Pollen season is underway in Philly. Here’s why pollen is so hard to predict and track.

The snowdrops are popping, daffodils are blooming, birds are frantically nest-scouting, and as sure as a red sea of fans will be filling Citizens Bank Park to see the Phillies, for perhaps hundreds of thousands of local allergy sufferers the tissue and eye-drop season has commenced.

The annual six-month cavalcade of pollen is underway, starting with the bud-swollen trees that are preparing to leaf out. The typical peak period still is a few weeks away, but the tree pollen is evident in the automated daily counts. Later in the spring on into September, the two other prongs of the pollen trinity — grasses and ragweeds — will follow.

And in the 2025 season, it appears that the Philadelphia region will be without an official human pollen-counter for the first time in 30 years, with the area’s only National Allergy Bureau station going dark. The automated sensors are useful, but local pollen sufferers no longer will have official ground-truth readings of pollen concentrations, which are valuable for tracking the trends and intensities of a given season.

While various studies have asserted that pollen levels have been increasing with the warmth of the planet, for a variety of reasons predicting and gauging the severity of any given pollen season is something that no one has quite mastered.

One of those reasons — perhaps a surprising one given that the Centers for Disease Control and Prevention estimates that pollen allergies nationally generate about $3 billion annually in medical expenses in this country — is that pollen is poorly measured in the United States and that reliable data are sorely lacking.

The allergy bureau pollen-counting network “has been decreasing for the last decade,” said Fiona Lo, a researcher at the University of Washington. The bureau has well less than one station per three million noses, far behind the networks of Germany, Japan, and other countries, according to a study by a German researcher.

And the authors of a 2022 study paper published in the International Journal of Environmental Research and Public Health observed that “the scarcity of pollen data in the U.S. presents a major challenge” to pollen research.

Counting is labor-intensive, for one thing. And, traditionally, the counts have been supplied on a voluntary basis by allergists, including Donald Dvorin, who practiced in Mount Laurel and was part of the allergy bureau network for over 30 years.

During time that he could have been seeing patients, the ritual required him to set up a pollen trap on the roof of his practice to catch a sample of what was in the air. Then, 24 hours later, he would remove the adhesive-coated slide from the trap, analyze the findings under a microscope, and post a report.

Dvorin said he will not be counting this year. After he learned his practice was purchased by a major health concern, he decided to retire. The nearest allergy bureau site to Philadelphia is in northern New Jersey.

The Asthma Center, an allergy practice in Center City, posts daily counts around 7 a.m., but those are the work of a widely used Pollen Sense AI sensor that the allergy bureau has not yet approved, said Pamela E. Gabrish, the bureau’s program director.

The sensors “show a lot of promise,” Lo said, and are likely to get better, but they evidently are having some indigestion issues.

To an extent, yes. But, Lo said, “The biggest weakness is the accuracy of pollen type identification.”

Regarding Pollen Sense, that 2022 Environmental Research study concluded that while the sensor’s data were “strongly correlated” with those of a traditional device, “the sensor’s weed and grass pollen identification algorithms require further improvement.”

The Asthma and Allergy Foundation of America uses Pollen Sense data for its annual “Allergy Capitals” reports, which include other factors such as the presence of allergy specialists in a given area.

In its report released last week, Wichita ranked No. 1 for the third year in a row (Philly was No. 77 out of 100). The foundation declined to share its data, but in one report it noted that the sensor reported high grass concentrations in a February in which temperatures were nearly five degrees below normal in Wichita.

“That’s just ridiculous,” Dvorin said.

The foundation’s research director, Hannah Jaffee, said that the Pollen Sense AI — which Lo says dominates the U.S. market — has “advanced greatly over the last few years as the network density increased.”

However, the Philadelphia machine appeared to produce some questionable results last spring. During a 14-day stretch starting on May 21, it reported “extreme levels” of grasses and trees.

“It is highly unlikely that trees and grass were extreme at the same time,” Dvorin said.

Lo is confident that the automated sensors eventually will improve, “as the data on which AI is trained gets larger and better.”

Pollen, the reproductive gametes of trees and grasses, has been a known affliction since at least the Roman Empire. Caesar Augustus was an allergy sufferer who was known to sneeze — the body’s reaction to what it perceives as a hostile invader.

The discovery that pollen grains were the sneezing impetus is credited to 19th-century British researcher Charles H. Blacksley. After a sneezing spell set off by a cloud of dust, he examined the offending particulates under a microscope and knew he was looking at pollen.

Just to be sure, he later experimented by inhaling the stuff and rubbing it in his eyes, with predictable results.

Pollen forecasts are readily available online, but be aware they are driven by algorithms based on weather forecasts and pollen histories in a given area.

Pollen.com forecasts for Philadelphia are apt to be the same as they are for surrounding areas, even though the region has quite a diversity of tree species. The homogeneity in the forecast is in part the result of the data scarcity, the company says on its site.

It notes that it is “rare to find multiple pollen counting stations throughout an entire metro area. Pollen counting is a time-consuming and costly commitment.”

Forecasts can and do go awry. Trees don’t always follow script.

The future, said Fiona Lo, may belong to the machines.

“The technology is here,” she said, “although it still needs to improve for it to be reliable and accurate. But I think that’s just a matter of time.”