When Do Hatches Really Start?

Every fly fisher has a hatch chart. It might be printed on the back of a fly shop business card, laminated in a vest pocket, or pinned to a corkboard in the garage. They all say roughly the same thing: "Caddis: June–August. PMDs: July. Green Drakes: late June."

A month-based range is a starting point, but it's too coarse to plan a trip around. Caddis might emerge June 3 on one river and July 22 on another — same species, same "June–August" chart entry. The difference is water temperature, and that's what we set out to track.

We built a Growing Degree Day (GDD) engine that uses actual USGS water temperature data to predict when hatches will start on a given river in a given year. Then we validated it against expert-sourced hatch charts from fly shops, guides, and fishing publications across the country.

From Months to Days: How the Engine Works

A hatch chart gives you a month. A GDD model gives you a date — and that date updates as the season unfolds. The idea is simple: bugs need a specific amount of accumulated heat in the water before they're ready to emerge. Track the heat, and you can project when emergence will happen on a specific river in a specific year.

The GDD Model

Growing Degree Days are simple: each day, take the water temperature, subtract the species' base temperature, and add it to a running total. When the total hits the emergence threshold, the bugs are ready.

Daily GDD = max(0, water_temp − base_temp)
Emergence when: accumulated GDD ≥ threshold

For example, caddis have a base temp of 40°F and a threshold of 620 GDD. On a day when the water is 52°F, you accumulate 12 degree-days. When the running total crosses 620, expect caddis.

Beyond Simple GDD: Three Modifiers That Matter

The GDD approach works — but a simple daily-mean formula is only the starting point. After running 671,000 data points across 293 rivers and 10 years of history, we added three modifiers to the engine and measured their impact.

1. Diurnal Temperature Range (Baskerville-Emin Model)

A river that averages 42°F might spend the night at 36°F and the afternoon at 48°F. The simple-mean method says 42 − 40 = 2 GDD. But the Baskerville-Emin sine-wave model (1969) recognizes that the hours above base temp contribute more than the hours below subtract — it traces a sine curve between min and max, integrating only the area above base. The result is typically less GDD than the simple method when overnight temps drop below base, because those cold hours genuinely slow development.

94% of our stations had min/max data available. The average diurnal swing was 4.2°F — but that average hides enormous variation:

2. Flow Suppression

Even when water temperature is sufficient, high flows mechanically disrupt emergence. Nymphs can't complete the swim to the surface in raging current. We computed per-station flow normals from 10 years of USGS discharge data (93,524 daily median values across 259 stations), then applied suppression multipliers when current flow exceeds historical medians.

15.1% of all days in the dataset had active flow suppression. The impact peaks during spring runoff: in April and May, over 21% of days are flow-suppressed, reducing effective GDD by 5.4%. By August, it drops to 10%.

The suppression scales by river type. Tailwaters have regulated releases that dampen the effect. Spring creeks barely feel it at all — their flow is spring-fed and nearly constant.

3. Photoperiod Gate

This turned out to be the biggest single factor. Many aquatic insects won't emerge until day length crosses a species-specific threshold, regardless of water temperature. It's a hard biological gate — photoperiod is the most reliable seasonal signal an organism can evolve against, because it's the same every year regardless of weather.

We compute day length from latitude and solar declination for each station. In mid-March 2026, 57% of all tracked species are photoperiod-blocked — their GDD is met, but the days aren't long enough yet.

This matters because many species can't emerge until the days are long enough, no matter how warm the water gets. A pure thermal model would predict salmonflies in April on warm rivers — but photoperiod holds them until mid-May at the earliest. The gate explains why some rivers accumulate "enough" GDD weeks before anyone sees adults on the water.

The Combined Impact

Across 671,000+ data points, the enhanced model reduces effective GDD accumulation by 4.2% on average. That may sound small, but 4.2% of a 700-GDD threshold is ~30 degree-days — roughly 2–4 days of delay in early season, and 1–2 days in summer. On freestone rivers during spring runoff (April–May), the reduction hits 5.5%.

The river-type breakdown tells the real story: freestone rivers feel the full force of all three modifiers (wild temp swings + runoff + exposure). Tailwaters are buffered by dam regulation. And spring creeks are nearly immune.

An Important Caveat: GDD Marks Readiness, Not the Rise

There's a nuance in the GDD model that matters for practical fishing: the emergence threshold marks when nymphs are physiologically ready to begin the transition to adulthood — not the moment adults appear on the surface.

When we calibrated our thresholds against Silver Creek, Idaho — one of the best-documented spring creeks in the West — the model initially predicted brown drake emergence 10–14 days before anglers actually reported seeing adults. The GDD math was right: the nymphs were ready. But they were still in the nymph stage, staging in the substrate, beginning the behavioral and physiological changes that precede a hatch.

In practice, this means GDD-predicted emergence dates run roughly 1–2 weeks ahead of the fishable adult hatch. The exact gap depends on the species (tricos are fast emergers, brown drakes are slow) and water conditions. We adjusted our thresholds upward to account for this — PMDs gained +120 GDD, tricos +250 GDD based on Silver Creek field reports — but you should still treat a GDD "emergence date" as the start of a window, not the exact morning you'll see duns on the water.

The Spring Creek Exception: Why Silver Creek Breaks the Rules

Most rivers in this study are freestone or tailwater systems where water temperature is at the mercy of snowmelt, air temp, and (for tailwaters) dam releases. Spring creeks are a fundamentally different animal — and now we have the data to prove it.

Silver Creek near Picabo, Idaho is fed by underground springs that maintain water temperatures between 48–56°F year-round. While the Yellowstone River near Livingston might sit at 33°F in February, Silver Creek is already at 50°F — quietly accumulating GDD all winter.

When we ran the enhanced model across all 293 rivers, spring creeks stood out immediately:

The Metolius River (another spring creek) showed 0.0% GDD change between the simple and enhanced models for BWO. Silver Creek: also 0.0%. Compare that to Deer Creek (−17.9%), the Clearwater (−13.7%), or the McKenzie (−9.2%) — all freestone rivers hammered by flow suppression during spring runoff.

The reason is straightforward: every modifier we added targets variability, and spring creeks don't have it.

• Diurnal range: Silver Creek's daily min-to-max swing averages 3.5°F. Hot Creek (another spring creek) averages just 0.4°F. Compare Convict Creek at 14.6°F. When the swing is small, Baskerville-Emin converges to the simple mean method — there's nothing to correct.
• Flow suppression: Spring-fed flow is nearly constant. Silver Creek's average flow ratio is 1.0x median — essentially no deviation. The flow suppression multiplier stays at 1.0 (no suppression) virtually all year. The Clearwater River, by contrast, averages 0.86x in some periods and spikes above 1.5x during runoff.
• Snowpack immunity: A record snowpack year will delay hatches on the Madison by weeks as frigid melt floods the river. Silver Creek's spring-fed flow is buffered from surface snowmelt — its GDD accumulation stays on a remarkably consistent trajectory year over year.
• The GDD curve is smoother. Freestone rivers accumulate GDD in bursts (warm spell, then cold snap resets progress). Spring creeks accumulate steadily — their curves look like gentle ramps rather than staircases. This makes predictions more reliable and is why Silver Creek served as our primary calibration reference.
• Hatches come earlier. Species that don't emerge until June on freestone rivers can appear in April or May on a spring creek, because the water has been banking thermal energy since January. BWO GDD on Silver Creek was already at 619 by mid-March 2026 — double the emergence threshold — while many freestone rivers hadn't cleared 300.
• Multi-generation hatches are common. The stable warm temps mean species like callibaetis and tricos can cycle through multiple generations per season — a pattern that's rare on colder freestone water.

What This Means for You

1. Water temperature trumps the calendar

A hatch chart tells you what species exist on a river. Water temperature data — accumulated over the season — tells you when they'll actually show up. Two rivers with the same "June–August" chart entry for caddis can be weeks apart in actual emergence depending on elevation, snowpack, and water source.

2. Spring creeks and warm rivers track charts closest

Rivers with stable, warm water temperatures — spring creeks like Silver Creek, geothermally-influenced rivers like the upper Madison — accumulate GDD predictably. On cold, variable freestone rivers, the GDD engine adds the most value because conditions vary the most year to year.

3. "Month" is a starting point, not a plan

Hatch charts are useful for knowing what to expect on a river. They're less useful for knowing when to show up. A GDD-based model that tracks actual water temperature turns a month-wide guess into a day-specific projection — and our validation shows it lands within the expert-sourced window 81% of the time.

Checking Our Work: Validated Against Expert Knowledge

We didn't just build the model and ship it. We validated it against independent ground truth: 473 expert-sourced hatch charts compiled from fly shops, guides, and fishing publications across 238 rivers. These charts represent decades of accumulated angler knowledge — the best available record of when species actually emerge on specific water.

For each river-species pair where we had both a hatch chart entry and historical GDD data, we compared: does the engine's predicted emergence date fall within the chart's month window? Across 392 year-level comparisons, the answer was yes 67% of the time exactly, and 81% within ±15 days.

The Hatches That Matter Most

Nine species hit 100% accuracy against the expert charts — and they include the hatches anglers plan trips around:

• Salmonfly — the marquee Western hatch. The engine nails the timing.
• Golden stonefly — often follows the salmonfly wave by 1–2 weeks. Validated.
• Mother's Day caddis — the spring caddis emergence that kicks off the dry fly season.
• Skwala — first stonefly of the year on many Western rivers.
• Hendrickson — the classic Eastern mayfly hatch.
• Yellow sally, callibaetis, mahogany dun, October caddis — all validated.

The remaining species (PMD, green drake, BWO, trico, and others) score between 60–90%. We're continuing to refine these as we accumulate more seasons of data.

What This Looks Like in Practice

Take the Madison River salmonfly hatch — arguably the most anticipated hatch in the West. Right now, a hatch chart says "June." The GDD engine can tell you something more useful: "Salmonfly GDD is at 847 of 1,200 as of today. Based on current water temps and the 7-day forecast, projected emergence is June 14–18. Photoperiod gate clears June 2."

That prediction updates daily as new water temperature data comes in. A month out, you might see a 2-week window. Two weeks out, it narrows to days. The closer you get, the more precise it becomes — because the model is tracking what the water is actually doing, not what it did on average over the last decade.

If you're booking a guide or planning time off work, the difference between "June" and "June 14–18" is the difference between hoping and planning.

Try It

This is live in Riffle today. Pick a river, and the engine shows you where each species stands — accumulated GDD, projected emergence date, photoperiod status, and how this year compares to the last 10. The prediction updates daily as new USGS water temperature data comes in.

Check hatch predictions for any of 700+ rivers at app.riffle.fish.

Methodology

GDD thresholds are based on published entomological research, calibrated against observed emergence data from angler field reports (Silver Creek, ID as primary reference), and validated against 473 AI-researched hatch charts spanning 238 rivers. Thresholds were adjusted upward from initial entomological values because raw GDD marks nymph readiness (pre-hatch staging), which precedes the fishable adult hatch by approximately 1–2 weeks depending on species. Post-validation recalibration (March 2026) raised thresholds for 6 species and added photoperiod floor gates for 2 fall species, improving chart-match accuracy from 42% to 67% exact and 62% to 81% within ±15 days across 392 year-level comparisons. Base temperatures and emergence thresholds vary by species (e.g., midges: 32°F base, 150 GDD; caddis: 40°F base, 620 GDD; golden stoneflies: 42°F base, 1,000 GDD).

Enhanced model (March 2026): The original 11-river analysis used simple daily mean GDD. The enhanced model adds three modifiers computed across 671,516 daily records from 293 USGS stations spanning 2016–2026. (1) Baskerville-Emin sine-wave GDD uses daily min and max water temperatures (USGS statistic IDs 00001 and 00002) available at 94% of stations, avg diurnal range 4.2°F. (2) Flow suppression uses per-station per-DOY median CFS computed from 10 years of discharge data (93,524 records across 259 stations); suppression applied when daily CFS exceeds 1.5x median, scaled by river type (freestone: full effect, tailwater: 30%, spring creek: 15%). (3) Photoperiod gate uses species-specific minimum daylight hours computed from station latitude via CBM solar declination model; 15 of 23 tracked species have photoperiod gates ranging from 11.0h (skwala) to 15.5h (hex). Combined modifier impact: 4.6% GDD reduction on freestone rivers, 1.1% on tailwaters, 0.4% on spring creeks. "Chart start" is defined as the first day of the earliest month in the conventional hatch chart for that river-species pair. Mahogany dun and October caddis use hard day-of-year gates (DOY 244 / Sep 1) rather than minimum daylight gates, as these fall species are triggered by declining photoperiod. Validation (March 2026): Model predictions were compared against 473 expert-sourced hatch charts (fly shops, guides, fishing publications) compiled by Riffle's AI research engine across 238 stations. 392 year-level comparisons (species × river × year) were scored as exact (predicted emergence DOY falls within chart month window) or ±15 days (within 15 days of window boundary). Pre-calibration: 42% exact, 62% ±15d. Post-calibration: 67% exact, 81% ±15d. Nine species achieve 100% chart-match accuracy. Citizen science observations (8,261 records from GBIF and iNaturalist across 30 species) were used for species distribution validation but proved unreliable for timing validation due to systematic summer observer bias.