NC, and AMZ are mainly affected by biomass combustion, while the DU in ME (Middle East), NWC (Northwestern China), and SD has a greater effect on AAOD, and the TAAOD in NEA (Northeastern Asia), WEU (Western Europe), EUS (Eastern United States), SC, SA (Southern Asia), and other regions is mainly affected by both DU and BC + OC (in which OC refers to organic carbon). The interannual variations of BCAAOD and OCAAOD tend to be flat before 2000, and then show an increasing trend. BCAAOD has the largest relative contribution (at about 60 %), followed by DUAAOD (at about 30 %), and then OCAAOD has the smallest contribution (at less than 10 %). From a global perspective, AAOD shows different increasing trends during 1980–2018, 1980–1992, and 1993–2005, and decreases or even completely reverses during 2006–2018. This paper provides the distribution characteristics and evolutionary trends of different chemical components of AAOD, which can improve scientific understanding of global- and regional-scale aerosols and their climatic effects.

A multi-method and multi-duration trend analysis of temperature and precipitation in Istanbul, Turkey, by using meteorological records, MERRA-2 reanalysis, and IMERG estimations

In this study, temperature and precipitation trends in Istanbul were analyzed using data from a ground-based station for the 1960–2019 interval, Integrated Multi-SatellitE Retrievals for Global Precipitation Measurement (GPM IMERG) for 2000–2019, and Modern-Era Retrospective Analysis for Research and Applications (MERRA-2) temperature reanalysis for 1980–2019. Single-duration analysis revealed significant increasing temperature trends with minor downward and major upward tendencies in precipitation. Multi-duration analysis resulted with dominant decreasing and increasing temperature trends, respectively, in 1980–1999 and 2000–2019, alongside decreasing precipitation trends in 1980–1999 winter and 2000–2019 spring, which were undetected in the single-duration analysis. As a novel attempt, trends based on MERRA-2 and IMERG data were compared with ground-based records in Istanbul and revealed better compatibility in MERRA-2, especially in 2000–2019 compared to IMERG data. By comparing various statistical methods and data sources, this study offers valuable insights into hydrometeorological trend analysis and aids transition from conventional to novel data sources.

How Skilful Are Cloud Cover Products in Representing Observed Cloudiness in Québec?

ABSTRACT In this study, we performed, for the first time, a detailed analysis of cloudiness in Québec using station based observations and evaluated the trustworthiness of various satellite and reanalysis based cloud cover products. We found only 12 stations with observational time series long enough for providing robust analysis due to various issues related to observing methods and recording of information during the 1990s. Our results showed that Québec can be fairly cloudy throughout the year with annual mean cloud cover fraction ranging between 0.5 and 0.7 with autumn being the cloudiest season at most station locations. We also found a significantly increasing trend in cloudiness in the winter season over Québec during 1982–2012 (∼2–6% per decade depending on the location). Among the cloudiness products evaluated in this study, a satellite based cloudiness product (i.e. based on AVHRR) performed best in representing the observed cloudiness over Québec including the wintertime increasing trend. Two reanalysis based cloudiness products (i.e. ERA5 and NARR) also performed well in representing the observed cloudiness indicated by high correlations, relatively lower mean biases and smaller root mean square errors. However, the reanalysis products did not capture well the observed seasonal trends. All products showed relatively larger biases in the winter season except JRA-25 reanalysis product. During the warmer months, however, JRA-25 showed largest biases followed by MERRA-2 reanalysis product. JRA-25 and MERRA-2 also had lower correlation for annual and winter means and highest root mean square errors for all seasons. Furthermore, based on two skill scores, we found that the ERA5, AVHRR and NARR performed relatively better than JRA-25 and MERRA-2. Based on these results we conclude that the satellite product and the two reanalysis products can be useful resources (as observational proxies) along with station based observations for understanding long term changes in cloudiness of this region and also potentially evaluating regional climate model simulations.

Spatiotemporal evolution of dust over Tarim Basin under continuous clear-sky

The unique terrain and complex atmospheric boundary layer (ABL) processes result in a distinctive spatiotemporal distribution of dust in the Tarim Basin; however, this distribution remains unclear under continuous clear-sky conditions. In this study, 382 cases were selected to investigate the spatiotemporal evolution of dust and its potential mechanisms based on MERRA-2 and ERA5 reanalysis datasets combined with MODIS satellite observations during the warm seasons from 2000 to 2023. Taking the typical case of a completely cloudless on July 24–27, 2016, the dust aerosol optical depth (DAOD) at the margin of the Tarim Basin increased with time. The climatological characteristics showed a high DAOD in the northern, western, and southwestern regions and a relatively low DAOD in the central area. Nocturnal low-level jets dominated by northeasterly winds enhance the low-level westward airflow in weak anticyclonic systems, causing dust accumulation in the west and north of the basin. Vertical mixing within the ABL during the daytime increases dust loading in the residual layer, and these dust particles can ascend to high altitudes after breaking through the ABL by the vertical circulation. The dust loading at the lower level during the daytime was higher than that at night, whereas the opposite was true for the upper level. The downward airflow in the northwest slope of the Tibetan plateau weakens at night, leading to dust being uplifted to higher altitudes and transported outside the Tarim Basin by the westerlies. These results enhance our understanding of dust distribution and related mechanisms in Tarim Basin and support the development and utilization of climatic resources in this region.

Modeling the contribution of leads to sea spray aerosol in the high Arctic

Abstract. Elongated open-water areas in sea ice (leads) release sea spray particles to the atmosphere. However, there is limited knowledge on the amount, properties and drivers of sea spray emitted from leads, and no existing parameterization of this process is available for use in models. In this work, we use measurements of aerosol fluxes from Nilsson et al. (2001) to produce an estimate of the location, timing and amount of sea spray emissions from leads at the scale of the Arctic Ocean for 1 year. Lead fractions are derived using sea ice data sets from numerical models and satellite detection. The proposed parameterization estimates that leads account for 0.3 %–9.8 % of the annual sea salt aerosol number emissions in the Arctic Ocean regions where sea ice concentration is greater than 80 %. Assuming similar size distributions to those from emissions from the open ocean, leads account for 30 %–85 % of mass emissions in sea ice regions. The total annual mass of sea salt emitted from leads, 0.1–2.1 Tg yr−1, is comparable to the mass of sea salt aerosol transported above sea ice from the open ocean, according to the MERRA-2 reanalysis. In addition to providing the first estimates of possible upper and lower bounds of sea spray emissions from leads, the conceptual model developed in this work is implemented and tested in the regional atmospheric chemistry model WRF-Chem. Given the estimates obtained in this work, the impact of sea spray from leads on Arctic clouds and radiative budget needs to be further explored.

Artificial intelligence for predicting arctic permafrost and active layer temperatures along the Alaskan North Slope

Climate change in Arctic regions poses a threat to permafrost stability, potentially leading to issues such as infrastructure damage, altered ecosystems, and increased greenhouse gas emissions. The challenge of monitoring permafrost temperatures and identifying critical environmental factors is accentuated by the scarcity of subsurface thermal data in remote Arctic locales. To alleviate this lack of soil thermal data we set out to combine in situ observed soil temperatures with MERRA-2 reanalysis data and Machine Learning (ML) to develop local and season-specific subsurface soil temperature predictions in Alaska. First, reanalysis features are selected based on surface energy budget physics. Coupled with Julian calendar dates, reanalysis features are preprocessed and then fed to the ML prediction model. We conducted a series of computational experiments and tested the results against performance benchmarks to determine the optimal prediction models, length of look-back period of model inputs, and the training set size. Six conventional ML models (e.g., GBDT, RF, SVR) and five statistical baselines (e.g. ARIMA) using in situ time series of soil temperature data ranging from 0 - 1 m depth are considered for the prediction task. The models are trained using in situ soil temperatures at depths between 0 and 1 m, spanning 16+ years, from field sites at Deadhorse and Toolik Lake, Alaska. All prediction performances are assessed using root mean squared error (RMSE) and mean absolute error (MAE). Results show that locally trained ML models can estimate shallow soil temperatures with an average error of RMSE=1.308∘\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$RMSE=1.308^{\\circ }$$\\end{document}C.

Modeling of Biologically Effective Daily Radiant Exposures over Europe from Space Using SEVIRI Measurements and MERRA-2 Reanalysis

Ultraviolet solar radiation at the Earth’s surface significantly impacts both human health and ecosystems. A biologically effective daily radiant exposure (BEDRE) model is proposed for various biological processes with an analytical formula for its action spectrum. The following processes are considered: erythema formation, previtamin D3 synthesis, psoriasis clearance, and inactivation of SARS-CoV-2 virions. The BEDRE model is constructed by multiplying the synthetic BEDRE value under cloudless conditions by a cloud modification factor (CMF) parameterizing the attenuation of radiation via clouds. The CMF is an empirical function of the solar zenith angle (SZA) at midday and the daily clearness index from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) measurements on board the second-generation Meteosat satellites. Total column ozone, from MERRA-2 reanalysis, is used in calculations of clear-sky BEDRE values. The proposed model was trained and validated using data from several European ground-based spectrophotometers and biometers for the periods 2014–2023 and 2004–2013, respectively. The model provides reliable estimates of BEDRE for all biological processes considered. Under snow-free conditions and SZA < 45° at midday, bias and standard deviation of observation-model differences are approximately ±5% and 15%, respectively. The BEDRE model can be used as an initial validation tool for ground-based UV data.

On the Biases of MERRA-2 Reanalysis and Ground-Based Measurements of Black Carbon Aerosols Over India

The precise measurement of black carbon (BC) and assessment of involved biases are very important considering its serious health and climatic effects. MERRA-2 reanalysis data (modern-era retrospective analysis for research and applications, version 2) of surface black carbon (SBC) and ground measurements by optical photometers (mostly Aethalometer) have been used widely to study the temporal and spatial variation of BC aerosols over India. However, a significant bias (more than 70%) between MERRA-2 SBC concentration and ground measurements of BC (by Aethalometer, referred as MBC (AE)) was observed by Malik and Aggarwal, (2021). In the current study, the possible reasons for this bias are further evaluated by comparing the MERRA-2 SBC and Aethalometer measurements against a reference photometer, i.e., continuous soot monitoring system (COSMOS). Three-year (2020-2022) SBC concentrations of MERRA-2 reanalysis data are compared against simultaneous measurements of BC by COSMOS (MBC (COSMOS)). Additionally, the biases in ground measurements are further investigated by comparing one-month (March 2023) observations of Aethalometer (MBC (AE)). These comparison results revealed a 40% underestimation in MERRA-2 reanalysis during 2020-2022. The higher underestimation (50%) of SBC during increased biomass emissions months (October to May), in contrast to the 25% underestimation during conventional emissions months (June to September), emphasizes the underestimation of biomass emissions inputs in MERRA-2 SBC reanalysis. The initial large bias between raw data of Aethalometer (MBC (AE-raw) = 1.92-1.98 times MBC (COSMOS)) get reduced significantly when correction due to multiple scattering in the filter media and mass absorption cross-section (MAC) are incorporated to correct the Aethalometer data (MBC (AE-corr) = 1.15-1.18 times MBC (COSMOS)). The findings of this study underscore the need to harmonize the correction methods for multiple scattering and redefine the MAC values for Aethalometer measurements as well as updating the anthropogenic emission inventories in MERRA-2 SBC reanalysis.

Post-process correction improves the accuracy of satellite PM2.5 retrievals

Abstract. Estimates of PM2.5 levels are crucial for monitoring air quality and studying the epidemiological impact of air quality on the population. Currently, the most precise measurements of PM2.5 are obtained from ground stations, resulting in limited spatial coverage. In this study, we consider satellite-based PM2.5 retrieval, which involves conversion of high-resolution satellite retrieval of aerosol optical depth (AOD) into high-resolution PM2.5 retrieval. To improve the accuracy of the AOD-to-PM2.5 conversion, we employ the machine-learning-based post-process correction to correct the AOD-to-PM conversion ratio derived from Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2) reanalysis model data. The post-process-correction approach utilizes a fusion and downscaling of satellite observation and retrieval data, MERRA-2 reanalysis data, various high-resolution geographical indicators, meteorological data, and ground station observations for learning a predictor for the approximation error in the AOD-to-PM2.5 conversion ratio. The corrected conversion ratio is then applied to estimate PM2.5 levels given the high-resolution satellite AOD retrieval data derived from Sentinel-3 observations. The region of study is central Europe during the year 2019. Our model produces PM2.5 estimates with a spatial resolution of 100 m at satellite overpass times with R2 = 0.55 and RMSE = 6.2 µg m−3. The corresponding metrics for monthly averages are R2 = 0.72 and RMSE = 3.7 µg m−3. Additionally, we have incorporated an ensemble of neural networks to provide error envelopes for machine-learning-related uncertainty in the PM2.5 estimates. The proposed approach can produce accurate high-resolution PM2.5 data that can be very useful for air quality monitoring, emission regulation, and epidemiological studies.

Effect of PM2.5 on burden of mortality from non-communicable diseases in northern Thailand.

Particulate pollution, especially PM2.5from biomass burning, affects public and human health in northern Thailand during the dry season. Therefore, PM2.5exposure increases non-communicable disease incidence and mortality. This study examined the relationship between PM2.5and NCD mortality, including heart disease, hypertension, chronic lung disease, stroke, and diabetes, in northern Thailand during 2017-2021. The analysis utilized accurate PM2.5data from the MERRA2 reanalysis, along with ground-based PM2.5measurements from the Pollution Control Department and mortality data from the Division of Non-Communicable Disease, Thailand. The cross-correlation and spearman coefficient were utilized for the time-lag, and direction of the relationship between PM2.5and mortality from NCDs, respectively. The Hazard Quotient (HQ) was used to quantify the health risk of PM2.5to people in northern Thailand. High PM2.5 risk was observed in March, with peak PM2.5concentration reaching 100 µg/m3, with maximum HQ values of 1.78±0.13 to 4.25±0.35 and 1.45±0.11 to 3.46±0.29 for males and females, respectively. Hypertension significantly correlated with PM2.5levels, followed by chronic lung disease and diabetes. The cross-correlation analysis showed a strong relationship between hypertansion mortality and PM2.5at a two-year time lag in Chiang Mai (0.73) (CI [-0.43-0.98], p-value of 0.0270) and a modest relationship with chronic lung disease at Lampang (0.33) (a four-year time lag). The results from spearman correlation analysis showed that PM2.5concentrations were associated with diabetes mortality in Chiang Mai, with a coefficient of 0.9 (CI [0.09-0.99], p-value of 0.03704). Lampang and Phayao had significant associations between PM2.5 and heart disease, with coefficients of 0.97 (CI [0.66-0.99], p-value of 0.0048) and 0.90 (CI [0.09-0.99], p-value of 0.0374), respectively, whereas Phrae had a high coefficient of 0.99 on stroke.

Spatial variation, sources, and trajectory of black carbon in the South Sumatra Region of Indonesia using MERRA-2 reanalysis data

Spatial variation, sources, and trajectory of black carbon in the South Sumatra Region of Indonesia using MERRA-2 reanalysis data

Consequences of weakening of dynamic barrier of the Arctic polar vortex

The dynamic barrier is one of the physical characteristics of the polar vortices; it prevents subpolar air masses from penetrating into the vortex and contributes to a temperature decrease inside the vortex in the lower stratosphere. In the presence of a dynamic barrier in winter, chlorine compounds involved in the ozone destruction cycle accumulate on particles of polar stratospheric clouds (PSCs) and heterogeneous reactions occur with the formation of molecular chlorine, and with the appearance of solar radiation over the polar region, photochemical reactions begin, leading to large-scale ozone depletion. When the dynamic barrier is weakened in winter, the temperature inside the vortex rises, PSC melts and, thus, the accumulation of chlorine cycle reagents on PSC is interrupted. We proposed dividing the Arctic polar vortex dynamics into 3 types according to the consequences: (1) the strong vortex, whose activity results in ozone depletion, (2) the weak vortex with breakdown in winter, marked by a sudden stratospheric warming, and (3) the stable vortex with an episode (episodes) weakening of the dynamic barrier in winter without ozone depletion in the period from late winter to spring. We have for the first time proposed a characteristic of the dynamic barrier of the polar vortex at all pressure levels from 100 to 1 hPa and described the consequences of its weakening. Using the vortex delineation method based on the data of the ERA5 and MERRA-2 reanalyses, we showed that in all cases when the polar ozone depletion was not recorded from late winter to spring under the conditions of the stable polar vortex, the dynamic barrier weakening and PSС melting was observed in midwinter.

Reconstructing 10-km-resolution direct normal irradiance dataset through a hybrid algorithm

Evaluating, mapping, and monitoring high-quality Direct Normal Irradiance (DNI) is crucial for the design, financing, and operation of solar power plants, especially those utilizing concentrating technologies. In China, limited number of first-class meteorological stations provide DNI observations. To bridge this gap, this study has reconstructed a comprehensive 41-year (1982–2022) daily DNI dataset for China (CHDNI). This dataset integrated CMA ground-based observations with ERA5 and MERRA-2 reanalysis data, employing a spatial resolution of 10 km. The data was processed using the REST2 radiation transfer model combined with Stacking ensemble machine learning. Validation with ground-based measurements indicated that the Stacking model exhibited high and consistent performance. This was evidenced by its sample-based cross-validation results, showing a correlation coefficient (R) of 0.92, a root-mean-square error (RMSE) of 32.04 W/m2, a mean absolute error (MAE) of 23.68 W/m2, and a global performance indicator of 3.86. Among the four analyzed DNI products (CERES, ERA5, SWGDN, and CHDNI), CHDNI demonstrated the closet alignment with ground observations, evidenced by the least deviation (MAE = 21.67 W/m2 and RMSE = 31.00 W/m2) and the highest R value (0.92). Aerosol optical depth and cloud cover emerged as the primary factors influencing the quality and accuracy of DNI products under clear-sky and total-sky conditions. The spatial distribution of DNI across China is complex, showcasing DNI values ranging from 25.82 W/m2 to 194.22 W/m2, with an average value of 98.08 W/m2 over 41-year. This dataset is a valuable resource for analyzing regional climate change, photovoltaic applications, and solar energy resources assessment.

Prediction of Tropical Pacific Rain Rates with Overparameterized Neural Networks

Abstract The prediction of tropical rain rates from atmospheric profiles poses significant challenges, mainly due to the heavy-tailed distribution exhibited by tropical rainfall. This study introduces overparameterized neural networks not only to forecast tropical rain rates but also to explain their heavy-tailed distribution. The investigation is separately conducted for three rain types (stratiform, deep convective, and shallow convective) observed by the Global Precipitation Measurement satellite radar over the west and east Pacific regions. Atmospheric profiles of humidity, temperature, and zonal and meridional winds from the MERRA-2 reanalysis are considered as features. Although overparameterized neural networks are well known for their “double descent phenomenon,” little has been explored about their applicability to climate data and capability of capturing the tail behavior of data. In our results, overparameterized neural networks accurately estimate the rain-rate distributions and outperform other machine learning methods. Spatial maps show that overparameterized neural networks also successfully describe the spatial patterns of each rain type across the tropical Pacific. In addition, we assess the feature importance for each overparameterized neural network to provide insight into the key factors driving the predictions, with low-level humidity and temperature variables being the overall most important. These findings highlight the capability of overparameterized neural networks in predicting the distribution of the rain rate and explaining extreme values. Significance Statement This study aims to introduce the capability of overparameterized neural networks, a type of neural network with more parameters than data points, in predicting the distribution of tropical rain rates from gridscale environmental variables and explaining their tail behavior. Rainfall prediction has been a topic of importance, yet it remains a challenging problem for its heavy-tailed nature. Overparameterized neural networks correctly captured rain-rate distributions and the spatial patterns and heterogeneity of the observed rain rates for multiple rain types, which could not be achieved by any other previous statistical or machine learning frameworks. We find that overparameterized neural networks can play a key role in general prediction tasks, with potential expanded applicability to other domains with heavy-tailed data distribution.