← Back

Microclimate methodology

How CropHUD uses your microclimate settings to refine disease and pest model predictions.

Most weather data comes from open-field conditions measured at standard weather stations. Actual field microclimate varies at the sub-kilometre scale depending on exposure, drainage, nearby features, and terrain. A grower who has worked a field for a season knows these things; the weather data alone doesn't. CropHUD uses microclimate settings to refine disease and pest model predictions for your specific field.

Each setting below is optional. The defaults assume neutral, open-field conditions — these are safe starting values that produce honest model output. Adjusting them gives the models more accurate inputs for your field.

Wind exposure

Wind dries leaves and soil and breaks up the boundary layer of still humid air that pathogens need to infect plants. A heavily sheltered field stays wetter for longer after rain or dew; a heavily exposed field dries out faster. CropHUD adjusts wet-period estimates and disease risk accordingly.

  • Typical — open field with normal exposure for the region. Default.
  • Heavily sheltered — surrounded on most sides by trees, buildings, or hills.
  • Sheltered — partly enclosed; one or two sides have a windbreak.
  • Exposed — situated on a ridge, plateau, or otherwise unobstructed area.
  • Heavily exposed — coastal, hilltop, or otherwise constantly windy.

Slope aspect

The direction a sloped field faces changes how much direct sunlight it receives across the day, which in turn affects soil temperature, leaf drying, and disease development. South-facing slopes (in the northern hemisphere) warm faster in spring; north-facing slopes stay cooler and damper. Flat fields receive uniform sunlight and are the default.

  • Flat — level ground or barely sloped. Default.
  • North / South / East / West (and combinations) — the compass direction the slope faces (i.e. the direction water would flow).

Slope steepness

Steeper slopes drain faster and amplify the aspect effect (a steep south-facing slope warms much more than a gentle one). On level ground, aspect is largely cosmetic. CropHUD's adjustment for aspect scales with steepness.

  • Minimal — level or barely sloped. Default.
  • Moderate — noticeable slope; surface water runs off.
  • Significant — steep enough to affect equipment access or irrigation choices.

Proximity to water

Standing water nearby raises local humidity and increases the chance of dew and fog. A field next to a small pond is slightly wetter than the regional weather suggests; a field next to a large lake or river can be substantially wetter. This directly affects the wet-period estimate the disease models depend on.

  • None — no significant standing water nearby. Default.
  • Small — pond, creek, or drainage ditch within a few hundred metres.
  • Large — river, lake, or sea within a kilometre or so.

Drainage

How quickly water moves through the soil after rain. Slow-draining fields hold moisture at the root zone and at the leaf surface for longer, extending wet periods. Well-drained fields shed water and dry faster. This setting works alongside soil type but captures the field-level reality, not just the soil's textbook behaviour.

  • Typical — neither notably fast nor slow. Default.
  • Well drained — sandy, sloped, or tile-drained; surface dries within a day after rain.
  • Slow draining — heavy clay, low-lying, or otherwise prone to puddling and saturation.

Elevation context

Where the field sits within the local terrain. Cold air pools in low spots overnight, causing late spring frosts and prolonged dew. High points dry quickly, see less frost, and have stronger winds. The regional weather report doesn't capture this — only the grower knows.

  • Typical — middle of the local elevation range. Default.
  • Low point — bottom of a valley or hollow; cold air settles here.
  • High point — top of a hill or ridge; dries quickly and stays warmer at night.

Windbreak proximity

Tree lines, hedges, and shelterbelts alter airflow at the field scale. They reduce wind speed, slow drying, and can trap humid air on their downwind side. Heavier windbreaks have larger effects. This setting overlaps with wind exposure but captures discrete features rather than overall conditions.

  • None — no significant tree lines or shelterbelts adjacent to the field. Default.
  • Some — single tree line, hedge, or building edge along one side.
  • Heavy — mature shelterbelt, woodland edge, or windbreak on multiple sides.

These settings are inputs to the models, not decisions by themselves. Adjusting them does not override the science — it calibrates it. Initial adjustment magnitudes are drawn from agricultural meteorology literature and will be refined as sensor-calibration data from CropHUD fields accumulates.

Wet-period estimation

How CropHUD decides whether leaves are wet at a given hour, and why it matters.

Most foliar disease infections require liquid water on the leaf surface for several hours. Knowing when leaves are wet — not just whether it rained — is what lets CropHUD predict disease pressure instead of just reporting weather. Standard weather stations don't measure leaf wetness directly: a CropHUD wet-period estimator infers it from the variables that are measured.

The default method is Dew-Point Depression (DPD). DPD is the gap between the air temperature and the dew point. When that gap is small, the air is close to saturation and dew forms; when it's large, the air is dry and any leaf moisture evaporates quickly. CropHUD uses two thresholds — an onset threshold below which dry leaves become wet, and a dryoff threshold above which wet leaves become dry. The gap between the two models the physical lag between condensation and evaporation. Any hour with measurable rainfall is treated as wet regardless of DPD.

DPD (Southern Ontario)

dpd_ontario

Applicable: Southern Ontario · Great Lakes climate zone

Dew-Point Depression (DPD) wet-period estimator calibrated for Southern Ontario climates. Uses onset 1.8 °C / dryoff 2.2 °C thresholds from Rao et al. 1998, validated against sensor data at Elora, Ontario by Sentelhas et al. 2008 (mean absolute error ~1.74 hours/day). Hysteresis between the two thresholds models the physical lag between leaves becoming wet and drying off. Any hour with measurable precipitation is treated as wet regardless of DPD.
Citation: Rao et al. 1998; validated Sentelhas et al. 2008 (Elora, Ontario)
Validation notes: MAE ~1.74 hours/day at Elora, Ontario (Sentelhas et al. 2008). Calibration outside Southern Ontario is uncalibrated — fall back to DPD-Generic.

DPD (Generic)

dpd_generic

Applicable: Default fallback for any region without a calibrated estimator

Conservative Dew-Point Depression (DPD) wet-period estimator used wherever a region-specific calibration has not yet been validated. Onset 2.2 °C / dryoff 2.5 °C — slightly wider than the Ontario calibration so the estimator errs toward "leaves are wet" in ambiguous cases. For disease-pressure decisions this is the safer default: missing a wet period understates risk; falsely flagging an extra dry hour as wet at most overstates it. Hysteresis between the two thresholds models the lag between leaves becoming wet and drying off. Any hour with measurable precipitation is treated as wet regardless of DPD.
Citation: DPD method (Rao et al. 1998); generic conservative thresholds pending regional calibration
Validation notes: Generic conservative thresholds. Switch to a region-specific estimator once one has been validated for your area. Pending sensor-calibration data from CropHUD growers.

Canopy LWS sensor

canopy_sensor

Applicable: Any field with a registered, active canopy sensor reporting recent readings

Direct in-canopy leaf-wetness measurement from a Bob-family NB-IoT sensor. The hour is wet when the calibrated LWS reading is at least 50% — the same threshold CropHUD applies to LWS percent-wet across the rest of the pipeline. When a reading is missing for an hour (NB-IoT dropout, dead battery, sensor failure), that single hour is delegated to the regional DPD estimator so wet-period continuity is preserved across short outages instead of voiding a whole day.
Citation: In-field LWS sensor (Bob/LilyGo T-SIM7080G family); sensor-derived wet flag replaces modeled DPD on hours where a reading is present and falls back to the configured DPD estimator on coverage gaps.
Validation notes: Per-device calibration validation pending. Sensor-source wet-flag is preferred over modeled DPD where both are available; coverage gaps within a day fall back to the field's regional DPD estimator hour-by-hour and are logged.

CropHUD picks the estimator for each field automatically based on the field's GPS location. Southern Ontario fields (roughly the Great Lakes climate zone) get the calibrated DPD-Ontario method; everywhere else gets the conservative DPD-Generic fallback. A grower can override the automatic choice on the field form if there's a specific reason to prefer a different estimator — for example, a Southern Ontario field where the calibrated thresholds underperform at a particular site, or an out-of-region field that consistently behaves like the Ontario climate.

Microclimate adjustments shift both DPD thresholds equally so the hysteresis gap is preserved. The total adjustment from all seven microclimate factors is clamped to ±1.5°C to keep grower input from overwhelming the underlying calibration. Adjustment magnitudes are initial values pending sensor calibration — they reflect agricultural meteorology principles, not yet field-validated coefficients. As CropHUD accumulates sensor-calibration data through year one and beyond, the magnitudes will be refined empirically.