When pollen monitoring meets hyperspectral imaging

When spring arrives, it brings beautiful flowers and vibrant colors, but also some discomfort and allergy symptoms. These symptoms may be caused by pollen, tiny particles composed of the male reproductive cells of plants that can be dispersed by wind, insects, or other animals to pollinate other plants. Pollen exists in the air and may enter our respiratory tract, potentially affecting our health.

Pollen concentration monitoring indicators and influencing factors are quite complex, involving not only pollen concentration, pollen particle size, and distance from pollen emission sources, but also local wind speed, wind direction, sunshine duration, precipitation type and intensity, and atmospheric stability.This is related to meteorological factors.

Traditional pollen monitoring mainly uses Manual monitoringand Automatic instrument monitoringThere are two main methods. Meteorological bureaus mostly use instrument monitoring.

In most practices, pollen measurements are performed using physical samplers, such as rotating rod devices or Hirst-type bulk spore traps. These devices capture pollen from the air and then analyze it under a microscope to count individual pollen grains. While this process has been standard practice for decades, its reliance on human intervention limits its accuracy and geographic coverage.

The range of pollen species covered is limited:Because traditional methods have a wide range of applications, the pollen level data measured using these methods have significant limitations. In many cases, monitoring stations only focus on a few pollen species, resulting in data gaps.

 Geographical distribution issues:Traditional pollen monitoring stations are concentrated in specific areas, mainly urban areas, while there is no monitoring in rural or suburban areas. This limited distribution results in poor geographical coverage.

Manual counting:Manually counting under a microscope is a tedious process prone to human error. Furthermore, it heavily relies on the availability of skilled personnel, impacting timelines and consequently affecting the reliability and validity of the data.

Sampling variability: The placement of pollen collectors significantly affects readings. Even moving the collector a few meters or changing its orientation can alter the results. Different sites use poles or plates of different sizes, affecting the number of pollen grains captured. These variables can lead to inconsistencies in data collected from different areas.

Environmental and technological factors:External factors (such as rain, wind, or machine malfunction) can interfere with data collection. In addition, the unpredictability of machine malfunctions or human delays in data collection may lead to data loss or inaccuracies.

Inconsistent time resolution:Traditional pollen counting is typically conducted daily or weekly, providing little to no real-time information. In many cases, monitoring stations do not report exact counts, but instead issue general categories such as "low," "medium," or "high" pollen levels, which may not be detailed enough for people with severe allergies.

Site-specific scope:Pollen count ranges are sometimes site-specific and may differ from national or international standards, making it difficult to compare pollen levels in different regions.

Limited point data:Each pollen monitoring station covers a specific small geographical point. However, pollen can spread over long distances, meaning that point measurements cannot account for the broader environmental factors that influence pollen dispersal.

Figure 1. Pollen monitoring in Yulin City (Source: Yulin Meteorological Bureau)

Hyperspectral imaging can capture spectral data of surface reflected light in multiple continuous narrow bands, and it has a wide coverage and rich spectral information. It can monitor pollen by identifying pollen grains or vegetation signals related to pollen. 

Figure 2. Spectral curves of pollen from different trees.

Different plants release different types of pollen during their flowering period, and these... The spectral reflectance characteristics of plants (such as chlorophyll, carotenoids, and water content) can be obtained through hyperspectral data analysis.Satellites can detect the spectral signals of vegetation, combined with ground-based observation data (such as pollen sampling). Establish pollen type and spectral characteristics Relationship between signs 。

In addition, it can be combined with Ground sensors 、 Remote sensing satellite imagesand Global proprietary sensor networkIts functionality allows it to obtain reliable data from multiple sources. Through... Model Measurement 、 Processing and sorting AnalysisMassive amounts of data from different sources ensure the accuracy of model monitoring.

Figure 3. Tree pollen monitoring using satellite imagery (Xi'an area)