Reconstructing pollen deposition in Northeastern China: increased deposition intensity without prolongation of pollen season

Abstract

Introduction The burden of pollen-related allergic diseases continues to increase. Characterizing pollen exposure and its long-term changes is therefore important for allergy risk warning and public health protection. Methods Utilizing daily pollen deposition observations and multisource environmental data from 15 cities in Northeastern China during 2022-2024, we developed a random forest model and evaluated its performance through cross-year and leave-one-city-out validation. Shapley Additive Explanations (SHAP) were used to identify the principal predictors. We then reconstructed daily pollen deposition for 2000-2024 and examined long-term changes in pollen season characteristics, maximum daily pollen deposition (MP), and seasonal pollen deposition integral (SPIn). Results Cross-year validation yielded R ² values of 0.542-0.593, whereas leave-one-city-out validation yielded R ² values of 0.247-0.655, indicating relatively stable temporal generalizability and some spatial transferability. DOY and 30-day growing degree days were important predictors in both spring and autumn; forest cover and NDVI made relatively large contributions in spring and autumn, respectively. Model estimates indicated that the start and end of the spring pollen season tended to advance by 0.133 and 0.149 days/year, respectively (0.05 ≤ p < 0.10), while season duration remained largely stable. In autumn, the start, peak, and end dates were significantly delayed by 0.236, 0.133, and 0.115 days/year, respectively ( p < 0.05), whereas season duration tended to decrease by 0.117 days/year (0.05 ≤ p < 0.10). Spring MP and SPIn increased significantly by 12 and 277 grains/(1,000 mm²·year), respectively ( p < 0.001), whereas autumn MP increased significantly by 2 grains/(1,000 mm²·year) ( p < 0.001) and SPIn tended to increase by 24 grains/(1,000 mm²·year) (0.05 ≤ p < 0.10). Discussion Overall, the model estimates suggest that long-term changes in pollen exposure across the study area may be characterized more by increasing deposition intensity than by prolonged pollen seasons. These findings provide environmental context for regional pollen exposure assessment and seasonal allergy-risk awareness.