On-Course Met Station vs Grid Forecast — When Local Sensors Win

What it is

This entry documents when on-course weather station data should be preferred over a regional grid forecast, and when the grid forecast is the more reliable source. The distinction matters directly for Eagle's disease-pressure and irrigation models: a grid forecast from Met Éireann or Met Office may diverge significantly from conditions on a specific fairway or green, particularly for leaf wetness, temperature inversions, and frost.

The microclimate problem

Golf courses are not uniform environments. A single course can contain microclimates that differ by 3–5°C air temperature and 2–4 hours leaf wetness duration between:

• Low-lying greens (cold air pooling, persistent frost, extended dew) vs high-exposure tees (faster drying, earlier frost clearance)
• Sheltered greens surrounded by trees (humid, low air movement, extended leaf wetness) vs open fairways (faster canopy drying)
• North-facing slopes (shadowed, cooler, wetter) vs south-facing slopes (warmer, drying faster)

A grid forecast at 2 km or even 1 km resolution cannot resolve these intra-course differences. An on-course met station placed centrally or at the most disease-prone location captures the actual microclimate [bigga:local-weather].

When local sensors are essential

Disease pressure models:

• Microdochium leaf wetness threshold (~6 hours) — a grid forecast that estimates 4 hours of leaf wetness may miss an actual 8-hour event on a sheltered green; the difference crosses the infection threshold
• Frost prediction — grid frost forecasts can be off by 30–90 minutes relative to actual frost formation on a low-lying green; local sensors are the authoritative source for the 8" probe / 1pm decision (see frost-risk-thresholds)
• Spray wind threshold — 16 km/h limit applies at canopy level, not at grid-forecast height; a course-level anemometer near the application zone is the relevant measurement

Irrigation scheduling:

• Evapotranspiration (ET) models use local temperature, humidity, wind speed, and solar radiation; a well-calibrated on-course station producing these four inputs provides better ET estimates than the nearest synoptic station which may be 5–20 km away

When grid forecast is sufficient (or better)

• Multi-day forecast planning — on-course stations provide real-time measurement but not forecast; 3–7 day planning uses NWP (Met Éireann / Met Office MEPS or UKV) as the authoritative source
• Rainfall accumulation forecasts — grid models for rainfall are generally reliable at 24h+ horizons; local tipping-bucket sensors measure what fell but cannot forecast what will fall
• Temperature anomaly detection — when a local station reading diverges significantly from the grid forecast, it is often a sensor fault rather than a genuine microclimate; cross-checking grid vs local is a data-quality step

Equipment considerations

Common on-course met stations (UK/IE):

• Davis Vantage Pro2 and Envoy8X — most widely deployed; includes leaf wetness sensor (requires calibration)
• Prodata Weather Systems — commercial installation with API; Prodata supplies data directly to some agronomic advisory services
• Campbell Scientific — research-grade; typical for STRI course- consulting deployments

Leaf wetness sensors specifically: Standard capacitance leaf- wetness sensors (flat plate or simulated-leaf) require site-specific calibration; they measure surface wetness state (wet/dry binary or a wetness scale) rather than VWC or RH directly. Accurate leaf-wetness data is the most valuable single input for Microdochium, Dollar Spot, and Anthracnose risk models.

Eagle data integration

Eagle's disease model uses a hybrid approach:

• Grid forecast for temperature and 24h outlook
• On-course sensor data (where available) for leaf wetness, actual air temperature, and wind speed at canopy level
• Where no on-course station exists, grid relative humidity + dew point is used to estimate probable leaf wetness duration

The hybrid approach degrades model confidence when on-course data is absent; Eagle surfaces this in confidence annotations on disease- pressure outputs for courses without local sensors.

When to deviate

• Station maintenance gaps: A poorly maintained or uncalibrated on-course station produces noise worse than the grid; data quality monitoring (checking for sensor drift, blocked rain gauges) is a prerequisite for local-sensor preference
• Remote/small courses without on-site equipment: Default to grid forecast with explicit confidence downgrade in Eagle outputs

Related

• leaf-wetness-and-disease-pressure — leaf wetness as a disease-trigger input
• frost-risk-thresholds — local frost prediction vs grid forecast
• spray-weather-decisions-uk — wind threshold measured at local level
• pogo-turfpro-vwc-methodology — soil sensor complements met station data
• multi-sensor-green-health-integration-framework — how met station and soil sensor data integrate in the Eagle model