Questionable hats. Serious data. Experimental forecasts.
Somewhere behind the raft shop, a black hat is making public wagers with Mother Nature. The data are real, the hat remains under investigation, and the misses stay where everyone can see them.
Start with the plain-language call. Open the notebook when you want the snow stations, gauge behavior, analog years, model rules and all the other hydrological nerdery hiding under the hat. For current official conditions, use the Pagosa Springs River Desk.
The Black Hat Lab is Pagosa Outside’s public river-prediction experiment. It turns San Juan River history, water stored in the mountain snowpack, live flow at the Pagosa Springs gauge and incoming weather into testable forecasts. Some answers concern the whole season; others answer the question a river watcher is most likely to have today.
One timely question occupies the main workbench as the water year changes: When will snow stick? How large will the snow reserve become? When will runoff begin and peak? How strong is the season? When will tubing flow arrive? Can summer rain give the river a boost? The answer updates as new evidence arrives.
Pick any date in the next 90 days to see an experimental flow estimate, a realistic range and the odds of crossing useful river-flow thresholds. A daily graph shows the same forecast from tomorrow through day 90.
Eight predictions are published at useful moments during the water year, locked so they cannot be improved after the fact, and scored when the river supplies the answer. The scoreboard shows whether the formulas actually beat ordinary historical expectations.
The eight scored questions follow the river from one winter to the next: When will the first snow stick? How much water will the snowpack hold at peak? When will the San Juan first reach rafting levels? Will it be a good rafting season? When and how high will spring runoff peak? How long will rafting season last? When will tubing season start? How long will tubing season last? The cards keep the technical definitions and scoring rules underneath those plain-language questions. CFS means cubic feet of water passing the Pagosa gauge each second.
Three tools, three jobs. Their formulas stay separate. A changing Right Now answer never rewrites a locked game prediction, and the 90-day future-date engine is not quietly used to improve either one.
What the Lab does not decide: A flow threshold is not a trip schedule. Weather, water temperature, access, debris and daily field judgment still matter. The Lab is an experiment—not an operating decision or official forecast.
This is experimental forecasting and should not be relied on for anything other than entertainment. Use the River Desk and official sources for current conditions and published forecasts.
Start here. The Lab changes jobs with the river and puts today’s useful answer first. Open the notebook only when you want to inspect the evidence or argue with the formula.
The Lab is comparing the newest river readings with forecast headwater weather and similar moments from previous water years.
The active specialist is building its current best answer.
Waiting for enough evidence.
Waiting for enough evidence.
Waiting for the newest observations.
The forecast feed is still loading.
The Lab is finding comparable seasons.
New observations, forecast weather or a meaningful change in the river’s direction can move the live estimate.
Run the Lab again after the next source update to see whether the evidence moved.
Right Now records how a locked seasonal guess meets new evidence. It never rewrites the scoreboard prediction.
Each prediction is published before the answer is known and stays where we put it. A hit scores for the Black Hat. A miss scores for Mother Nature. Open any card for the exact hydrology and scoring rule.
Every October 1 starts a clean water-year game. A prediction is calculated from information available at its posted date or event benchmark, locked, and never moved after better evidence arrives. Right Now follows the target-specific seasonal prediction to its finish line; Ask + 90 Days and practice-squad challengers remain separate experiments. The official card cannot move.
Before October 1, all eight questions are upcoming. Once the water year opens, the tracker follows each locked prediction to its finish line without moving the original guess.
The first question asks when a meaningful headwater snow reserve will arrive and persist for a full week.
The October 1 formula uses the two in-basin SNOTEL stations and the historical arrival pattern. KCPW stays in Right Now as a live freeze-persistence challenger.
A challenger can earn a future promotion only by beating the incumbent in a no-peeking historical test.
The models below are the strongest target-specific methods the Lab has actually justified. No 2027 answer is pre-issued: each card locks on its posted date using only evidence available then.
Gold will be the locked Black Hat guess, blue is the historical reference and green will be what Mother Nature delivers. Right Now follows the same target-specific question after it locks; separate challengers may run in the background without replacing the starter.
What we’re predicting: The first date the two monitored headwater stations build a meaningful snow reserve that survives for a full week.
Status: This prediction has not been issued yet. The validated starter method is selected, but the answer will lock only with information available on its posted date.
Win condition: Land within 14 days of the first date the two-station average reaches one inch of stored water and remains above one-half inch for seven days.
What we’re predicting: The highest daily average snow-water equivalent at Upper San Juan and Wolf Creek Summit—an early clue to runoff strength and season length.
Status: This prediction has not been issued yet. The validated starter method is selected, but the answer will lock only with information available on its posted date.
Win condition: Predict the peak amount within 20% or two inches, whichever gives the wider scoring window.
What we’re predicting: The first calendar day the Pagosa gauge’s daily average reaches 400 CFS, the river’s general transition into rafting flow.
Status: This prediction has not been issued yet. The validated starter method is selected, but the answer will lock only with information available on its posted date.
Win condition: Land within 10 days of the first calendar day whose average flow reaches at least 400 CFS; a brief spike does not count.
What we’re predicting: The May 1–June 30 daily-average flow, translated into a mild, dependable average or high-water rafting season.
Status: This prediction has not been issued yet. The validated starter method is selected, but the answer will lock only with information available on its posted date.
Win condition: Predict the May 1–June 30 average within 30% or 100 CFS. Mild is below 75% of the historical average; high-water begins above 125%.
What we’re predicting: The calendar day with the highest daily-average flow from April 1 through July 31, plus that day’s mean CFS. The biggest instantaneous gauge reading is highlighted separately as the High-Water Mark, but it does not decide the contest.
Result: The April 15 card will publish one best peak-flow day, the full likely date window and a daily-mean CFS estimate rounded to the nearest 50.
Win condition: Put the best date within seven days of the actual peak-flow day and the predicted daily mean within 30% or 250 CFS, whichever gives the wider flow window.
What we’re predicting: The number of May 1–Labor Day daily averages at or above 300, 600 and 800 CFS—the three useful rafting-flow benchmarks.
Status: This prediction has not been issued yet. The validated starter method is selected, but the answer will lock only with information available on its posted date.
Win condition: Finish within 10 days on at least two of the three counts after Labor Day.
What we’re predicting: The first post-peak calendar day whose final Pagosa daily mean is below 400 CFS.
Status: One completed daily mean is the whole-day check. The Lab waits until the spring crest is likely past, then finds the first later calendar day below 400 CFS. It does not require every instantaneous reading to stay below 400.
Win condition: Land within seven calendar days of the first post-peak USGS daily mean below 400 CFS. The value normally becomes final the next day, but the recorded answer is the calendar day it describes.
What we’re predicting: The number of 30–399 CFS daily averages remaining after the automatically detected opening day, beginning with the next full calendar day and ending on Labor Day.
Status: In a 15-season replay, the simple native-flow candidates produced 10 hits in 15 years; the best combined-flow version had an 8.9-day average miss. Assuming every remaining day qualified produced 8 hits and an 11.9-day average miss.
Win condition: Finish within 10 days of the total number of daily averages from 30 through 399 CFS beginning the first full calendar day after the automatic opening and ending on Labor Day.
All eight predictions are settled. The final 2026 score is Black Hat Lab 3, Mother Nature 5.
All eight predictions are complete. Final score: Black Hat Lab 3, Mother Nature 5. Open the Film Room for every bet, the five misses and the model changes carried into 2027.
The water year opened with two rain-driven flood crests from October 10–14, 2025. The larger reached about 8,570 CFS and 12.82 feet at the Pagosa gauge—the third-largest flood in the local record. Winter then reversed the story: persistent snow arrived late, Wolf Creek Ski Area opened November 22, and the two-station headwater snowpack peaked at only 14.3 inches of stored water on March 6. Spring runoff was short and mild, and the San Juan reached its High-Water Mark at 903 CFS at 1:15 a.m. MDT on May 15 before dropping toward tubing flow unusually early. Low summer water removed many potential tubing days. The year’s flood-to-drought swing was exactly the kind of outlier that exposes formulas trained mostly on more ordinary seasons.
The official 2026 score remains Black Hat 3, Mother Nature 5. That record never moves. This separate retrospective runs the newest playbook from the information available at each issue date or event trigger. It is a learning exercise—not a replacement score and not a blind test, because the formulas were improved after the Lab had seen 2026.
What changed: The newer threshold-day counter turns Prediction 6 into a hit, and the event-triggered tubing-opening model turns Prediction 7 into a hit. Prediction 7 now scores the first post-peak USGS daily mean below 400 CFS rather than demanding a perfect 24-hour gauge window or a manual operator date. For Prediction 8, the new native-flow baseline replays 2026 at 53 days versus 38 actual. That is still 15 days high and still a miss.
The retrospective was reconstructed from information available on each posted issue date. Hits stay hits and misses stay misses. The two season-total cards reached their fixed Labor Day finish line and are now final.
What we’re predicting: When winter will stop teasing us and leave a lasting snow reserve in the San Juan headwaters.
Result: Mother Nature waited 24 days longer than our guess.
Win condition: Land within 14 days of the first date the two-station average reaches one inch of stored water and remains above one-half inch for seven days.
What we’re predicting: How much water winter will bank in mountain snow before melting begins—an early clue to runoff strength and season length.
Result: We overshot the peak by 23 inches. That miss taught the Lab not to pull an exceptionally poor snow year back toward average; the current method includes a stricter warm-and-dry guardrail.
Win condition: Predict the peak amount within 20% or two inches, whichever gives the wider scoring window.
What we’re predicting: When the river will first climb out of winter flow and reach Pagosa’s general rafting crossover.
Result: The river arrived nine days earlier than our guess—close enough to put one on the Black Hat side.
Win condition: Land within 10 days of the first calendar day whose average flow reaches at least 400 CFS; a brief spike does not count.
What we’re predicting: Whether Pagosa is headed for a high-water year locals should not miss, a dependable average rafting year or a mild season when guests should get on the river early instead of waiting for warm June weather.
Result: Only 35 CFS high. The Lab correctly warned that this would be a mild rafting year—one to catch early rather than postponing until June warmed up.
Win condition: Predict the May 1–June 30 average within 30% or 100 CFS. Mild is below 75% of the historical average; high-water begins above 125%.
What we’re predicting: One complete timestamp and the flow of the spring crest—the classic runoff question and valuable high-water planning context.
Result: The predicted crest arrived 25 hours early and the flow was 81 CFS high—a clean Black Hat win.
Win condition: Get the flow within 30% or 250 CFS and place the complete predicted timestamp within seven days (168 hours).
What we’re predicting: How many days will reach the flow benchmarks behind the season’s Splash & Dash, Mesa Canyon and West Fork planning.
Result: Only the 300-CFS total finished within 10 days of the locked guess. The card needed two of the three counts, so Mother Nature takes the point.
Win condition: Finish within 10 days on at least two of the three counts after Labor Day.
What we’re predicting: When post-peak runoff will hand the river from spring rafting to warm-weather tubing.
Result: Tubing conditions arrived 10 days earlier than our guess, putting this point on Mother Nature’s side.
Win condition: Land within seven days of the first complete post-peak 24-hour period with every available gauge reading from 30 through 399 CFS.
What we’re predicting: How many recreation-season days will land in Pagosa’s broad tubing range—an estimate of the season’s usable tubing supply.
Result: The season finished with 45 tubing-range days, 32 fewer than the locked guess. Low water—not rafting water—removed most of the missing days, so Mother Nature takes the point.
Win condition: Finish within 10 days of the total number of May 1–Labor Day daily averages from 30 through 399 CFS.
Early-season timing can be undone by a long warm, dry stretch. KCPW freeze persistence now runs beside the locked model in Right Now, but has not yet earned a fixed-card promotion.
The old tree model pulled an exceptionally poor winter back toward normal because it could not extrapolate beyond the low-snow years in its training history. This was the clearest model failure of the season.
The old May 10 issue date arrived too early to see how quickly the small runoff would collapse. The new event trigger waits for the post-peak handoff; it predicts June 3 against the automatic June 1 answer and flips this card to a hit.
Only one original threshold-day guess finished within 10 days. The new low-snow/CBRFC consistency guard replays the card at 35/17/4, close enough on all three counts.
Under the new prospective question, 98 calendar days remained after the automatic June 1 opening; 60 later averaged below 30 CFS, leaving 38 tubing-range days. The current snow-state playbook would have forecast 53. That is still 15 days high and still a miss. In the six-year CPC comparison, adding the broad precipitation tilt did not improve the hit count, so it stays an uncertainty signal rather than moving the center.
The rule going forward: Right Now may learn continuously. A challenger only enters the starting lineup after it beats the incumbent in a no-peeking historical test.
Choose a date, get the answer beside the controls and see that day highlighted immediately on the river map below. Change the date and ask again without chasing the answer up and down the page.
Once the Lab finishes loading, your answer will use the newest river and snowpack inputs.
These are experimental flow-threshold odds. Trip and product guidance is shown only for dates inside Pagosa Outside’s May 1–Labor Day operating season.
Calculating the short-range experiment.
Calculating the 30-day experiment.
Scanning the 90-day river map for the next meaningful move.
The short version: the Lab does not ask one giant formula to understand the entire river. It gives each kind of question its own job, its own evidence and its own test. This section explains what changes, what stays locked and why each prediction is made when it is.
It follows the most useful question for the present part of the water year and updates the answer as fresh snow, flow and weather observations arrive.
It estimates flow on a date chosen by the visitor and maps the full daily path from tomorrow through day 90.
It publishes eight forecasts before their outcomes are known, locks them, and measures each against a rule written in advance.
The separation is deliberate. A method that is useful for estimating next Tuesday’s flow may be poor at predicting the season’s highest crest or total number of threshold days. The 90-day engine therefore supplies only Ask + 90 Days. Right Now and the eight game cards use specialist methods built for their particular questions.
Right Now is the Lab’s active notebook. As the seasons turn, its main bench moves from lasting snow to peak snowpack, the first 400-CFS day, spring peak, May–June season strength, tubing opening and the Monsoon Rain-Flow Watch. After Labor Day it becomes the Season Review, pairing the full actual-versus-normal water-year hydrograph with the year’s snow, runoff, recreation and notable local events.
When a game card has already locked, Right Now is allowed to use measurements that arrived later. That is the point: it shows what the Lab believes today, records how the estimate moves toward the final answer and reveals which new evidence was genuinely useful. The original card remains unchanged on the scoreboard.
The late-season watch changes its emphasis with the measured river. Below 30 CFS it asks whether rain can lift flow back into tubing range; from 30–399 CFS it confirms the river is already in that broad range and watches for a useful bump; at 400+ CFS it follows the rain-fed surge and its likely return toward tubing flow. After Labor Day, Pagosa Outside is closed and the game moves to the Film Room. Right Now leads with the broader Season Review while a smaller “One More September Flow Bump?” panel continues through September 30 without making trip recommendations or creating a ninth prediction.
This engine builds one continuous daily forecast from tomorrow through day 90. Twelve separately trained forecast distances—1, 3, 5, 7, 10, 14, 21, 30, 45, 60, 75 and 90 days—form the backbone. The Lab blends the space between them, calibrates each date for its calendar month and lightly smooths neighboring days so the graph does not develop artificial corners where one forecast distance hands off to another.
Every point carries a center estimate, an 80% experimental range and rounded probabilities for tubing flow, 300+, 600+ and 800+ CFS. In chronological testing across 77,074 historical future dates, the displayed range contained the eventual flow about four times out of five. Compared with simply using the historical average, typical error improved by about 41% at seven days, 28% at 30 days and 11% at 90 days. That shrinking advantage is why the far end should be read as seasonal direction, not a calendar appointment. Sudden warmups, cold reversals and localized summer storms remain difficult.
The game follows one complete Southern San Juan water year. Each lock point balances two competing needs: enough evidence to make an informed forecast, but enough time remaining for the answer to still be useful. Six predictions use calendar dates. The final two wait for river benchmarks because a fixed date would sometimes talk about tubing in the middle of good rafting season and sometimes arrive after the transition had already happened.
The new water year begins before the headwater snow reserve usually becomes established, allowing the Lab to forecast the first meaningful snow that survives a full week.
Enough winter has passed to measure the year’s character, while February and March still leave real uncertainty about the final inches of water stored in snow.
Late-winter snow and climate signals are available, but the spring rise has usually not yet answered when Pagosa will cross its 400-CFS daily-average benchmark.
Most of the snow reserve is visible and the mild, dependable average or high-water call can be made before the May–June period being scored begins.
CBRFC’s dated spring guidance is available while the peak is normally still ahead. The card scores the peak-flow day and its daily mean; the instantaneous High-Water Mark is highlighted separately.
Opening day creates a clean May 1–Labor Day counting window for daily averages at 300+, 600+ and 800+ CFS.
The forecast waits until the crest is likely past, flow is below 800 CFS and the three-day direction is falling—late enough to focus on the handoff without writing off rafting too early. Mother Nature’s answer is the first later calendar day whose final USGS daily mean is below 400 CFS.
When that completed daily mean is detected, the Lab automatically locks the number of additional 30–399 CFS daily averages expected before Labor Day. The count begins with the next full calendar day.
The scoreboard is the honesty mechanism. Every card has a finish line and a win rule before Mother Nature answers it. Right Now may keep revising the active estimate, but the locked prediction cannot move and a near miss cannot be reclassified after the fact.
The eight questions are different mathematical problems. Snow-sticking and threshold crossings are timing problems. Peak snow and season strength are amount problems. Rafting and tubing days are counts. For each target, the Lab compares reasonable candidate methods and keeps the simplest one that improves on the ordinary historical reference.
Prediction 8 now starts with the number of calendar days remaining after the automatically detected opening, then estimates what share should stay from 30 through 399 CFS. Peak two-station SWE as a percentage of its date-normal sets the native-flow baseline: 100%+ snow years receive a higher survival rate and below-normal years a lower one. Across the 2012–2026 automatic-opening replay, simple native-flow candidates reached 10 of 15 hits with about nine days of average error; assuming every remaining day survived produced 8 of 15 hits with 11.9 days of average error.
Testing is chronological and “no peeking.” To replay a past season, the method receives only information that would have existed on that season’s issue date; later years cannot leak backward into the forecast. A challenger joins the public lineup only if it improves the target-specific historical test. Local Pagosa flow and the two snow stations inside the contributing headwaters form the foundation. Regional snow stations, high-country weather and broad NOAA outlooks are allowed only where they helped that exact target in held-out years.
The spring-peak card shows why restraint matters. The new method stops trying to outsmart the dated CBRFC guidance: the ESP 50% peak-flow forecast becomes the predicted daily mean, and the midpoint of CBRFC’s earliest-to-median peak dates becomes the single best date. The complete earliest-to-median span stays visible as the likely window. The instantaneous gauge maximum remains the card’s highlighted High-Water Mark, but it is secondary information rather than a scoring target.
The exact CBRFC replay is available for 2021–2026. It won four of six strict date-and-flow contests—67%, and the actual peak-flow day landed inside the published likely window in five of six seasons—83%. Those are encouraging archived results from a short record, not a promise about the next unusual water year.
The present design grew from mistakes as much as hits. One all-purpose model could not beat specialists at every job. Calendar dates were awkward for the late-season handoff, so tubing predictions moved to observable river benchmarks. The finish line is now a completed daily mean: once runoff has likely peaked, the first later day below 400 CFS becomes the automatic answer. That is a looser and more reproducible whole-day rule than requiring every instantaneous reading to remain below 400. Extreme low-snow years exposed formulas that pulled unfamiliar conditions too aggressively back toward normal, leading to a conservative snow-state baseline for Prediction 8. Rain-flow years showed that the late-summer question should be about the next boost and its arrival in Pagosa—not just a drought countdown.
CPC’s broad July–August precipitation tilt was also tested. It is useful context, but it did not improve the hit count in the six seasons with archived issue-date outlooks; the May 2026 outlook leaned wetter before an exceptionally dry summer. CPC’s one- and two-month outlooks therefore widen or tilt the plausible range and remain on the practice squad; they do not move the locked center yet. The Lab treats 2025 and 2026 as important low-tail examples, not automatic proof of a permanent new climate regime.
That is also why the Lab preserves two answers at once: the locked forecast shows what was knowable at the decision point; Right Now shows what becomes knowable afterward. Comparing the two gives the experiment a useful memory. A promising new signal—freeze persistence, melt energy, a regional snow pattern or forecast rain—can be tracked immediately, but it earns a future locked-card role only by improving a no-peeking replay of the same question.
U.S. Geological Survey Pagosa gauge supplies daily mean and instantaneous river flow. Its first post-peak daily mean below 400 CFS automatically settles Card 7 once the crest is likely past; the value normally becomes available the following day. Upper San Juan and Wolf Creek Summit Snow Telemetry stations measure the water stored in snow inside the contributing headwaters. KCPW at Wolf Creek Pass and KPSO near Pagosa Springs give the live Watch a high-country and lower-basin weather comparison for temperature, refreezing, humidity, wind and storm type; they do not replace the snow or river gauges. Middle Creek and Lily Pond are outside the drainage and appear only as regional pattern checks on the specific locked cards where chronological testing improved. NOAA Climate Prediction Center outlooks provide broad one- and two-month wet/near/dry context. For Card 8 they affect uncertainty only; the limited comparison did not justify moving the center estimate. National Weather Service observations and seven-day forecasts feed the live Watch. Colorado Basin River Forecast Center guidance supplies the April 15 ESP peak-flow magnitude and earliest-to-median timing window for the 2027 spring-peak card. Josh Kurz’s South San Juans streamflow writing provides an independent human-powered perspective. Beartown, Grayback and the combined outside-station snow set were tested and rejected.
The Season Review and historic-year catalog use that same USGS daily-mean record; the comparison curve is the date-matched 1991–2020 median. The Western Water Assessment rapid flood report documents the October 2025 flood sequence. The Pagosa Springs SUN records Wolf Creek’s Nov. 22, 2025 opening, while the Durango Herald records its first-in-the-nation Oct. 22, 2024 opening. Those local milestones appear only in years where a reliable source verifies them.
Trade the sparks and speculation for actual San Juan River flow, water temperature, short-term forecasts and today’s river options.
Stuff you want to know about Pagosa Outside's Black Hat Hydrology Lab Experiment
The Black Hat Hydrology Lab is Pagosa Outside’s public-facing river-forecasting experiment. It combines historical San Juan River flows, headwater snowpack, current conditions and weather information to make longer-range flow estimates and seasonal predictions. It is part river science, part local experience and part ongoing attempt to make better educated guesses than we used to make over coffee.
Read the whole backstory on our blog HERE.
The name is a wink, not a confession. In our version, “black hat” means taking public river, snowpack, weather and climate information traditionally scattered across separate sources, hacking it together in unconventional ways and testing whether it can produce a better-informed home-field advantage.
We are not breaking into government computers or hiding the methods: every source is public, every official guess is locked and every miss stays on the scoreboard. The only thing we are trying to outsmart is Mother Nature—and she still has the administrator password.
The River Desk shows current measurements, published government forecasts, snowpack, weather and today’s river options. The Black Hat Lab goes beyond those published forecasts and experiments with what might happen 30 to 90 days from now or later in the river season.
The Lab’s predictions are independent Pagosa Outside experiments—not official government forecasts. They should be treated as entertainment rather than instructions for river use, travel or business decisions. Use the River Desk for current conditions and published short-term forecasts.
The Lab estimates San Juan River flow 30, 60 and 90 days ahead and displays an experimental 90-day hydrograph. It calculates the probability of reaching Pagosa’s general tubing and rafting-flow ranges, operates a Live Seasonal Watch and publishes eight fixed predictions about important events during the water year.
Those seasonal predictions cover headwater snowpack, the beginning and strength of spring runoff, peak flow, rafting-flow days, Tubing Season Opening Day and tubing-range days.
The honest answer is: promising, but still experimental. In historical testing, the seasonal prediction methods scored 73 hits in 96 development tests. A later five-season test—using methods selected without seeing those results—scored 30 hits in 40 predictions. That works out to roughly three correct calls out of four.
The 30-day flow model reduced its typical error by approximately 28% compared with using historical averages alone, while the 90-day model improved by approximately 11%. The experimental ranges contained the actual result about four times out of five during historical testing.
Accuracy still changes considerably by season, forecast distance and type of prediction. One unusual storm, heat wave or stubborn snowpack can embarrass the entire contraption. That is why the scoreboard displays every public hit, miss and pending result instead of showing only the guesses that make us look clever.
The Lab uses river flow from the U.S. Geological Survey gauge at Pagosa Springs; snowpack from the Upper San Juan and Wolf Creek Summit Snow Telemetry stations; temperature and precipitation forecasts from the National Weather Service; selected climate outlooks from the National Oceanic and Atmospheric Administration; and historical guidance from the Colorado Basin River Forecast Center.
Middle Creek and Lily Pond may be used as regional storm-pattern checks on predictions where historical testing showed an improvement. Other stations—including Beartown and Grayback—were tested but did not improve the Pagosa predictions enough to earn a regular seat at the workbench.
The Lab does not ask a chatbot to stare into the river and invent a number. Its predictions come from repeatable formulas, historical comparisons and backtested models. The calculations run automatically, but the targets, inputs, scoring rules and operating thresholds were developed around the actual questions Pagosa river managers and guides try to answer.
The Lab compares current river flow, recent movement, seasonal timing, headwater snowpack, temperature and precipitation with similar conditions from previous water years. It produces a center estimate and an experimental range for each future date. The historical average is displayed as a baseline, not as a second prediction.
The center estimate is the Lab’s single best guess. The experimental range shows a broader group of reasonably plausible outcomes and is calibrated to contain the actual result about four times out of five during historical testing. It is not a best-case and worst-case limit, and the river is under no contractual obligation to remain inside it.
Each horizon is tested separately. A 60-day range can occasionally be wider than a 90-day range when the 60-day date lands during a volatile runoff or seasonal transition and the 90-day date lands during historically steadier flow.
The probability table estimates the chance that flow on each forecast date will fall within Pagosa’s general operating ranges:
30–399 cubic feet per second: tubing range
300 cubic feet per second or higher: possible Splash & Dash rafting flow
600 cubic feet per second or higher: possible Mesa Canyon rafting flow
800 cubic feet per second or higher: possible West Fork rafting flow
These are flow probabilities—not promises that a particular trip will operate. Water temperature, weather, debris, access and current river conditions still influence Pagosa Outside’s daily lineup.
Right Now in the Lab changes jobs as the water year progresses. During winter, Spring Runoff Watch follows headwater snowpack development and early runoff potential. As spring runoff begins rising, Predict the Peak Runoff Watch continuously updates the Lab's best guess for the peak date, time and flow. After a sustained decline indicates that the spring crest has likely passed, it becomes Tubing Opening Day Watch and looks for the first complete 24 hours with every available Pagosa gauge reading from 30 through 399 cubic feet per second. After the tubing transition, Monsoon Flow Watch follows possible rain-driven tubing boosts, low-water recoveries and returns to rafting flow. After Labor Day, it becomes Bonus Flow Watch.
Unlike the eight official predictions, Right Now in the Lab updates whenever the Lab runs and new information becomes available. Its changing estimates never rewrite the fixed guesses on the scoreboard.
The eight official predictions are:
First persistent headwater snowpack
Peak headwater snowpack
First 400-plus cubic-feet-per-second day
May–June river strength
Spring peak date, time and flow
Rafting-flow days from May 1 through Labor Day
Tubing Season Opening Day
Tubing-flow days from May 1 through Labor Day
Each question uses its own combination of snowpack, river flow, weather, climate outlooks, seasonal timing and historical comparisons. Predicting peak snowpack is not the same problem as predicting spring peak flow or tubing days, so one magic formula would not be very magical.
Each official call also has a fixed issue date selected to balance useful lead time with historical accuracy. Once published, the guess is locked and cannot be quietly changed when better information arrives.
Every prediction has a measurable finish line and scoring rule established before the outcome is known. A hit earns one point for the Black Hat Lab. A miss earns one point for Mother Nature. Unfinished predictions remain pending.
The scoreboard follows the October 1 through September 30 water year and divides the season into four quarters. A new game begins each October, while the previous season’s record remains visible. The Lab does not get to quietly drag its misses behind the shed.
A retrospective season recreates earlier predictions using only information that would have been available on each stated issue date. The model is trained on previous years and is not allowed to peek at what happened later. Retrospective results are clearly labeled because they were historical tests—not predictions publicly posted at the time.
Sometimes. Archived National Oceanic and Atmospheric Administration one-month and three-month outlooks improved selected predictions, including the first 400-plus cubic-feet-per-second day, May–June river strength and spring peak flow.
Those outlooks did not improve every part of the Lab. They are excluded from the 30-, 60- and 90-day flow estimates, hydrograph, operating-threshold probabilities and Monsoon Flow Watch because historical testing found that the other methods performed as well or better. The Live Seasonal Watch still uses the National Weather Service’s seven-day headwater forecast for short-range weather and storm information.
More data does not automatically produce a better prediction, so every new ingredient has to earn its way into the formula.
Absolutely. This is a continuing public experiment, and the build number shows which version is running. New water years create new tests, misses expose weak formulas and better data sources may improve individual predictions.
A method changes only when backtesting shows a real improvement without peeking at the answer. Past guesses and results remain visible so a new formula cannot rewrite history or polish an old miss into a hit.
Because once we put the whole river story onto one desktop, somebody asked the dangerous question: What happens after the official forecast ends—and can we guess it better than usual?
The Lab combines 25 years of river and snowpack history with current conditions, weather forecasts, tested formulas and the PO River Crew’s accumulated hunches to make longer-range and seasonal predictions. It is where actual science, river-guide instinct and old-timer theories are forced to share one questionable workbench while Mother Nature controls the scoreboard.
We built it to find out whether our educated guesses can beat ordinary historical averages—and because arguing about peak runoff at the coffee shop was not producing enough charts.
Read the whole backstory on our blog HERE.
