Growing Degree Days: The Number That Tells You What Your Plants Are Actually Doing

If you have ever wondered why your forsythia bloomed three weeks earlier than your neighbor's last year but only five days earlier this year, or why the extension service says to spray dormant oil "in late winter" without telling you what late winter actually means when February hits 55 degrees one week and 33 the next, you are asking the right question. The calendar is a terrible predictor of plant development. The thermometer is a much better one, but only if you know how to read it the right way.

Growing degree days (GDD) are that right way. They are the accumulated heat units that drive every biological event in your landscape: bud break, bloom, leaf-out, pest emergence, fruit development, leaf drop. The math takes thirty seconds. Once you understand it, you will know why spray timing windows open and close regardless of what month it is, why some years your magnolia blooms alongside your forsythia and other years it lags by two weeks, and why a frost in March matters more than a frost in January.

HortGuide tracks GDD daily for the Kent, WA reference station, and the season tracker on every plant profile page uses it to show where your landscape stands right now.

The Idea in Plain Language

Plants do not read calendars. They read temperature. Every species has a base temperature below which no meaningful growth or development occurs. Every degree of average daily temperature above that base counts as one "degree day" of development. Add those up from January 1 and you get a cumulative heat sum that tracks how much thermal energy the plant has received so far this season.

That cumulative number is what GDD measures. Think of it as a biological clock that runs on heat instead of hours. A warm January accelerates the clock. A cold snap in March slows it. But the sequence of events never changes: corneliancherry dogwood always blooms before forsythia, forsythia always blooms before star magnolia, star magnolia always blooms before redbud. The calendar dates shift from year to year, but the GDD thresholds at which these events occur remain remarkably consistent.

The Calculation

The daily GDD formula is simple:

Daily GDD = (Daily High + Daily Low) / 2 - Base Temperature

If the result is negative, you record zero for that day (plants do not "lose" development on cold days; they just stop accumulating). You add each day's value to a running total starting January 1.

For example, if the high is 52°F and the low is 38°F, the daily average is 45°F. Using a base of 32°F, that day contributes 13 growing degree days. Using a base of 50°F, that same day contributes zero, because the average did not reach the threshold.

That distinction between base temperatures is not academic. It is the central reason that GDD models built for one climate can fail completely in another, and it is why HortGuide uses a different base temperature than most published references.

Kent, WA weather station, January 5-11, 2026. Seven days of biological activity invisible to the base-50 model.

Why Base Temperature Matters (and Why We Use Base 32)

Most published GDD phenology data in the United States uses a base temperature of 50°F. The Ohio State phenology calendar developed by Dr. Daniel Herms, the University of Maryland IPMnet Pest Predictive Calendar, the UMass Extension landscape phenology tables: all base 50. That base works well in continental climates with cold winters and sharp spring transitions. In Ohio, Michigan, and Maryland, winter temperatures sit well below 50°F for months. When spring arrives, it arrives fast: temperatures jump above the threshold and GDD accumulates rapidly. The model tracks reality because the base temperature and the actual onset of biological activity happen to coincide.

This region breaks that assumption. Our maritime climate delivers winter daily averages that commonly range from 35°F to 48°F. That is above freezing. Plants and soil organisms are doing meaningful physiological work at those temperatures: root growth, cell division, slow metabolic processes that prime the spring flush. But a base-50 model registers all of that as zero. From January 1 through early March 2026, Kent accumulated only 9.9 GDD at base 50. The model said nothing was happening. Meanwhile, corneliancherry dogwood had already bloomed, red maple was in full flower, and forsythia was opening along every fence line in the valley.

The field observation that made this inescapable: star magnolia was blooming in Kent by March 5, 2026. Published research from the University of Maryland places star magnolia bloom at 89 GDD₅₀. Kent had accumulated 7.5 GDD₅₀ on that date. The model was not just wrong; it was blind. It predicted bloom would not occur until mid-May, two full months after the tree was already covered in flowers.

At base 32, the same date showed 703 GDD₃₂. That number tracked reality. The accumulated warmth from all those 35-to-48 degree winter days, invisible to the base-50 model, was exactly the heat sum driving actual plant development.

This is not a novel discovery. Researchers have long understood that the optimal base temperature depends on the organism and the climate. The 50°F convention became standard because most published phenology research originated in the Midwest and Mid-Atlantic, where it works. For maritime climates with mild winters, a lower base captures the biological reality that those "cold" days are still contributing to development.

HortGuide uses GDD base 32°F for all phenological tracking at the Kent reference station. Kent is the anchor because it is home: the Green River valley clay soils, the microclimates along the valley floor, the specific trees and shrubs in the landscape are all under daily observation year-round. Most of the field observations that calibrate this system come from walking the same routes and watching the same plants week after week. A second set of observations comes from Issaquah, at the Cascade foothills, where I keep an office. The two stations bracket the east side of the lowlands and give a sense of how elevation and proximity to the foothills shift the timing. Every threshold on every plant profile page is calibrated to the Kent base. The calibration is anchored to the March 5, 2026 star magnolia observation and validated against the known bloom sequence of 25 indicator species across the full growing season.

The Conversion Problem: Translating Published Research

Here is where honesty matters more than polish. Nearly all the pest emergence data and plant phenology thresholds published by university extension programs are in GDD₅₀. That is the data we need. But we display in GDD₃₂ because that is what tracks reality in this climate. Translating between the two is harder than it looks, and we want you to understand why.

The naive assumption is that you can apply a fixed multiplier. If 100 GDD₅₀ equals 730 GDD₃₂ in Kent, then 200 GDD₅₀ equals 1,460, right? No. The ratio between GDD₃₂ and GDD₅₀ changes dramatically through the season because the two bases accumulate at different rates depending on the daily temperature distribution.

A Real Example: Kent, March 2026

Walk through the actual numbers from the Kent weather station this year.

On January 15, Kent had accumulated 3.6 GDD₅₀ and 172 GDD₃₂. The ratio: 47.8 to 1. Almost every winter day contributed to GDD₃₂ (daily averages between 35-45°F, each day banking 3-13 degree-days above base 32) while contributing nothing to GDD₅₀ (those same averages fell below 50°F).

By February 10, the ratio had climbed to 60.9 to 1. GDD₃₂ was at 457; GDD₅₀ was at 7.5.

By March 1, the ratio peaked at 85 to 1. Kent had accumulated 638 GDD₃₂ but only 7.5 GDD₅₀. The base-50 model had barely started counting while the base-32 model had already tracked enough heat to push corneliancherry dogwood past bloom.

By late June, the ratio drops to about 4 to 1, because summer daily averages (65-80°F) are well above both thresholds and both bases accumulate rapidly. By September, it settles near 3 to 1.

The ratio is not a constant. It is a curve shaped by this region's specific temperature distribution. That is why a simple multiplier produces errors, and why the conversion has to be calibrated to local weather data.

Three Approaches We Tested

We tested three methods for converting published GDD₅₀ thresholds to GDD₃₂. The results were instructive.

The simplest approach works reasonably well for directional guidance. Method 3 is the only method that asks the right question: not "when does the math say it should happen?" but "when does it actually happen here, and what is the GDD₃₂ at that moment?"

Method 3 is the gold standard. It is also the most labor-intensive, because it requires a local observation for every species and event. We have validated 210 plant phenological thresholds using this approach. For pest profiles, we are building toward it. Every field observation logged at the Kent station, every confirmed bloom date or pest sighting, replaces a converted estimate with a measured value. The field observation system is the calibration engine.

What This Means for the Numbers You See

Every GDD₃₂ threshold on a HortGuide profile page falls into one of three categories, and we want to be transparent about which is which:

Locally observed. Anchored to a field observation at the Kent station. These are the most accurate values on the site. Star magnolia at 703 GDD₃₂, red maple at 552, forsythia at 624. As the observation dataset grows, more thresholds move into this category.

Kent-calibrated estimate. Converted from published GDD₅₀ data using the bloom-date mapping method with the Kent accumulation curve. These cover 210 plant phenological stages. Accuracy depends on how well the estimated local bloom date matches reality. They are good approximations, validated against the indicator species sequence, but not field-confirmed for each individual species.

Converted estimate. Translated from published GDD₅₀ using the piecewise ratio method. These carry a mean error of about 17%. Most pest emergence thresholds currently fall in this category. They are directionally correct (they place events in the right part of the season) but should not be treated as precise. When a pest profile shows a GDD₃₂ threshold, the source and conversion method are documented in the profile's source citation. Every season of field observations moves more thresholds from this category into the locally observed category.

Where the Source Data Comes From

The GDD thresholds on this site are not invented. They are drawn from published university research, with each value traced to a specific institution, study, and (where possible) the individual paper or fact sheet. This matters because GDD data is only as trustworthy as the research behind it.

For plant phenology, the primary source is the Ohio State University phenology calendar developed by Dr. Daniel Herms at the Dow Gardens and OSU. This calendar covers over 200 landscape plants and events, with GDD₅₀ thresholds validated over multiple years in northern Ohio. The University of Maryland Extension IPMnet Pest Predictive Calendar (Dr. Stanton Gill and colleagues) provides parallel data from the Mid-Atlantic. UMass Extension landscape phenology tables add a New England data point.

For pest emergence, HortGuide draws from more than a dozen university extension programs. Each pest profile cites its specific sources, and most cite two or more independent institutions. Apple maggot, for example, carries GDD data from MSU Extension, UC Davis IPM, and Wisconsin Extension, each reporting from different study regions. Where institutions report different thresholds, the profile documents all of them rather than silently picking one. Conflicting numbers usually reflect different base temperatures, different phenological events (first emergence versus peak emergence), or different study climates, and those distinctions matter.

UC Davis IPM is particularly valuable because their pest models often report both lower and upper developmental thresholds (the temperature range where development occurs, not just the lower bound). Their pear psylla model, developed from Yakima, WA and Medford, OR field data, is one of the few published GDD models validated in Pacific Northwest conditions.

Every source URL linked on a pest or plant profile page points to the specific page for that species, not an index page. When you click through to an MSU Extension apple maggot fact sheet or an OSU phenology event page, you land directly on the data that supports the threshold you are reading. If the source is a published research paper, the DOI link is included.

A single-source threshold could reflect the quirks of one study site. A threshold confirmed across three or four institutions in different climates represents a more robust biological constant, and calibrating those constants to this region's accumulation curve is what makes them useful here.

Year-to-Year Variability

One more piece of honesty. The GDD₃₂ accumulation curve for any single year reflects that year's weather, not a long-term average. Compare 2025 and 2026 at the Kent station:

By March 1, 2025: 404 GDD₃₂.
By March 1, 2026: 638 GDD₃₂.

That is a 58% difference by the same calendar date. A bloom threshold calibrated to 2026 data alone would predict blooms arriving much earlier than they actually did in 2025, and vice versa. Single-year calibration is a starting point, not a finish line.

The path to stable thresholds requires multi-year averaging. HortGuide has complete daily data for Kent from January 2020 through the current date, seven full years backfilled from the Open-Meteo historical archive. That dataset is now large enough to compute multi-year mean accumulation curves and year-to-year variance bounds. As those averages replace single-year calibrations, the conversion ratios and threshold estimates will reflect the range of normal variability rather than the specific pattern of any one year. The field observation system compounds this: if star magnolia blooms on March 5 in 2026 and March 18 in 2027, both observations anchor the GDD₃₂ threshold to what actually happened, and the multi-year weather data shows how the accumulation curve differed between those two seasons.

How It Works on the Site

Every plant profile on HortGuide that has GDD data includes two features powered by this system.

The Phenological Calendar

This is the table you see in the "Phenological Calendar" section of each profile. It lists the major developmental stages for that species (bud break, leaf emergence, bloom start, full bloom, petal fall, and others depending on the plant) along with the GDD₃₂ threshold at which each stage occurs and the typical date window for the Kent area.

The stages are ordered by their GDD threshold, not by calendar date. This matters for species where bloom precedes leaf-out: forsythia, star magnolia, redbud, and red maple all flower before their leaves emerge. The calendar reflects the actual sequence driven by heat accumulation, not the generic "spring timeline" you find in gardening books.

The Season Tracker

The dark green card at the top of the phenological calendar section is the season tracker. It reads the latest weather data from the Kent reference station, compares the current GDD₃₂ accumulation against the thresholds for that specific plant, and tells you:

Where you are right now. The current GDD₃₂ value and which phenological stage is active (marked "NOW" in the table) or which stage is coming next (marked "NEXT").

When the next stage is predicted to arrive. Using a 16-day weather forecast from the Open-Meteo API, the tracker projects GDD₃₂ accumulation forward and estimates the date for upcoming thresholds. Predictions within the forecast window are labeled "forecast." Predictions beyond 16 days use historical climate averages for the Kent area and are labeled "avg."

What the weather looks like. Current temperature, soil temperature, chill hours, and spray window status.

The table rows themselves reflect this status: completed stages are dimmed, the current stage has a warm amber border, and the next upcoming stage has a green border. Predicted dates appear on upcoming rows so you can plan ahead.

How GDD Predicts Pest Timing

The original reason extension researchers developed GDD phenology calendars was not to track bloom. It was to predict pest emergence. Insects are ectotherms. Their development is driven entirely by ambient temperature. Every insect pest has a GDD threshold for egg hatch, larval emergence, adult flight, and each subsequent generation.

The practical power of this system is that plant phenology and pest phenology are synchronized by the same heat accumulation. When forsythia reaches full bloom, the eastern tent caterpillar is hatching. When Vanhoutte spirea blooms, birch leafminer adults are laying eggs. When black locust flowers, bronze birch borer adults are emerging. These correlations hold across years because both the plant and the pest are responding to the same cumulative temperature signal.

For the homeowner or landscape professional, this means you can use the plants you already see blooming as indicators of pest activity windows, even without running any calculations. If your neighbor's forsythia is in full bloom, it is time to scout for tent caterpillars. If the black locust down the street is flowering, check your birches for borer activity. The GDD model is running in the background of every spring whether you measure it or not. The plants are the readout.

HortGuide will eventually tie pest emergence windows directly to the season tracker, so that each plant profile shows not just its own phenological stages but also the pest and disease management windows that coincide with them. That integration is in development. For now, the seasonal action summaries on plant profiles use month ranges calibrated to the same GDD-based timing.

The Indicator Species Sequence

The following bloom sequence holds reliably for the Puget Sound lowlands. These are the species you can watch as a living calendar for your landscape. Each one signals a temperature milestone that corresponds to specific pest and management windows.

Getting Started

You do not need a weather station to use GDD. You need a thermometer and a notebook, or just your eyes and the indicator species listed above.

The minimum approach: Watch forsythia and star magnolia. When forsythia blooms, your early-spring management windows are open. When star magnolia blooms, your dormant-spray windows are closing. Those two species alone give you the most important timing signal for residential landscape management in this region.

The thermometer approach: Record the daily high and low temperature. Average them, subtract 32, and add it to your running total (if the average is below 32, add zero). Do this daily from January 1. Compare your running total to the thresholds on HortGuide plant profiles to see where your specific microsite stands relative to the Kent reference station. South-facing walls, urban heat islands, and hilltop exposures accumulate faster. North slopes, frost pockets, and sites near large water bodies accumulate slower.

The HortGuide approach: Let us do the math. The season tracker on every plant profile page reads the Kent station data automatically and shows you the current GDD₃₂, active stage, and predicted dates for upcoming events. As we expand the station network, you will be able to select a station closer to your location. For now, the Kent station serves as a regional baseline for the Puget Sound lowlands, and the forecast-driven predictions give you a rolling look ahead at what is coming next.

---

GDD₃₂ thresholds are calibrated to the Kent, WA reference station (47.38°N, 122.23°W, Zone 8b) using 2026 field observations and projected bloom timing. All thresholds are estimates pending multi-year validation. Converted thresholds from published GDD₅₀ data carry approximately 17% mean error; locally observed thresholds are anchored to field data. The conversion methodology, validation data, and accuracy analysis are maintained in the HFG knowledge library.

Sources: Ohio State University phenology calendar (Dr. Daniel Herms, OSU/Dow Gardens); University of Maryland Extension IPMnet Pest Predictive Calendar (Harding, Klick, Shrewsbury); UMass Extension Landscape IPM Tools phenology tables; Michigan State University Extension landscape pest GDD tables; UC Davis IPM pest phenology models (Horton et al. 1992 for pear psylla; apple maggot multi-model comparison); Rutgers Plant & Pest Advisory; Wisconsin Extension landscape and fruit pest GDD tables; Penn State Extension pest fact sheets; Cornell Cooperative Extension GDD tables; Morton Arboretum Plant Health Care Reports; HortGuide Kent, WA weather station (Open-Meteo API daily data); field observations, Kent, WA, 2025-2026 seasons.