Forecast Accuracy & Assumption of Risk

1.1 What "Forecast" Means in PeakScout

PeakScout produces forecasts by combining public data feeds — federal agency APIs, USGS gauge readings, SNOTEL snowpack stations, NWS grid point weather models, and state agency publications — into synthesized outputs: trail condition scores, river flow predictions, wave height estimates, campsite Go Scores, avalanche loading indicators, and activity condition ratings.

These outputs are the product of automated data aggregation and mathematical modeling. They are not issued by trained meteorologists, hydrologists, or field inspectors. They reflect the state of the public data at the time of computation — not ground-truth conditions at your specific location.

1.2 Known Sources of Forecast Error

• Sensor lag. USGS stream gauges report every 15-60 minutes. SNOTEL stations report hourly. NWS grid forecasts update every 1-6 hours. Rapidly changing conditions (flash floods, afternoon convection, dam releases) can occur and substantially alter the forecast before the next data cycle.
• Point-measurement extrapolation. Sensors measure a single point. A SNOTEL station at a sheltered clearing 8 km away does not represent snowpack on an exposed northeast-facing couloir. A USGS gauge 25 km downstream does not represent current flows at your access point.
• Grid resolution vs. terrain complexity. NWS grid forecasts and Open-Meteo weather models operate at 2.5–13 km horizontal resolution. A mountain valley, ridgeline, and adjacent slope can experience dramatically different conditions at the same forecast grid cell.
• Model divergence. When multiple weather models disagree (HRRR, NAM, GFS, ECMWF), PeakScout blends or selects outputs — any blend inherits the uncertainty of all inputs.
• Unmeasured variables. Wind redistribution of snow, aspect-specific solar radiation, drainage characteristics, soil saturation, and land management decisions (dam operations, trail clearing) are either partially modeled or not modeled at all.

1.3 What PeakScout Forecasts Do Not Include

• Field inspection of actual trail, river, or lake conditions
• Real-time wildfire behavior or smoke trajectory at your specific location
• Individual slope stability assessment (avalanche terrain)
• Rock quality, ice condition, or route-specific climbing hazards
• Wildlife behavior at a specific location on a specific day
• Real-time campsite or trailhead crowding beyond reservation data

PeakScout assigns confidence ratings to forecasts based on the proximity and recency of the underlying data source — not on the inherent predictability of the weather or conditions being forecast.

2.1 What Confidence Ratings Do Not Measure

• Accuracy probability. A "High" confidence forecast is not 90% likely to be correct. There is no statistical calibration behind the confidence rating — it reflects data proximity, not forecast skill.
• Safety level. A "High" confidence "favorable" forecast does not mean conditions are safe. It means we have good data to work with. Conditions at your specific location, aspect, and elevation may still be dangerous.
• Completeness. Confidence applies to the measured variables (temperature, flow, snowpack). Unmeasured variables (ice formation, route-specific debris, recent wind slab formation) are not reflected in any confidence rating.

Always treat Low confidence forecasts as planning hints, not planning inputs. When confidence is Low, verify directly with the managing agency, local guide services, or other recent on-the-ground reports before committing to the activity.

3.1 USGS Gauge Data Lag

PeakScout river intelligence is built on USGS stream gauge data, polled at regular intervals. The USGS National Water Information System (NWIS) typically reports instantaneous values every 15–60 minutes. Actual flow conditions at your access point may be 15 to 60 minutes ahead of what PeakScout displays. In rapidly rising flow events, this lag can represent the difference between safe and dangerous conditions.

3.2 Dam Release Variability

Many rivers in Colorado and Montana are regulated by upstream reservoirs operated by the Bureau of Reclamation (BOR), US Army Corps of Engineers (USACE), irrigation districts, or water utilities. Dam release schedules are not published in a machine-readable format that PeakScout can integrate programmatically.

• PeakScout does not have advance knowledge of dam release decisions.
• A dam release can raise river flows by 200–1,000+ CFS within minutes of the decision, far faster than USGS gauges will reflect and far faster than PeakScout's polling interval.
• Regulated rivers (e.g., below Horsetooth Reservoir, below Dillon Reservoir, below Canyon Ferry) should be treated with additional caution regardless of current gauge readings.
• Check with the relevant reservoir operator or BOR district office before any river activity on regulated waterways.

3.3 Snowmelt Model Uncertainty

PeakScout's river intelligence incorporates snowmelt-driven runoff models based on SNOTEL snowpack data and temperature forecasts. These models estimate peak runoff timing and magnitude for seasonal planning — they are not real-time flow sensors.

• Snowmelt rate is highly sensitive to solar radiation, wind, and temperature — all of which are modeled, not directly measured, at most river corridors.
• Peak runoff can arrive days earlier or later than modeled, and can significantly exceed modeled magnitude in high-snowpack years with rapid warming.
• Irrigation diversions can drop flows rapidly during summer months, particularly on heavily appropriated rivers in Colorado's Front Range and Rio Grande basins.

3.4 Flash Flood Risk

Convective thunderstorms can spike tributary flows and raise mainstem river levels within minutes — well before any upstream gauge can register the event and far before PeakScout can reflect it. Flash flood risk is highest in slot canyons, canyon bottoms, and at river crossings. Monitor NWS flash flood watches and warnings independently. Do not enter flood-prone river corridors when thunderstorms are in the upstream watershed, regardless of current gauge readings.

3.5 Wading Safety Classifications

PeakScout's wading safety classifications (safe / use caution / dangerous / impassable) are generated from USGS CFS readings cross-referenced against species and reach-specific flow thresholds. These thresholds are generalizations:

• A "safe wading" flow on a wide, shallow freestone river may be dangerously fast in a narrow bedrock canyon reach of the same river.
• Individual angler skill, wading equipment (felt vs. rubber soles, wading staff, PFD), and water temperature all affect actual safety at any given flow.
• Cold water shock and hypothermia risk are not quantified in PeakScout's wading classification.

4.1 Bretschneider Wave Model

PeakScout estimates wave height on mountain lakes using the Bretschneider wave model, which calculates significant wave height from wind speed, wind duration (fetch), and water body geometry. This is a standard oceanographic approximation method adapted for inland lakes.

Known limitations of this model on mountain lakes:

• Wind data is from the nearest NWS grid point or NOAA station — not from a sensor on or adjacent to the water surface. Mountain wind is highly variable across short distances. A ridge sheltering one end of a lake can produce calm conditions while the opposite end experiences dangerous chop.
• Fetch assumptions are based on lake geometry from geodata — actual effective fetch varies with wind direction and can change significantly within a single wind event.
• Model does not capture wind shadow, orographic acceleration, or thermals common on high-altitude lakes surrounded by terrain.
• Seiches and standing waves caused by sustained wind on small lakes are not modeled and can create conditions substantially different from the wave height estimate.
• Typical uncertainty: ±40%. A forecast of 1-foot waves may represent actual conditions ranging from 0.6 to 1.4 feet — or potentially outside that range.

4.2 Cold Water Immersion Risk

PeakScout estimates water temperature and cold-water hypothermia incapacitation times on lake pages. These estimates use USGS gage height and temperature sensor data where available, supplemented by seasonal models and adjacent river gauge temperatures.

• Lake surface temperature and temperature at depth can differ significantly — cold deep water can cause cold shock even when surface temperature appears reasonable.
• Hypothermia incapacitation time estimates are population averages. Individual variation in body composition, fitness, and experience materially affects actual cold tolerance.
• Cold water shock (cardiac response to sudden cold immersion) can occur before hypothermia regardless of incapacitation time estimates.

4.3 Craft Safety Ratings

PeakScout's lake craft safety ratings (favorable / marginal / hazardous) are derived from the modeled wave height and wind speed relative to general small-craft thresholds. They do not account for:

• Your specific vessel type, stability, and freeboard
• Paddler/operator experience and skill
• PFD availability and use
• Distance from shore and ability to self-rescue
• Rapidly deteriorating conditions not yet reflected in the wind forecast

5.1 Planning Tool, Not Safety System

PeakScout is designed to help you plan outdoor activities more effectively — not to replace the judgment, skill, and situational awareness required to execute them safely. Using PeakScout to decide when and where to go is appropriate. Using PeakScout as a substitute for reading actual conditions when you arrive is not.

The following decisions are yours alone to make, based on conditions you observe directly:

• Whether to launch a boat or kayak given actual lake conditions on arrival
• Whether to attempt a river crossing given actual current, depth, and flow you observe
• Whether to continue a hike given actual weather conditions developing around you
• Whether to descend a slope given actual snowpack and runout characteristics you assess
• Whether to set up camp given actual flooding, fire, or wildlife risk you observe on-site

5.2 Inherent Risks of Outdoor Recreation

Outdoor recreation carries inherent risks that exist regardless of forecast accuracy or data quality. These risks include but are not limited to: rapidly changing weather, terrain hazards, wildlife encounters, equipment failure, human error, flooding, avalanche, rockfall, lightning, hypothermia, and dehydration. PeakScout's forecasts do not eliminate these risks and do not reduce the likelihood that you will encounter them.

By using PeakScout to plan or conduct any outdoor activity, you acknowledge that:

• Forecast data is imperfect and may not reflect actual conditions at your location
• You are voluntarily assuming the inherent risks of outdoor recreation
• PeakScout is a planning aid and you remain personally responsible for all decisions made in the field
• No forecast — regardless of confidence rating — removes the obligation to assess conditions directly

5.3 Tie to Existing Assumption of Risk Framework

This assumption of risk disclosure supplements the existing typed-signature assumption of risk in PeakScout's Federal Land Liability Disclaimer. Users who have completed that disclaimer have explicitly acknowledged assumption of risk for federal land activities. This forecast accuracy disclaimer extends that acknowledgment to the specific limitations of PeakScout's automated forecast outputs — including river flow predictions, wave height estimates, trail condition scores, and weather forecasts across all activity types.

State-specific assumption of risk statutes that apply to forecast reliance in PeakScout's coverage areas are identified in Sections 8 (Colorado) and 9 (Montana) of this disclaimer.

6.1 Scope of This Limitation

This limitation applies to all forecast outputs produced by PeakScout, including:

• Trail condition scores and trail open/closed status estimates
• River flow predictions, wading safety classifications, and whitewater class ratings
• Wave height estimates and lake craft safety ratings
• Weather forecasts, temperature predictions, and precipitation estimates
• Avalanche storm loading indicators and danger proxies
• Mountain IQ™ scores, Go Scores, and all composite condition ratings
• Hatch predictions, fish activity forecasts, and fishing condition ratings
• Seasonal open/close estimates for roads, campgrounds, and trailheads
• Fire restriction status, road condition summaries, and AQI estimates

6.2 Relationship to Master Terms

This limitation supplements and is consistent with the full limitation of liability in PeakScout's Master Terms of Service (Section 8). The aggregate cap on PeakScout's liability — the greater of amounts paid in the prior 12 months or $100 — applies equally to forecast-related claims.

7.1 Known Reliability Issues by Source

• USGS stream gauges — Sensors can malfunction, freeze, or be washed out during flood events. Anomalous readings (sudden spikes or drops to zero) occur and PeakScout applies basic sanity filters but cannot catch all errors. USGS provisional data ("P" flag) is subject to revision after quality review.
• SNOTEL stations — Snow pillow sensors can be affected by bridging (snow sticking above the sensor), sensor drift, and telemetry outages during storms when data is most critical. SNOTEL network density in Montana is lower than Colorado — station spacing is larger and data gaps are more common.
• NWS / NOAA APIs — Government APIs experience maintenance windows, rate limiting, and occasional data feed disruptions. PeakScout caches NWS data for up to 15 minutes and may serve stale data during API outages.
• CAIC / GNFAC / WCMAC / FNAC — Avalanche center data is polled on a schedule. Mid-day updates, special danger statements, and emergency bulletins may not appear in PeakScout immediately. During rapidly evolving avalanche cycles, official center websites should be checked directly.
• CPW / FWP / Recreation.gov — Campsite availability, reservation status, and facility data can lag by 1–24 hours depending on the data source. Reservation cancellations and last-minute availability changes are often not reflected in real-time.

7.2 PeakScout's Response to Data Degradation

When PeakScout detects that a data source is unavailable or returning anomalous values, it will typically: display a reduced confidence rating, show a data availability warning on the affected page, or fall back to the next-best available data source. These degraded-state displays are intended to be transparent about data limitations — they are not a guarantee that all degradation is detected or communicated.

If you notice a forecast that appears wildly inconsistent with observed conditions, treat the observed conditions as authoritative and report the discrepancy to peakscout@polsia.app.

This disclaimer supplements PeakScout's Master Terms of Service — specifically Section 1b (Forecast Services), Section 7 (Disclaimer of Warranties), and Section 8 (Limitation of Liability). In the event of any conflict, the Master Terms control.

State-specific addenda in Sections 8 (Colorado) and 9 (Montana) supplement but do not replace the Master Terms. The Master Terms limitation of liability (§8), indemnification (§9), and dispute resolution (§10) provisions apply in full to all forecast features, condition ratings, and data displayed on PeakScout pages.

This disclaimer is one part of PeakScout's legal framework. Related disclaimers:

• Storm Intelligence Disclaimer — NWS alert attribution, storm prediction limits, alert delivery
• Federal Land Liability Disclaimer — USFS, NPS, BLM assumption of risk
• Data Sources — Full attribution for all PeakScout data integrations

For questions about this disclaimer, contact peakscout@polsia.app.