Poisson, negative binomial, and zero-inflated negative binomial regression models for predicting daily airborne pollen concentration levels in Sinop (Türkiye)


YİĞİTER A., Demir C. C., Hamurkaroğlu C., Özler H., Kaplan A., Danacıoğlu N., ...More

Environmental Monitoring and Assessment, vol.198, no.1, 2026 (SCI-Expanded, Scopus) identifier identifier

  • Publication Type: Article / Article
  • Volume: 198 Issue: 1
  • Publication Date: 2026
  • Doi Number: 10.1007/s10661-025-14871-0
  • Journal Name: Environmental Monitoring and Assessment
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, ABI/INFORM, BIOSIS, Compendex, EMBASE, Environment Index, Geobase, Greenfile, MEDLINE, Public Affairs Index, Urban Studies Abstracts
  • Keywords: Meteorological conditions, Negative binomial regression, Poisson regression, Pollen, Sinop (Türkiye), Zero-inflated negative binomial regression
  • Hacettepe University Affiliated: Yes

Abstract

Pollen, produced during the flowering period of plants, especially anemogamous plants that produce high volumes of pollen, poses a risk to individuals with pollen allergies when it is present in the atmosphere. Meteorological factors are known to affect the duration, distribution, and amount of pollen in the air. The remarkable increase in allergic cases in recent years has led to many studies investigating the relationship between pollen and spores that cause allergies and meteorological factors in Türkiye as well as in the world. In this study, meteorological factors and their influence on pollen concentrations in the air were examined for the Sinop region in northern Türkiye. First, descriptive statistics for pollen obtained from plant taxa were obtained and interpreted. Precipitation, humidity, temperature, and wind speed were considered as meteorological parameters, and the effects of these variables on pollen counts and their annual changes were modelled using Poisson, negative binomial, and zero-inflated negative binomial (ZINB) regression models. The estimation results for all pollen taxa were then discussed. In the models obtained for each pollen type, the statistical significance of the independent variables such as temperature, precipitation, relative humidity, wind speed, time, and lag 1 was found to be different according to the pollen type.