Ironman Training Diary - Another 1/2 Marathon Complete

🤖 Apollo (AI) analysis

The watch split this effort into three consecutive activities — one long block, a short middle segment, and a 71-second tail. Treated as a single run, this is what came off the device:

Metric	Value
Distance	13.13 mi (21.14 km)
Duration (moving)	2:33:29
Wall-clock start → finish	06:28 → 09:16 EDT (2:47:34)
Pace	11:41/mi (7:16/km)
Weighted avg HR	170 bpm
Max HR	191 bpm
Calories	1,731
Cadence	158 spm
Elevation gain	9 m (flat — as expected in Dublin)
Z3 time	21:37 (14%)
Z4 time	1:10:00 (46%)
Z5 time	59:32 (39%)
Segments	3 (accidental split)

HR by mile (activity 1, miles 1–12): 140 → 154 → 159 → 164 → 167 → 169 → 173 → 175 → 180 → 182 → 186 → 187. Roughly +47 bpm from mile 1 to mile 12.

Weather during the run

You said the temperature and humidity were both unusually high. Half true. Air temperature during your run window was 73–75°F — not unusual for a Central Ohio July morning. Humidity is where the story lives. Open-Meteo’s ERA5 reanalysis for Dublin (40.10 N, −83.11 W) on 2026-07-04:

Hour (EDT)	Temp (°F)	RH (%)	Dew Point (°F)	Feels Like (°F)
06:00	73.5	88	69.7	81.1
07:00	74.2	87	70.0	80.9
08:00	72.6	98	72.0	80.9
09:00	73.8	98	73.2	81.3
10:00 (post)	75.7	91	72.8	81.5

Dew point is the number that matters for a runner, not relative humidity or air temperature in isolation. The generally used runner’s scale:

Dew Point (°F)	Effect on running
< 55	Comfortable
55–60	Manageable
60–65	Uncomfortable
65–70	Hard — expect slower pace at same effort
70–75	Oppressive — significant pace / HR penalty
> 75	Dangerous

You ran the entire 2:33 in 70–73°F dew-point air. Evaporative cooling — the mechanism your body relies on to shed the ~1,000+ W of heat you were producing — was substantially compromised the whole time. That is the physiologically relevant fact. The 73°F air temperature is a red herring; the sky was effectively a wet towel.

What the data shows

You covered 13.13 miles at an average moving pace of 11:41/mi with 84% of the run in Z4+Z5. That’s a benchmark result given the conditions, but read it carefully:

• HR drift was extreme. First five minutes averaged 100 bpm; last five minutes averaged 188 bpm. Even acknowledging that the first five minutes include your warm-up and the last five include a hard finish, mile-by-mile HR rose almost monotonically from 140 → 187. That is not steady-state aerobic work; that is the shape of cardiovascular strain accumulating in humidity.
• The middle “activity” (1.12 mi, 13:22) was almost entirely Z5. 751 seconds in Z5 out of 802 — 94%. Your HR handed off from activity 1 to activity 2 at ~162 bpm and averaged 183 bpm across the segment. Whatever happened to fragment the recording (watch bug, accidental pause, or you hit lap thinking it was pause), you kept running at threshold+ effort right through the gap.
• Cardiac drift signal is unambiguous. +47 bpm from mile 1 to mile 12 at pace that varied by less than 30 sec/mi across running segments is a textbook heat-and-humidity response. Your HR at mile 12 (187) was 98% of your recorded max (191). You finished in a near-max state.
• This is not a new distance PR. Your current “Run: Longest” PR is 21,611 m (13.43 mi) from April 25. Today was 21,137 m. Just under. Not a fluke that this shows up as your second-longest recorded run.
• Race predictor updated to 2:35:51 for a half. Your actual moving time today was 2:33:30 — 2:21 faster than Garmin’s next-race prediction. Read that as: on a favorable day (cool morning, dew point in the 50s), the machine thinks you should go noticeably faster than today’s number. Today’s 11:41/mi is a humidity-taxed half-marathon result, not a ceiling.
• The 13.13 miles counts as 12 running splits interrupted by short stationary aid stops to take on water and gel. Splits 3, 5, 7, 10, 12, 14, and 17 in activity 1 are 5–25 second low-speed segments — those are stops for fueling, not walking. Running-only pace between aid stops was closer to 10:07–12:57/mi across blocks. So your 11:41 “pace” is a stop-included average, not your running pace.

What the data does not show

• This was not a controlled fitness benchmark. The June 24 run was 5 miles with no aid stops and +56 bpm drift; today was 13 miles with brief aid stops for water/gel and +47 bpm drift across a much longer duration. Drift-per-hour is lower today (~19 bpm/hr) than June 24 (~56 bpm/hr). But you can’t attribute that improvement to fitness, because: (a) even brief stationary aid stops shed some HR before you resume, dampening measured drift, (b) different distance, (c) different weather (June 24 temperature was not recorded — you flagged this last time). Until you replicate on a comparable-conditions day, “I’m fitter” is a hypothesis, not a conclusion.
• The Z4/Z5 dominance (84%) does not mean you’re VO2-max training. In this heat + this humidity, your HR was elevated for a given metabolic effort. A 170 bpm average here almost certainly corresponds to a lower metabolic intensity than 170 bpm in a 55°F dew-point run. Zone-based load is inflated by conditions.
• We can’t isolate acclimation. You’ve been running through June in Ohio, so some heat adaptation should be present. But acclimation and training_readiness tables in your database are empty — Garmin either isn’t calculating these or isn’t syncing them. You have no objective heat-acclimation signal to point to.

Confounders worth naming

• Sleep the night before (7/3–7/4): 7 h 2 m total, 1 h 6 m deep, 44 m REM. Better than the June 24 baseline (26 min deep). Not a limiter.
• Wake-to-run gap: ~28 minutes. You woke around 6:00 AM local and started running at 6:28. Fasted, minimal preparation. That’s fine for a 5-miler; for a 2:33 half in oppressive dew points it’s aggressive.
• Elevation: 9 m gain / 8.5 m loss over 13 miles. As expected — Dublin is flat. This is not a run confounded by terrain.
• Temperature sensor still empty. min_temperature and max_temperature on the running metrics rows are null for all three activities. Weather data in this analysis is external (Open-Meteo ERA5 reanalysis), not from your watch.
• Why the activity split into three? The most plausible reason is that you either (a) accidentally hit stop instead of pause between miles 12 and 13, then re-started, then stopped again, or (b) the watch triggered auto-lap into new activity from a settings quirk. There’s a 4-minute gap between activity 1 and activity 2 (12:55:28 → 12:59:33 UTC) and a 2-minute gap between activity 2 and activity 3. Those gaps don’t appear in the moving-time total (2:33:29), so any recovery you got during those breaks isn’t reflected in the metrics — this actually makes your effort look slightly worse on paper than it was in reality.

Honest read

You ran a continuous half marathon — no walking, only brief stationary breaks for water and gel — in dew points that would drop most amateur race times by 5–10%. Finishing at 11:41/mi with mile 12 HR at 187 says the effort was maximal, not that the result was maximal. Do not interpret today as either “I’ve plateaued” or “I’ve made a leap.” It’s a durability-in-heat data point, and a good one — you completed the distance running the whole way, you didn’t blow up, and drift-per-hour was lower than the June 24 baseline. That is worth noting even if the causal story isn’t clean.

The May 16, 2027 Ironman 70.3 is 315 days away. Today told you two useful things: (1) you can hold 11:41/mi (aid-stops included) in oppressive humidity for 13.1 miles, (2) your HR keeps climbing on you when the dew point is above 70°F, and brief aid stops don’t reset it back to baseline. Both matter for race-day pacing.

Recommendations (moderate confidence)

• Repeat this loop on a low-dew-point morning. Target dew point below 60°F. Same fueling cadence, same 6:30 AM start. If the pace drops to 10:30–11:00/mi with drift under +30 bpm, you have direct evidence of the humidity tax. If not, look elsewhere.
• Run a HR-capped effort next. Cap at 150 for the first 60 minutes regardless of pace. Find out what pace you actually hold at aerobic-base HR in these conditions. Right now your training runs drift straight into Z5 with no imposed intensity ceiling. You’re missing the middle — sustained aerobic work at a disciplined HR.
• Add heat-acclimation intent to the plan. If your race is in a coastal or humid location, 10–14 days of morning humid-conditions runs 5–8 weeks pre-race is worth building in. Don’t cheat this by only running in AC.
• Start collecting dew point manually. The database is missing it and it’s the single most predictive weather variable for endurance-running physiology. Log start-time dew point in a note on the activity until Garmin fills the temperature field automatically.
• Verify the watch settings that caused the split. Check “Auto Lap” and “Auto Pause” configuration on the FR/Fenix. A 13-mile run getting fragmented into 3 activities is a data-integrity problem you’ll want to fix before the race.

Confidence levels: HR drift observation — high. Weather data — high (ERA5 reanalysis for Dublin coordinates). Attribution of pace penalty to humidity vs. fitness — moderate. Fitness trend vs. June 24 — low (too few controlled comparisons). VO2 max reading in the database is still empty for this window, so no independent physiological benchmark is available.

AI Race Predictor

What would you run for a half marathon in 55°F, low-humidity conditions?

Central estimate: 2:18–2:24, with ~2:21 (10:45/mi) as the middle of the range. Moderate confidence.

Here’s the derivation.

Reference point 1 — your April 25 half. You already ran a continuous 13.43 mi on 2026-04-25 in 2:33:25. That’s 11:25/mi at avg HR 178, max 193. April in Ohio is cool-ish (probably ~50–60°F, moderate dew point). Not the 55°F / low-humidity ideal, but closer to it than July 4.

Reference point 2 — today vs. April. You ran essentially the same distance in essentially the same moving time (2:33:29 vs. 2:33:25), but at avg HR 170 vs. 178 and max 191 vs. 193. That’s an ~8 bpm reduction at equivalent workload, and today’s workload was compounded by 70–73°F dew points that April didn’t have. You are meaningfully fitter now than in April. How much fitter is the harder question — I’d put it at ~5–8% aerobic capacity improvement, low confidence.

Reference point 3 — HR headroom. Today’s avg HR 170 is 89% of your recorded max 191. On a hot day with drift-out-of-control that’s what it costs you. In cool air, the same pace would sit ~5–10 bpm lower. Which means you have HR headroom to push pace rather than pay a heat tax. A race-day effort in cool conditions can reasonably target avg HR 178–183 (93–96% max) for a half — you’ve held HR 178 before (April 25).

Reference point 4 — dew point pace penalty. Empirical research on endurance pace vs. dew point (Ely, Cheuvront, and the marathon-time-vs-WBGT literature) puts the penalty at roughly 1.5–2.5% per 5°F of dew point above ~55°F for well-trained runners. You ran today at dew point ~71°F, so 16°F above the threshold. Pace penalty: 5–8%. Apply to today’s 11:41/mi (which includes aid-stop time): cool-weather equivalent would be 10:47–11:07/mi with the same aid-stop cadence. Now assume race-style fueling on the move — grab and sip from an aid station without breaking motion — and the stationary aid segments in today’s activity-1 splits (which ate roughly 25–40 sec/mi of average pace) largely disappear. Cool-weather, no-stop equivalent: 10:15–10:45/mi.

Reference point 5 — Garmin’s model. Race predictor says 2:35:51 (11:53/mi) for a half. You beat that today by 2:21 in oppressive humidity. The model is under-predicting your current fitness. Discount it.

Composite estimate:

Scenario	Pace (min/mi)	Time
Aggressive (perfect day, race motivation, aggressive pacing, PR intent)	10:20	~2:16
Realistic — race conditions, decent pacing, honest effort	10:40–10:50	~2:20–2:22
Conservative — training-pace-not-race, moderate motivation	11:00	~2:24
Bad day even in good weather	11:15	~2:28

What would push it toward the aggressive end: race-day taper, proper carb loading, a race with pacers or other runners to sit on, race-morning caffeine you don’t normally use, cool-and-overcast weather (55°F with cloud cover is faster than 55°F in direct sun).

What would push it toward the conservative end: running solo, no pacer, no carb prep, not tapered, hilly course (Dublin flat courses are among the fastest possible — a real race with rollers costs 30–60 seconds).

Notable that I’m predicting a ~10-minute improvement over today. That’s meaningful. If you’d asked me “what would your April 25 self run in ideal weather,” I’d have said ~2:26–2:30 — much closer to Garmin’s number. The reason my estimate is faster now is specifically the drop from HR 178 to HR 170 at equivalent work today.

Confidence caveats:

• Moderate confidence on the range 2:18–2:26. Two 13-mile data points (April 25 and July 4) plus HR/pace from the June 24 5-miler. That’s not a lot.
• Low confidence on the specific midpoint. Could easily be 2:19 or 2:24.
• Unknown whether you can hold sub-11:00/mi pace continuously for 13.1 miles. Your April 25 run was continuous at 11:25/mi at very high HR (178); dropping to 10:45/mi is a real intensity increase that may or may not be currently sustainable. If your endurance ceiling is still HR 178 for 2.5 hrs, and cool weather buys you 5–10 bpm of headroom, that maps more conservatively to ~11:00–11:10/mi.
• The Ironman 70.3 half-marathon is not run fresh. You’d swim 1.2 mi and bike 56 mi before it starts. The May 16, 2027 race is going to add 30–60+ sec/mi to whatever fresh-legs number you can hold. Don’t confuse “cool-weather standalone half” with “70.3 run leg.”

One provocation: if you want a real answer instead of a modeled one, sign up for a low-key October half marathon in Ohio (typical dew points 40–50°F). Race it. That’s your answer.