Coming next: WALTER-R (ML)
v4.0 training data now being collected
Streamlet is now recording the variables needed for a future machine-learning-assisted catchment response model: WALTER-R — Wetness, Atmosphere, Land, Terrain and Event Rainfall — Runoff/Recharge Response.
W
wetness
rainfall memory & groundwater
A
atmosphere
VPD, pressure & drying
L
land
soils, vegetation & surfaces
T
terrain
flow paths & connectivity
E
event rainfall
intensity & uncertainty
R
response
runoff, recharge & stage
Why collect this now? Future WALTER-R (ML) will need a memory of ordinary days as well as storm events. The model will eventually need to learn from live Streamlet level data, rainfall uncertainty, EA gauge age/validity, possible high-flow gauge inundation, recent NFM works in the catchment, and dry-spate behaviour after prolonged dry weather. For now, this is a data-gathering exercise — not a live ML alarm.
Garden Milesight logger
The garden logger is a Milesight EM500-SWL/SWL500 water-level sensor. It sends low-power LoRaWAN uplinks to the SenseCAP M2 gateway, then via the internet to The Things Network. A Raspberry Pi sidecar logger subscribes to the TTN MQTT feed, saves each reading to SQLite, and publishes this website panel. The timer below keeps ticking between the nominal 15-minute uplinks.
Garden rate of rise
—
RoR30 / RoR60, m/h
ETA to action level
—
Only shown when rising meaningfully
Time since last point
—
Expected about every 15 min
Signal
—
RSSI / SNR / data rate
Garden stage — last 30 days—
Footbridge action 0.900 m
Channel full 0.977 m
Overtopping 1.384 m
Catchment stream context
Preferred context: Pophams / Colaton Raleigh EA stream signal where available, even if slightly stale. Garden logger now provides the local alarm.
Context level
— m
Pophams / Colaton Raleigh preferred
Rate of rise
—
RoR30 / RoR60, m/h
Context age
—
Fresh enough for context?
Official EA alert context: Fetching current Rivers Otter and Sid alert-area status…
Why this is separate: the garden logger is the local alarm. This panel is catchment / downstream context, so it prefers the Pophams / Colaton Raleigh EA signal within a reasonable stale window rather than switching too quickly to Hayes/Budleigh surrogate. In larger events WALTER-R will also need to check whether the EA signal has become stale, hydraulically suspect or affected by inundation.
Pophams / Colaton Raleigh context — last 48 h—
Dryness index & dry-air stress
This panel keeps two related but different signals separate: the local 0–10 catchment dryness index, and VPD as a crop/plant-relevant dry-air stress indicator.
VPD source: Open-Meteo general forecast; ET₀ uses a separate UKMO Seamless feed where available.
Dry-air stress (VPD)
— kPa
VPD band
VPD in plain English: Vapour pressure deficit is how “thirsty” the air is. Higher VPD means stronger pull from leaves, wet surfaces and exposed soil, so it is relevant to crop and vegetation stress as well as catchment drying.
| VPD | Crop demand band | Practical meaning |
|---|
VPD — previous 48 h + forecast 48 hOpen-Meteo kPa
Drying pressure, weather demand & storm forecast
This section walks through the atmospheric drying story: VPD shows dry-air stress, ET₀ shows broader weather-driven evaporative demand, and ADP combines them into a daily score against a 5-year seasonal baseline. Storm rainfall and pressure context are kept here too because they describe the weather system arriving next.
Source: Open-Meteo UKMO forecast/context at Streamlet coordinates.
How Streamlet reads drying: VPD asks “how dry is the air?” ET₀ asks “how much water could the weather remove from a reference grass surface?” Because one can be high while the other is lower, Streamlet combines them into Atmospheric Drying Pressure. COSMOS soil-moisture sites below provide a station-relative regional reality check on whether soils are approaching the dry end of their own records.
Forecast rain totals
—
Next 24 h / 48 h
Max hourly rainfall
—
Max 1 h in next 24 h / 48 h
Today’s ET₀
— mm
Reference evapotranspiration
Last 24 h ET₀
— mm
Modelled reference-crop water loss
Next 24 h ET₀
— mm
Forecast reference-crop water loss
Today’s drying pressure
— /100
Daily score
Lightning nearby
—
Optional API
Pressure trend values: the 3 h, 6 h and 24 h figures show the difference between the current pressure reading and the pressure recorded that many hours earlier. Negative values mean pressure has fallen since then; positive values mean it has risen.
VPD and ET₀ are ingredients, not rival scores: VPD shows dry-air stress: how strongly the atmosphere can pull moisture from leaves, soil and wet surfaces. ET₀ estimates reference evapotranspiration: the potential water loss from a standard well-watered grass surface. The FAO Penman–Monteith method uses weather data including air temperature, humidity, wind speed, solar radiation and air pressure. Streamlet uses Open-Meteo ET₀ as a modelled atmospheric-demand input, not as measured water loss from this catchment.
Atmospheric Drying Pressure: a prototype Streamlet indicator of atmospheric evaporative demand — in plain English, how strongly recent weather has favoured drying of exposed soil, vegetation and surface litter. It combines ET₀ and daily maximum VPD as seasonal percentiles, so the score shows whether drying pressure is low or high for this time of year. It is a daily context score, not an instantaneous sensor reading or rainfall forecast.
Why this score? waiting for ET₀ and VPD ingredients.
Today’s score: waiting for ET₀ and VPD context.
Surface Drying Build-up: waiting for ADP persistence and rainfall context.
Build-up
— /100
Catchment memory
High ADP days
— /14 d
ADP ≥ 75
Dry days
— /14 d
Rain < 1 mm
Above-normal drying
—
Σ max(0, ADP−50)
Soil waxiness risk: waiting for COSMOS, local dry-period and drying-intensity context.
v3.4.9 waxiness scale: COSMOS VWC is now scored relative to each comparator station. North Wyke around 20–22% VWC is treated as close to its lower-envelope context rather than merely “20% wet”. Sydling is scored on its own drier station scale. The driest usable comparator controls the screening flag.
Soil waxiness risk
— /4
Experimental
COSMOS score
—
0–4 station-relative COSMOS dryness
Dry-period score
—
0–4 local ADP/dry days
Drying intensity
—
0–4 ET₀ / VPD / temp
How Soil waxiness risk is scored
- Regional COSMOS soil moisture: >25% VWC = 0; 20–25% = 1; 15–20% = 2; 10–15% = 3; <10% = 4.
- Local dry-period score: <3 dry days = 0; 3–6 = 0.5; 6–10 = 1; 10–14 = 1.5; 14–21 = 2.5; 21–30 = 3.5; 30+ = 4.
- Drying intensity: 7-day ET₀ ≥20 mm adds 1 and ≥30 mm adds 2; VPD ≥1.3 kPa adds 1 and ≥2.0 kPa adds 2; maximum temperature ≥25°C adds 0.5 and ≥30°C adds 1. This component is capped at 4.
- Weighting: COSMOS 40%, local dry-period 40%, drying intensity 20%. If an input is missing, available inputs are re-normalised.
- Rainfall reset: recent rain reduces the score, but intense rain does not get full reset credit because runoff or preferential flow may occur before the soil matrix fully rewets.
- Bands: <0.75 Low; 0.75–1.49 Slight; 1.50–2.39 Moderate; 2.40–3.19 High; ≥3.20 Very high.
Atmospheric Drying Pressure source: waiting for Open-Meteo ET₀ and VPD context.
Atmospheric Drying Pressure — seasonal baseline, daily valuesScore /100
How to read Atmospheric Drying Pressure
- It is not a formal drought index and does not predict flooding by itself.
- ET₀ is weighted more heavily because it already includes radiation, temperature, wind and humidity.
- VPD is included as a dry-air stress modifier.
- The value shown is a daily score. For today it combines recent/modelled hours so far with forecast hours to midnight UTC.
- Historic/recent values, today full-day estimates and forecast values are separated so forecast data is not mistaken for observed evidence.
| Index | Band | Plain-English meaning |
|---|
Surface pressure — previous 7 d + forecastUKMO / Open-Meteo hPa
ET₀ — reference evaporative demandOpen-Meteo / FAO Penman–Monteith reference ET₀, mm per preceding hour
Soil moisture response — COSMOS regional check
The atmospheric indicators above describe the weather’s drying pressure. COSMOS provides the regional soil-moisture reality check: it does not measure the Streamlet catchment directly, but it helps show whether nearby indicator soils are actually wetting, drying or holding steady.
COSMOS regional comparator loading…
