Long-term and short-term forecast models for Poaceae (grass) pollen in Poznań, Poland,constructed using regression analysis

Airborne concentrations of Poaceae pollen have been monitored in Poznań for more than 10 years and the length of the dataset is now considered sufficient for statistical analysis. The objective of this paper is to produce long-range forecasts that predict certain characteristics of the grass pollen season (such as the start, peak and end dates of the grass pollen season) as well as short-term forecasts that predict daily variations in grass pollen counts for the next day or next few days throughout the main grass pollen season. The method of forecasting was regression analysis. Correlation analysis was used to examine the relationship between grass pollen counts and the factors that affect its production, release and dispersal. The models were constructed with data from 1994 to 2004 and tested on data from 2005 and 2006. The forecast models predicted the start of the grass pollen season to within two days and achieved 61% and 70% accuracy on a scale of 1–4 when forecasting variations in daily average grass pollen counts in 2005 and 2006, respectively. This study has emphasised how important the weather during the few weeks or months preceding pollination is to grass pollen production, and draws attention to the importance of considering large-scale patterns of climate variability (indices of the North Atlantic Oscillation) when constructing forecast models for allergenic pollen.     

Development and validation of a 5-day-ahead hay fever forecast for patients with grass-pollen-induced allergic rhinitis

One-third of the Dutch population suffers from allergic rhinitis, including hay fever. In this study, a 5-day-ahead hay fever forecast was developed and validated for grass pollen allergic patients in the Netherlands. Using multiple regression analysis, a two-step pollen and hay fever symptom prediction model was developed using actual and forecasted weather parameters, grass pollen data and patient symptom diaries. Therefore, 80 patients with a grass pollen allergy rated the severity of their hay fever symptoms during the grass pollen season in 2007 and 2008. First, a grass pollen forecast model was developed using the following predictors: (1) daily means of grass pollen counts of the previous 10 years; (2) grass pollen counts of the previous 2-week period of the current year; and (3) maximum, minimum and mean temperature (R ²?=?0.76). The second modeling step concerned the forecasting of hay fever symptom severity and included the following predictors: (1) forecasted grass pollen counts; (2) day number of the year; (3) moving average of the grass pollen counts of the previous 2 week-periods; and (4) maximum and mean temperatures (R ²?=?0.81). Since the daily hay fever forecast is reported in three categories (low-, medium- and high symptom risk), we assessed the agreement between the observed and the 1- to 5-day-ahead predicted risk categories by kappa, which ranged from 65 % to 77 %. These results indicate that a model based on forecasted temperature and grass pollen counts performs well in predicting symptoms of hay fever up to 5 days ahead.     

Exploring the spatio-temporal relationship between two key aeroallergens and meteorological variables in the United Kingdom

Constructing accurate predictive models for grass and birch pollen in the air, the two most important aeroallergens, for areas with variable climate conditions such as the United Kingdom, require better understanding of the relationships between pollen count in the air and meteorological variables. Variations in daily birch and grass pollen counts and their relationship with daily meteorological variables were investigated for nine pollen monitoring sites for the period 2000–2010 in the United Kingdom. An active pollen count sampling method was employed at each of the monitoring stations to sample pollen from the atmosphere. The mechanism of this method is based on the volumetric spore traps of Hirst design (Hirst in Ann Appl Biol 39(2):257–265, 1952). The pollen season (start date, finish date) for grass and birch were determined using a first derivative method. Meteorological variables such as daily rainfall; maximum, minimum and average temperatures; cumulative sum of Sunshine duration; wind speed; and relative humidity were related to the grass and birch pollen counts for the pre-peak, post peak and the entire pollen season. The meteorological variables were correlated with the pollen count data for the following temporal supports: same-day, 1-day prior, 1-day mean prior, 3-day mean prior, 7-day mean prior. The direction of influence (positive/negative) of meteorological variables on pollen count varied for birch and grass, and also varied when the pollen season was treated as a whole season, or was segmented into the pre-peak and post-peak seasons. Maximum temperature, sunshine duration and rainfall were the most important variables influencing the count of grass pollen in the atmosphere. Both maximum temperature (pre-peak) and sunshine produced a strong positive correlation, and rain produced a strong negative correlation with grass pollen count in the air. Similarly, average temperature, wind speed and rainfall were the most important variables influencing the count of birch pollen in the air. Both wind speed and rain produced a negative correlation with birch pollen count in the air and average temperature produced a positive correlation.     

Model for forecasting Olea europaea L. airborne pollen in South-West Andalusia, Spain

Data on predicted average and maximum airborne pollen concentrations and the dates on which these maximum values are expected are of undoubted value to allergists and allergy sufferers, as well as to agronomists. This paper reports on the development of predictive models for calculating total annual pollen output, on the basis of pollen and weather data compiled over the last 19 years (1982–2000) for Córdoba (Spain). Models were tested in order to predict the 2000 pollen season; in addition, and in view of the heavy rainfall recorded in spring 2000, the 1982–1998 data set was used to test the model for 1999. The results of the multiple regression analysis show that the variables exerting the greatest influence on the pollen index were rainfall in March and temperatures over the months prior to the flowering period. For prediction of maximum values and dates on which these values might be expected, the start of the pollen season was used as an additional independent variable. Temperature proved the best variable for this prediction. Results improved when the 5-day moving average was taken into account. Testing of the predictive model for 1999 and 2000 yielded fairly similar results. In both cases, the difference between expected and observed pollen data was no greater than 10%. However, significant differences were recorded between forecast and expected maximum and minimum values, owing to the influence of rainfall during the flowering period. Received: 25 October 2000 / Revised: 26 February 2001 / Accepted: 28 February 2001     

Influence of the North Atlantic Oscillation on grass pollen counts in Europe

Relationships between temporal variations in the North Atlantic Oscillation (NAO) and grass pollen counts at 13 sites in Europe, ranging from Córdoba in the south-west and Turku in the north-east, were studied in order to determine spatial differences in the amount of influence exerted by the NAO on the timing and magnitude of grass pollen seasons. There were a number of significant (P < 0.05) relationships between the NAO and start dates of the grass pollen season at the 13 pollen-monitoring sites. The strongest associations were generally recorded near to the Atlantic coast. Several significant correlations also existed between winter averages of the NAO and grass pollen season severity. Traditional methods for predicting the start or magnitude of grass pollen seasons have centred on the use of local meteorological observations, but this study has shown the importance of considering large-scale patterns of climate variability like the NAO.     

Forecasting the onset of the grass pollen season in Melbourne (Australia)

In Melbourne, a southern hemisphere city with a cool temperate climate, the grass pollen season has been monitored using a Burkard spore trap for 12 years (11 pollen seasons, which extend from October through January). The onset of the grass pollen season (OGPS) has been defined in various ways using both arbitrary cumulative scores (Sum 75, Sum 100) and percentages (10% Pollen Fly). OGPS, based on the forecast model of pollen season devised by Lejoly-Gabriel (Acta Geogr. Lovan., 13 (1978) 1–260) has been most widely used in efforts to forecast the beginning of the pollen season. OGPS occurred in Melbourne between 20 October to 24 November (average 6 November), a difference of 35 days. Duration of the pollen season ranged from 46 to 81 days, with a mean of 55 days, one of the longest reported. The relationships between onset and various weather parameters for July have enabled us to modify a model, using linear regression analysis, to predict onset. The prediction model is based on a negative correlation between date of onset and the sum of rainfall for July (a winter month). The error of prediction (Ep) is 24% and predicted day of OGPS was precisely predicted on 2 occasions, and on others with a range of accuracy of 3 to 14 days.     

Atmospheric Poaceae pollen frequencies and associations with meteorological parameters in Brisbane,Australia: a 5-year record, 1994–1999

Grass pollen is an important risk factor for allergic rhinitis and asthma in Australia and is the most prevalent pollen component of the aerospora of Brisbane, accounting for 71.6% of the annual airborne pollen load. A 5-year (June 1994–May 1999) monitoring program shows the grass pollen season to occur during the summer and autumn months (December–April), however the timing of onset and intensity of the season vary from year to year. During the pollen season, Poaceae counts exceeding 30 grains m^–3 were recorded on 244 days and coincided with maximum temperatures of 28.1 ± 2.0 °C. In this study, statistical associations between atmospheric grass pollen loads and several weather parameters, including maximum temperature, minimum temperature and precipitation, were investigated. Spearmans correlation analysis demonstrated that daily grass pollen counts were positively associated (P < 0.0001) with maximum and minimum temperature during each sampling year. Precipitation, although considered a less important daily factor (P < 0.05), was observed to remove pollen grains from the atmosphere during significant periods of rainfall. This study provides the first insight into the influence of meteorological variables, in particular temperature, on atmospheric Poaceae pollen counts in Brisbane. An awareness of these associations is critical for the prevention and management of allergy and asthma for atopic individuals within this region.     

Two statistical approaches to forecasting the start and duration of the pollen season of Ambrosia in the area of Lyon (France)

The aim of the present study was to forecast the start and duration of the pollen season of Ambrosia from meteorological data, in order to provide early information to allergists and allergic people. We used the airborne pollen data from Lyon (France), sampled using a Hirst trap from 1987 to 1999, and the meteorological data for the same period: air temperature (minimal, maximal, and average), rainfall, relative humidity, sunshine duration and soil temperature. Two forecasting models were used, one summing the temperatures and the other making use of a multiple regression on 10-day or monthly meteorological parameters. The start of the pollen season was predicted with both methods, results being more accurate with the regression (the errors between the predicted and the observed SDP ranging from 0 to 3 days). The duration of the pollen season was predicted by a regression model, errors ranging from 0 to 7 days. The models were later tested with satisfactory results from 2 additional years (2000 and 2001). Such forecasting models are helpful for allergic people, who have to begin their anti-allergic treatment before the start of the pollen season and not when the symptoms have appeared, since a preventive treatment is more efficient than a curative one. The regression allows predictions to be made 3-5 weeks in advance and so it is of particular interest. The forecasts will be broadcast on the Internet.     

Prediction of birch airborne pollen counts by examining male catkin numbers in Hokkaido,northern Japan

Birch pollen is a very common cause of pollinosis in Hokkaido, northern Japan. Birch airborne pollen concentrations vary each year; hence, the development of a method for predicting annual airborne pollen concentration is very important in preventing widespread symptoms of pollinosis. In the current study, we investigated airborne pollen counts and male catkin numbers (male flower index) of birch in four cities of Hokkaido between 2002 and 2008. Airborne pollen surveys were conducted using Durham’s sampler, and male catkin numbers determined for three major birch species (Betula platyphylla var. japonica, B. emanii, and B. maximowicziana). We found an annual variation in male flower index for all the three birch species investigated. This variation worked in combination with the amount of precipitation during the pollen season to influence total birch pollen counts. In conclusion, the male catkin numbers of three major birch species reliably predict airborne pollen counts in Hokkaido, but only when the effect of precipitation during pollen season is considered.     

Predicting the grass pollen count from meteorological data with regard to estimating the severity of hayfever symptoms in Melbourne (Australia)

In Melbourne, Australia, grass pollen is the predominant cause of hayfever in late spring and summer. The grass pollen season has been monitored in Melbourne, using a Burkard spore trap, for 13 years (1975–1981, 1985 and 1991–1997). Total counts for grass pollen were highly variable from one season to the next (approximately 1000 to >8000 grains/m³). The daily grass pollen counts also showed a high variability (0 to approximately 400 grains/m³). In this study, the grass pollen counts of the 13 years (12 grass pollen seasons, extending from October to January) have been compared with meteorological data in order to identify the conditions that can determine the daily amounts of grass pollen in the air. It was found that the seasonal total of grass pollen was directly correlated with the rainfall sum of the preceding 12 months (1 September–31 August): seasonal total of grass pollen (counts/m³)=18.161 × rainfall sum of the preceding 12 months (mm) −8541.5 (r _s=0.74,P<0.005,n=12). The daily amounts of grass pollen in the air were positively correlated with the corresponding daily average ambient temperatures (P<0.001). The daily amount of grass pollen which was to be expected with a certain daily average temperature was linked to the seasonal total of grass pollen: in years with high total grass pollen counts, a lower daily average temperature was required for a high daily pollen count than in years with low total grass pollen counts. As the concentration of airborne grass pollen determines the severity of hayfever in sensitive patients, an estimation of daily grass pollen counts can provide an indication of potential pollinosis symptoms. We compared daily grass pollen counts with the reported symptomatic responses of hayfever sufferers in November 1985 and found that hayfever symptoms were significantly correlated to the grass pollen counts (P<0.001 for nasal,P<0.005 for eye symptoms). Thus, a combination of meteorological information (i.e. rainfall and temperature) allows for an estimation of the potential daily pollinosis symptoms during the grass pollen season. Here we propose a symptom estimation chart, allowing a quick prediction of eye and nasal symptoms that are likely to occur as a result of variations in meteorological conditions, thus enabling both physicians and patients to take appropriate avoidance measures or therapy.     

Integration of flowering dates in phenology and pollen counts in aerobiology: analysis of their spatial and temporal coherence in Germany (1992–1999)

We studied the possibility of integrating flowering dates in phenology and pollen counts in aerobiology in Germany. Data were analyzed for three pollen types (Betula, Poaceae, Artemisia) at 51 stations with pollen traps, and corresponding phenological flowering dates for 400 adjacent stations (< 25 km) for the years 1992–1993 and 1997–1999. The spatial and temporal coherence of these data sets was investigated by comparing start and peak of the pollen season with local minima and means of plant flowering. Our study revealed that start of birch pollen season occurred on average 5.7 days earlier than local birch flowering. For mugwort and grass, the pollen season started on average after local flowering was observed; mugwort pollen was found 4.8 days later and grass pollen season started almost on the same day (0.6 days later) as local flowering. Whereas the peak of the birch pollen season coincided with the mean flowering dates (0.4 days later), the pollen peaks of the other two species took place much later. On average, the peak of mugwort pollen occurred 15.4 days later than mean local flowering, the peak of grass pollen catches followed 22.6 days after local flowering. The study revealed a great temporal divergence between pollen and flowering dates with an irregular spatial pattern across Germany. Not all pollen catches could be explained by local vegetation flowering. Possible reasons include long-distance transport, pollen contributions of other than phenologically observed species and methodological constraints. The results suggest that further research is needed before using flowering dates in phenology to extrapolate pollen counts.     

Regional and seasonal variation in airborne grass pollen levels between cities of Australia and New Zealand

Although grass pollen is widely regarded as the major outdoor aeroallergen source in Australia and New Zealand (NZ), no assemblage of airborne pollen data for the region has been previously compiled. Grass pollen count data collected at 14 urban sites in Australia and NZ over periods ranging from 1 to 17 years were acquired, assembled and compared, revealing considerable spatiotemporal variability. Although direct comparison between these data is problematic due to methodological differences between monitoring sites, the following patterns are apparent. Grass pollen seasons tended to have more than one peak from tropics to latitudes of 37°S and single peaks at sites south of this latitude. A longer grass pollen season was therefore found at sites below 37°S, driven by later seasonal end dates for grass growth and flowering. Daily pollen counts increased with latitude; subtropical regions had seasons of both high intensity and long duration. At higher latitude sites, the single springtime grass pollen peak is potentially due to a cooler growing season and a predominance of pollen from C₃ grasses. The multiple peaks at lower latitude sites may be due to a warmer season and the predominance of pollen from C₄ grasses. Prevalence and duration of seasonal allergies may reflect the differing pollen seasons across Australia and NZ. It must be emphasized that these findings are tentative due to limitations in the available data, reinforcing the need to implement standardized pollen-monitoring methods across Australasia. Furthermore, spatiotemporal differences in grass pollen counts indicate that local, current, standardized pollen monitoring would assist with the management of pollen allergen exposure for patients at risk of allergic rhinitis and asthma.     

The use of discriminant analysis and neural networks to forecast the severity of the Poaceae pollen season in a region with a typical Mediterranean climate

Biological particles in the air such as pollen grains can cause environmental problems in the allergic population. Medical studies report that a prior knowledge of pollen season severity can be useful in the management of pollen-related diseases. The aim of this work was to forecast the severity of the Poaceae pollen season by using weather parameters prior to the pollen season. To carry out the study a historical database of 21 years of pollen and meteorological data was used. First, the years were grouped into classes by using cluster analysis. As a result of the grouping, the 21 years were divided into 3 classes according to their potential allergenic load. Pre-season meteorological variables were used, as well as a series of characteristics related to the pollen season. When considering pre-season meteorological variables, winter variables were separated from early spring variables due to the nature of the Mediterranean climate. Second, a neural network model as well as a discriminant linear analysis were built to forecast Poaceae pollen season severity, according to the three classes previously defined. The neural network yielded better results than linear models. In conclusion, neural network models could have a high applicability in the area of prevention, as the allergenic potential of a year can be determined with a high degree of reliability, based on a series of meteorological values accumulated prior to the pollen season.     

Forecasting ragweed pollen characteristics with nonparametric regression methods over the most polluted areas in Europe

Nonparametric time-varying regression methods were developed to forecast daily ragweed pollen concentration, and the probability of the exceedance of a given concentration threshold 1 day ahead. Five-day and 10-day predictions of the start and end of the pollen season were also addressed with a nonparametric regression technique combining regression analysis with the method of temperature sum. Our methods were applied to three of the most polluted regions in Europe, namely Lyon (Rhône Valley, France), Legnano (Po River Plain, Italy) and Szeged (Great Plain, Hungary). For a 1-day prediction of both the daily pollen concentration and daily threshold exceedance, the order of these cities from the smallest to largest prediction errors was Legnano, Lyon, Szeged and Legnano, Szeged, Lyon, respectively. The most important predictor for each location was the pollen concentration of previous days. The second main predictor was precipitation for Lyon, and temperature for Legnano and Szeged. Wind speed should be considered for daily concentration at Legnano, and for daily pollen threshold exceedances at Lyon and Szeged. Prediction capabilities compared to the annual cycles for the start and end of the pollen season decreased from west to east. The order of the cities from the lowest to largest errors for the end of the pollen season was Lyon, Legnano, Szeged for both the 5- and 10-day predictions, while for the start of the pollen season the order was Legnano, Lyon, Szeged for 5-day predictions, and Legnano, Szeged, Lyon for 10-day predictions.     

Using meteorologic data to predict daily ragweed pollen levels

Pollen-related allergy is a common disease resulting in symptoms of hay fever and asthma. Control of symptoms depends (generally) on avoidance and pharmacological treatment. Both of these approaches could benefit from accurate predictions of pollen levels for future days. We have constructed a model that uses meteorological data to predict ragweed pollen levels based on air samples collected daily in Kalamazoo, MI from 1991 to 1994. Ragweed pollen counts were converted to pollen grains/m³ of air (24-h average). We used Poisson regression, which appropriately handles the heterogenous variance associated with pollen data. Using standard statistical model selection procedures, combined with biological considerations, we selected rainfall, wind speed, temperature, and the time measured from the start of the season as the most significant variables. Using our model, we propose a method that uses the weather forecast for the following day to predict the ragweed pollen level. This approach differs from most previous attempts because it uses Poisson regression and because this model needs to be fit iteratively each day. By updating the coefficients of the model based on the information to date, this method allows the fundamental shape of the pollen distribution curve to change from year to year. Application to the Kalamazoo data suggests that the method has good sensitivity and specificity for predicting high pollen days.     

Examining high magnitude grass pollen episodes at Worcester,United Kingdom,using back-trajectory analysis

Trajectory analysis is a valuable tool that has been used before in aerobiological studies, to investigate the movement of airborne pollen. This study has employed back-trajectories to examine the four highest grass pollen episodes at Worcester, during the 2001 grass pollen season. The results have shown that the highest grass pollen counts of the 2001 season were reached when air masses arrived from a westerly direction. Back-trajectory analysis has a limited value to forecasters because the method is retrospective and cannot be employed directly for forecasting. However, when used in conjunction with meteorological data this technique can be used to examine high magnitude events in order to identify conditions that lead to high pollen counts.     

Development of a symptom load index: enabling temporal and regional pollen season comparisons and pointing out the need for personalized pollen information

This study describes the development of an index, the symptom load index, to compare the severity of pollen seasons for pollen allergy sufferers during different seasons in different geographical regions. This is done by comparing symptom data with pollen data. This study is based on symptom data of the Patients’ Hayfever Diary (PHD) and on pollen data from the European Aeroallergen Network (EAN). The PHD allows objective assessment of the season severity. Season length calculated on the basis of pollen data from the EAN and all records of users in a specific region are included in calculation of the symptom load index. It is thus not influenced by pollen counts during the season. Its application to pollen seasons of the three important pollen types birch, grass and ragweed in Austria and Germany from 2009 to 2013 proved that the impact of a pollen season on allergy sufferers is not correlated with total pollen loads. The possibility to calculate the pollen season in the allergy sufferers’ point of view is a novelty that inaugurates new pathways for the future of pollen information. Furthermore, such an index will be of value for the establishment of personalized pollen information and for calculations of pollen thresholds.     

Spatial and temporal modeling of daily pollen concentrations

Accurate assessments of pollen counts are valuable to allergy sufferers, the medical industry, and health researchers; however, monitoring stations do not exist in most areas. In addition, the degree of spatial reliability provided by the limited number of monitoring stations is poorly understood. We developed and compared spatial models to estimate pollen concentrations in locations without monitoring stations. Daily Acer, Quercus, and overall tree, grass, and weed pollen counts, in grains/m³, were obtained from 14 aeroallergen monitoring stations located in the northeastern and mid-Atlantic region of the United States from 2003 to 2006. Pollen counts were spatially interpolated using ordinary kriging. Mixed effects and generalized estimating equations incorporating daily and seasonal weather characteristics, pollen season characteristics and land-cover information were also developed to estimate daily pollen concentrations. We then compared observed values from a monitoring station to model estimates for that location. Observed counts and kriging estimates for tree pollen differed (p = 0.04), but not when peak periods were removed (p = 0.29). No differences between observed and kriging estimates of Acer (p = 0.46), Quercus (p = 0.24), grass (p = 0.31) or weed pollen (p = 0.29) were found. Estimates from longitudinal models also demonstrated good agreement with observed counts, except for the extremes of pollen distributions. Our results demonstrate that spatial interpolation techniques as well as regression methods incorporating both weather and land-cover characteristics can provide reliable estimates of daily pollen concentrations in areas where monitors do not exist for all but periods of extremely high pollen.     

Trends in grass pollen season in southern Spain

The main characteristics of Poaceae pollen season at 8 sites in Andalusia were studied. Special attention was paid in the trends of grass pollen-season start and peak dates. Moreover, we analyse the intensity of the grass pollen season over the study period as well as potential temporal and spatial patterns in these data. Statistical analysis was performed to determine the possible influence of weather-related parameters on variations in the grass pollen season. Main results show an advance in the start and peak of grass pollen season and an increase in the annual Pollen Index and in the severity of the season (days > 25 pollen grains/m³). The future consequences of these changes in grass phenology could be related with changes in land use and also in pollinosis symptoms due to the higher concentrations recorded but also to the variations on pollen season dates.     

Poaceae pollen in the atmosphere of Tetouan (NW Morocco): effect of meteorological parameters and forecast of daily pollen concentration

The Poaceae pollen season has been characterized in Tetouan during a 7-year period, and the effect of weather conditions on daily concentrations was examined. The forecast models were produced using a stepwise multiple regression analyses. Firstly, three models were constructed to predict daily Poaceae pollen concentrations during the main pollen season, as well as the pre-peak and post-peak periods with data from 2008 to 2012 and tested on data from 2013 and 2014. Secondly, the regression models using leave-one-out cross-validation were produced with data obtained during 2008–2014 taking into account meteorological parameters and mean pollen concentrations of the same day in other years. The duration of the season ranged from 70 days in 2009 to 158 days in 2012. The highest amount of Poaceae pollen was detected in spring and the first fortnight of July. The annual sum of airborne Poaceae pollen concentrations varied between 2100 and 6251. The peak of anthesis was recorded in May in six of the other years studied. The regression models accounted for 36.3–85.7% of variance in daily Poaceae pollen concentrations. The models fitted best when the mean pollen concentration of the same day in other years was added to meteorological variables, and explained 78.4–85.7% of variance of the daily pollen changes. When the year 2014 was used for validating the models, the lowest root-mean-square errors values were found between the observed and estimated data (around 13). The reasonable predictor variables were the mean pollen concentration of the same day in other years, mean temperature, precipitations, and maximum relative humidity.