首页|Reconstructing multi-decadal airborne birch pollen levels based on NDVI data and a pollen transport model
Reconstructing multi-decadal airborne birch pollen levels based on NDVI data and a pollen transport model
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NSTL
Elsevier
Airborne birch pollen may elicit allergies and affect the public health badly. Timely spatially distributed information on current and forecasted pollen levels may help people with pollen allergies to take preventive measures. This requires a modelling approach. Here we reconstruct multi-decadal (1982-2019) daily spatially distributed airborne birch pollen levels by ingesting seasonal dynamic birch pollen emission source maps into the pollen transport model SILAM (System for Integrated modeLling of Atmospheric coMposition) in a bottom-up approach. We introduce seasonal variations in the birch pollen emission maps by combining a forest inventory based areal birch fraction map with four decades of spaceborne Normalized Difference Vegetation Index (NDVI) in a Random Forest statistical framework. The approach of combining the transport model with NDVI based pollen emission maps is applied and evaluated with birch pollen observations by Hirst method from the Belgian aerobiological surveillance network that go back to 1982. Transport in SILAM is driven by ECMWF ERA5 meteorological data. The mean seasonal R2 values between modelled and observed time series of airborne birch pollen levels in the period 1982-2019 range between 0.35 and 0.63, but can go up to 0.86 for individual seasons, indicating good performance of SILAM for Belgium. Here we show that the predicted amount of birch pollen in the air in Belgium has been increasing on average by 13.1% per decade based on the Sen slopes computed on the Seasonal Pollen Integral for the period 1982-2019. Analysis of the SILAM runs shows that this increase over time is mainly climate-induced (8.2% per decade), but it is amplified by the spatiotemporal variations of the birch pollen emission sources with 4.9% per decade.
NSTL
Elsevier
