Pinpointing the contribution of sources in complex urban areas, affected by large point sources such as oil refineries, is important for developing emission control strategies. Receptor models based ...on the chemical composition of particulate matter (PM), such as chemical mass balance (CMB) and positive matrix factorization (PMF), are useful means for source apportionment, but the inclusion of other gaseous pollutants need further consideration. The results of the multipollutant analyses using temporal variations in pollutant concentrations, chemical PM speciation and receptor modeling, PMF and conditional bivariate polar plots (CBPF), were used for determination of major pollutant sources of fine particulate matter (PM2.5) and less represented pollutants – hydrogen sulfide (H2S), nitrogen dioxide (NO2) and sulfur dioxide (SO2) in an urban area in Slavonski Brod, Croatia influenced by a large point source (an oil refinery) in Brod, Bosnia and Herzegovina. It is found that the composition of PM2.5 is dominated by carbonaceous combustion particles, mainly organic carbon (OC), with maximum values appearing during winter. Summer PM2.5 levels were dominated by sulfate and ammonium, which can be related to the industrial activities i.e., oil refinery. According to PMF analysis, the majority of OC is coming from biomass burning with ∼50% contribution to observed species concentration followed by ∼30% from industry/refinery and ∼10% from traffic. CPBF model showed that urban and highway traffic was the main source of NO2 concentrations while oil refinery was identified as the dominant source of SO2 and H2S. The CBPF receptor model combines concentrations of pollutants and meteorological parameters and emerged as a reliable complementary tool for the identification of sources for considered gaseous pollutants. Limitations of the CBPF method are in the application in stable atmospheric boundary layer conditions (SABL) as wind direction is not representative. Also, larger uncertainty is related to the representation of peak concentrations transported with higher wind speeds (>8 m/s) due to the lower number of events. This work uses various source apportionment methods in the assessment of PM but also for gaseous pollutants, such as NO2, SO2 and H2S that are less represented in the source apportionment studies and can be used for future scientific applications to assure more efficient air quality management.
•CBPF emerged as a reliable tool for source apportionment of gaseous pollutants and PM.•PMF is an integral and practical tool to determine pollution sources for PM.•An oil refinery was the dominant source of SO2 and H2S.•Emissions from city traffic and the highway were major NO2 sources.•Peak PM2.5 and NO2 values were observed during stable atmospheric conditions.
The Split wildfire in July 2017, which was one of the most severe wildfires in the history of this Croatian World Heritage Site, is the focus in this study. The Split fire is a good example of a ...wildfire–urban interface, with unexpected fire behavior including rapid downslope spread to the coastal populated area. This study clarifies the meteorological conditions behind the fire event, those that have limited the effectiveness of firefighting operations, and the rapid escalation and expansion of the fire zones within 30 h. The Split fire propagation was first reconstructed using radio logs, interviews with firefighters and pilots involved in the intervention, eyewitness statements, digital photographs from fire detection cameras, media, and the monthly firefighting journal. Four phases of fire development have been identified. Then, weather observations and numerical simulations using an
enhanced-resolution operational model are utilized to analyze the dynamics
in each phase of the fire runs. The synoptic background of the event
includes large surface pressure gradient between the Azores anticyclone
accompanied by a cold front and a cyclone over the southeastern Balkan Peninsula. At the upper level, there was a deep shortwave trough extending from the Baltic Sea to the Adriatic Sea, which developed into a cut-off low. Such synoptic conditions have resulted in the maximum fire weather index in 2017. Combined with topography, they also locally provoke the formation of the strong northeasterly bura wind along the Adriatic coast, which has been accompanied by a low-level jet (LLJ). The bura (downslope wind), with mid- to low-level gravity-wave breaking and turbulence mixing (as in the hydraulic jump theory), also facilitated the subsidence of dry air from the upper troposphere and rapid drying at the surface. This study demonstrates that numerical guidance that indicates the
spatial and temporal occurrence of a LLJ is highly capable of explaining the Split fire evolution from the ignition potential to its extinguishment stage. Thus, in addition to the conventional fire weather indices, such products are able to improve fire weather behavior forecasting and in general more effective decision-making in fire management.
Since changes in temperature and precipitation have different effects on (a) all developmental stages of grapevines in most of the wine regions worldwide (i.e., on their phenological characteristics) ...and (b) different varieties, a comprehensive database of bioclimatic indices has been calculated and analysed for Croatian wine producing regions. The database consists of the average growing season temperature, growing degree‐days, Huglin index, dryness index and cool night index that are based on all available meteorological measurements as well as the outputs of regional climate models (RCMs) from the EURO‐CORDEX database. The horizontal grid spacing of 0.11° from the RCM ensembles enabled a fine‐scale determination of bioclimatic indices for the present and future climate in Croatia. In addition, statistical analyses (standard statistical parameters and Bayesian method) were carried out to examine trends in sugar content, total acidity and date of harvest. Calculations were performed for the present and future climate on the basis of data from seven selected vineyards/wineries and four varieties (‘Graševina’, ‘Plavac mali’, ‘Chardonnay’ and ‘Merlot’). The results show whether the part of Croatia that is suitable for grape cultivation in the present climate will continue to be favourable in the future within the Mediterranean area. In general, projections suggest further warming and drying of the climate in Croatia and an earlier harvest, with some variations among varieties that show latitude dependence. Projections for the future climate also suggest that the existing viticultural zoning will be much less adequate for the Croatian territory because it reduces the economically sustainable production of wine in certain areas.
Spatial distributions of temperature‐based bioclimatic indices (e.g., growing degree‐days GDD) for high‐end climate change scenarios are pointing to further warming in the period 2041–2070. Differences between the two periods P2 (2041–2070) and P0 (1971–2000) clearly show that some indices in certain regions will likely cross into higher (warmer) classes within the index scale. All grape varieties examined indicated an increase in the number of earlier harvests and a reduction in the number of later harvests, regardless of location.
The uncharacteristically extreme outbreak of particulate matter took place over the Balkan region from 27 to 30 March 2020. Observations at air quality stations in Croatia recorded hourly PM10 ...concentrations up to 412 μgm−3. The meteorological analysis shows that the increase in PM10 concentrations was primarily due to the advection of dust from the deserts east of the Caspian Sea. The anticyclone north of Croatia and the cyclone over Anatolia formed a strong pressure gradient driving transport from the east. Both back trajectories and satellite products pointed to the dry Aral Sea as the main source of dust. A dust plume influenced the PM10 increase observed in Croatia, starting at the easternmost air quality stations. The modeling study shows that the vertical extent of the dust plume was up to ∼2 km. However, the chemical and morphological (scanning electron microscope analysis) composition of PM10 at the sites in the northeastern Adriatic Sea showed mainly the presence of Saharan dust. Prior to the advection of the Asian dust, the transport of Saharan dust, driven by Sharav cyclone, was observed in the PM10 values at several stations in the Adriatic Sea and on the Croatian mainland on 26 March 2020. Modeling results showed that the Saharan dust transport occurred at altitudes below ∼8 km. The mixing of the Asian and Saharan dust plumes over the Balkans was favored by the subsidence due to anticyclonic high‐pressure conditions and is the most likely explanation for the observed PM chemical and morphological results.
Plain Language Summary
The event of extreme air pollution in Croatia occurred at the end of March 2020. Exceptionally high aerosol concentrations were observed at several air quality stations. The outbreak was studied using a numerical chemical transport model. Chemical analysis was also performed for the aerosol sample and the sample was viewed with an electronic microscope. The analysis revealed that the composition of the aerosol was due to desert dust transported by air masses from the deserts east of the Caspian Sea and the Sahara Desert.
Key Points
The extraordinary dust episode over the Balkan region was simulated with the WRF‐Chem model
The dried Aral Sea was identified as the main source, but the dust plume from the Sahara was also present
Chemical and morphological analysis of the PM10 sample shows that the dust originated from the Sahara Desert
Over the past few decades the horizontal resolution of regional climate models (RCMs) has steadily increased, leading to a better representation of small-scale topographic features and more details ...in simulating dynamical aspects, especially in coastal regions and over complex terrain. Due to its complex terrain, the broader Adriatic region represents a major challenge to state-of-the-art RCMs in simulating local wind systems realistically. The objective of this study is to identify the added value in near-surface wind due to the refined grid spacing of RCMs. For this purpose, we use a multi-model ensemble composed of CORDEX regional climate simulations at 0.11° and 0.44° grid spacing, forced by the ERA-Interim reanalysis, a COSMO convection-parameterizing simulation at 0.11° and a COSMO convection-resolving simulation at 0.02° grid spacing. Surface station observations from this region and satellite QuikSCAT data over the Adriatic Sea have been compared against daily output obtained from the available simulations. Both day-to-day wind and its frequency distribution are examined. The results indicate that the 0.44° RCMs rarely outperform ERA-Interim reanalysis, while the performance of the high-resolution simulations surpasses that of ERA-Interim. We also disclose that refining the grid spacing to a few km is needed to properly capture the small-scale wind systems. Finally, we show that the simulations frequently yield the accurate angle of local wind regimes, such as for the Bora flow, but overestimate the associated wind magnitude. Finally, spectral analysis shows good agreement between measurements and simulations, indicating the correct temporal variability of the wind speed.
•GDD models proved to be slightly better than multiple linear regression models.•GDD models with base temperature 5 °C proved to be the best for budburst.•GDD with base temperature of 10 °C proved to ...be the best for other stages.•These models could be coupled with climate models for predction in future changes.
In recent decades, there have been significant changes in temperature and precipitation, as well as in the occurrence of phenological stages of the grapevine in most wine-growing regions around the world. These changes are not the same for each variety, nor in all locations. Due to the vulnerability of the viticulture sector, including the possible losses in production in the current winegrowing areas, as well as the planting of vineyards in new areas, it is of great importance to examine the trends in the occurrence of individual stages and to link them as successfully as possible with changes in meteorological parameters. The simplest approach to this is using agrometeorological indices (e.g., Growing degree day, GDD) which can determine the possibility of growing a certain variety. There is also the possibility of developing and testing simple statistical phenological models that serve to predict the occurrence of phenological stages. Four such models were tested for the prediction of four phenological stages (budburst, flowering, veraison, and harvest) for four grape varieties ('Graševina', 'Chardonnay', 'Merlot', and 'Plavac mali') in Croatia. The first two models are commonly used GDD models with a temperature base of 10 °C or 5 °C, and thresholds necessary for phenological stage to start depending on variety or variety and location. The other two models are based on the determination of the best multi-linear regression using as predictors monthly and multi-month averages of minimum temperature, maximum temperature, mean temperature, and total precipitation. The increase in temperature index values from the 1990s to today is particularly significant. Statistical phenological models also proved to be a good indicator of the occurrence of individual phenological stages. GDD models proved to be somewhat better in prediction, GDD models that use a temperature of 5 °C as a base proved to be better for predicting budburst, those that use a base of 10 °C proved to be better for the other stages and particularly for flowering (with agreement index d up to 0.8 and root mean square error of prediction RMSE from 5 to 10 days). Linear regression that uses temperature as a predictor and the same equation regardless of location proved to be very good in predicting the harvest of autochthonous varieties ('Graševina' and 'Plavac mali') with low RMSE (up to 10 days). The presented results indicate that these models could be applied to future scenarios and with that help to make decisions in the wine sector in Croatia and worldwide.
The main goal of this study is to present a recently developed classification method for weather types based on the vorticity and the location of the synoptic centers relative to the Adriatic region. ...The basis of the present objective classification, applied to the Adriatic region, is the subjective classification developed by Poje. Our algorithm considered daily mean sea-level pressure and 500 hPa geopotential height to define one out of 17 possible weather types. We applied the algorithm to identify which weather type was relevant in the generation of the two typical near-surface winds over the Adriatic region, namely Bora and Sirocco. Two high-resolution (0.11°) EURO-CORDEX regional climate models were used, SMHI-RCA4 and DHMZ-RegCM4, forced by several CMIP5 global climate models and analyzed for two 30-year periods: near-present day and mid-21st century climate conditions under the high-end Representative Concentration Pathway (RCP8.5) scenario. Bora and Sirocco days were extracted for each weather type and a distribution over the 30-year period was presented. Our results suggest that in the winter season, climate model projections indicate a reduction in the main cyclonic types relevant in the formation of Bora over the entire Adriatic region and an increase in the number of anticyclonic types relevant in Sirocco events. In contrast, for the summer season, an increase in the main anticyclonic Bora-related weather types is found in the ensemble over the northern Adriatic region.
•Minimum temperature in meteorological shelter that best describes frost is 2.5 °C.•Condition that the dew point temperature is less than 0 °C reduces false alarms.•Tmin threshold of 3 °C and Td of ...0 °C gave the best results for frost probability.•These method could be coupled with climate models for frost prediction in future.
Due to the earlier start of phenological cycles among fruit trees, frost represents one of the most notable hazards for agriculture. There is no unique method for forecasting frost, and different methods for describing frost under present and future climate conditions can be found in the literature. Often these methods are applied in a certain area without prior control. Five such frost detection methods were assessed in Croatia. In addition, five new frost estimation methods that rely on measurements of the daily minimum temperature (Tmin) and dew point temperature (Td), calculated using Tmin, relative humidity (RH), and the Clausius Clapeyron equation, as well as machine learning, were introduced in this research and compared to other methods. Overall, the frost prediction results showed that the minimum temperature measured at the meteorological shelter that best describes frost formation is 2.5 °C. Additionally, the condition whereby the dew point temperature is lower than 0 °C results in a reduction in the proportion of false alarms. Methods that introduce additional variables outperform those that rely solely on the temperature. The method in which days are classified as exhibiting frost using a Tmin threshold of 3 °C and Td threshold of 0 °C (ased on Tmin and daily mean RH) could capture the most frost days with the smallest error. This method is the most suited for continental areas with a high probability of detection (POD > 0.9) and a probability of false detection (POFD < 0.3) which conforms with the history of frost occurrence in this type of climate zone. These findings were corroborated by signal detection theory analysis, yielding high values of the accuracy index and beta values below 1, indicating a bias toward estimating frost events (with high hit rate values and high false alarm values). This method could be used to identify geographic areas most susceptible to frost formation and, if coupled with a climate model, enable the study of the frost vulnerability due to climate change.
In this study, a synoptic and mesoscale analysis was performed and Szilagyi’s waterspout forecasting method was tested on ten waterspout events in the period of 2013–2016. Data regarding waterspout ...occurrences were collected from weather stations, an online survey at the official website of the National Meteorological and Hydrological Service of Croatia and eyewitness reports from newspapers and the internet. Synoptic weather conditions were analyzed using surface pressure fields, 500 hPa level synoptic charts, SYNOP reports and atmospheric soundings. For all observed waterspout events, a synoptic type was determined using the 500 hPa geopotential height chart. The occurrence of lightning activity was determined from the LINET lightning database, and waterspouts were divided into thunderstorm-related and “fair weather” ones. Mesoscale characteristics (with a focus on thermodynamic instability indices) were determined using the high-resolution (500 m grid length) mesoscale numerical weather model and model results were compared with the available observations. Because thermodynamic instability indices are usually insufficient for forecasting waterspout activity, the performance of the Szilagyi Waterspout Index (SWI) was tested using vertical atmospheric profiles provided by the mesoscale numerical model. The SWI successfully forecasted all waterspout events, even the winter events. This indicates that the Szilagyi’s waterspout prognostic method could be used as a valid prognostic tool for the eastern Adriatic.
This study investigates the sensitivity of a high-resolution mesoscale atmospheric model in the model reproduction of thermally induced local wind (i.e., sea breezes, SB) on the development of deep ...convection (Cb). The three chosen cases are simulated by the Weather and Research Forecasting (WRF-ARW) model at three (nested) model domains, whereas the area of the interest is Istria (peninsula in the northeastern Adriatic). The sensitivity tests are accomplished by modifying (1) the model setup, (2) the model topography and (3) the sea surface temperature (SST) distribution. The first set of simulations (over the three 1.5-day periods during summer) is conducted by modifying the model setup, i.e., microphysics and the boundary layer parameterizations. The same events are simulated with the modified topography where the mountain heights in Istria are reduced to 30% of their initial height. The SST distribution has two representations in the model: a constant SST field from the ECMWF skin temperature analysis and a varying SST field, which is provided by hourly geostationary satellite data. A comprehensive set of numerical experiments is statistically analyzed through several different approaches (i.e., the standard statistical measures, the spectral method and the image moment analysis). The overall model evaluation of each model setup revealed certain advantages of one model setup over the others. The numerical tests with the modified topography showed the influence of reducing the mountains heights on the pre-thunderstorm characteristics due to: (1) decrease of sensible heat flux and mid-tropospheric moisture and (2) change of slope-SB wind system. They consequently affect the evolution and dimensions of SBs and the features of the thunderstorm itself: timing, location and intensity (weaker storm). The implementation of the varying SST field in the model have an impact on the characteristics and dynamics of the SB and finally on the accuracy of Cb evolution, duration and the intensity. SST variations emphasized the importance of the phase matching in both daytime cycles of SB and Cb due to their extremely strong nonlinear relationship.