Soil moisture based irrigation for farms and orchards

At a glance

Item	Typical answer
Problem	Blocks are irrigated on a fixed rota or by feel, so some fields are over-watered while others stress, pumping energy is wasted and nobody can show what each block received.
What is connected	Multi-depth soil moisture probes per irrigation block, a farm weather station, block valves, and the pump station: pump drive, pressure, flow, filter and well level.
Who uses it	The farm manager, the irrigation technician, the agronomist and, for water accounts, the owner or the irrigation association.
Typical first scope	Two blocks with contrasting soils, the pump station that feeds them and one weather station, for one season.
Time to first screen	Typically within a week of the first LoRaWAN gateway going up. Budget a full season before the rules are trusted to start pumps on their own.

The situation

Growers of orchards, vineyards, vegetables and field crops know their soils, but their irrigation decisions rest on a rota, a quick look at the canopy and a spade. Fields are large, far from power and often spread over several farms. A pump station runs one block after another, sometimes in the expensive tariff hours because that is when the operator is on site.

The consequences are familiar. Light soils at the top of a slope stress while heavy soils at the bottom stay waterlogged. Water runs below the root zone, taking fertiliser with it. A filter clogs and the far end of a drip line receives half the design flow. The irrigation association asks for abstraction figures and the farm can only offer the electricity bill. When a pump trips at 2 a.m., the block it fed is found dry in the morning.

This blueprint measures the soil at several depths and the weather on the farm over LoRaWAN, reads the pump station through a licensed gateway, and uses rules to irrigate when a block reaches its refill point, inside the cheap tariff window, with the pump station supervised throughout.

Goals and measures

• Root-zone water content per block = depth-weighted average of the probe readings across the root zone, compared with two block attributes set by the agronomist: the refill point and the full point.
• Stress hours = hours the root-zone value spent below the refill point, per block per week.
• Drainage indicator = number of irrigations after which the deepest sensor, below the roots, rose by more than an agreed amount. Rising deep moisture means water passed the roots.
• Applied depth per irrigation = m³ delivered to the block ÷ block area in hectares ÷ 10, in mm (1 mm over 1 ha is 10 m³).
• Specific energy = pump kWh ÷ m³ pumped. A rising value at the same pressure points to pump wear or a falling well level.
• Water account = season-to-date m³ against the licensed or allocated volume.
• Tariff share = share of pumped volume delivered inside the low-tariff window.

Minute, hourly and daily aggregates are kept for every key; daily volumes and energy come from the increase of the meter totals.

Architecture

In the fields

Battery soil probes in each block and a weather station on the farm report over LoRaWAN to Hermetry's own network server: EU868, OTAA, class A. One gateway on a water tower, silo or hill can cover several kilometres of open ground; in orchards and vineyards the canopy attenuates the signal in summer, so survey coverage in leaf and add a second gateway where blocks are marginal. Messages heard by several gateways are de-duplicated. Gateways speak LoRa Basics Station and need power and backhaul; on remote sites that means a small solar kit and a licensed 4G router. See LoRaWAN.

Block valves are hydraulic valves opened by latching solenoids on battery LoRaWAN valve controllers, or by wired solenoids from a licensed gateway's relay outputs where blocks are close to the pump station. Battery controllers are class A, so commands arrive after their next uplink; mains-powered controllers at the pump station can run as class C and receive commands almost at once.

At the pump station

A licensed gateway in the pump panel reads the variable-speed drive and the electromagnetic flow meter over Modbus RTU, the discharge pressure and well level on 4-20 mA inputs and the filter differential pressure switch and fertigation pump state on digital inputs, and publishes over MQTT with TLS. Models differ in inputs and links; see Hermetry Edge and the licensed devices. Where cellular coverage is weak but NB-IoT is available, a low-power cellular gateway model is an option.

Component	Role	Notes
Multi-depth soil probe (LoRaWAN)	Water content and temperature at, for example, 20, 40 and 60 cm	Codec from the library where the model is listed, or your own codec.
Farm weather station	Rain, temperature, humidity, wind, radiation	LoRaWAN, or Modbus into a gateway; one per microclimate.
LoRaWAN valve controller	Opens and closes block valves	Class A on battery; timed open preferred.
Licensed gateway at the pump station	Drive, flow, pressure, well, filter	Relay and Modbus write commands depend on model and firmware.
Pump drive and motor protection	Runs and protects the pump	Dry-run, overpressure and overload protection stay here.
LoRaWAN gateway with solar and 4G	Field coverage	Your own SIM today; eSIM management is in development.

Point list

One pump station, one weather station and, per block, a three-depth probe and a valve controller. Registers and inputs are examples.

Point	Telemetry key	Unit	Source	Interval	Notes
Soil water content, 20 cm	vwc_20	%	Soil probe	30 min	Codec key names vary by model.
Soil water content, 40 cm	vwc_40	%	Soil probe	30 min
Soil water content, 60 cm	vwc_60	%	Soil probe	30 min	Below the roots: drainage indicator.
Root-zone water content	vwc_root	%	Computed in the probe profile's root flow	30 min	Depth-weighted average.
Soil temperature, 20 cm	soil_temp_20	°C	Soil probe	30 min
Block valve open	valve_open	0/1	Valve controller	on change + 15 min	Read back, not assumed.
Air temperature	air_temp	°C	Weather station	10 min
Air humidity	air_hum	%	Weather station	10 min
Rain total	rain_total_mm	mm	Weather station counter	10 min
Rain over 24 hours	rain_24h_mm	mm	Computed in a flow	10 min
Wind speed and gust	wind_avg_ms, wind_gust_ms	m/s	Weather station	10 min	Sprinkler blocks hold in wind.
Solar radiation	solar_wm2	W/m²	Weather station, where fitted	10 min	Input for evapotranspiration.
Reference evapotranspiration	et0_mm	mm/day	Station output, where provided	daily	See Automation.
Pump running	pump_run	0/1	Drive status word, Modbus	on change + 5 min
Drive frequency	vfd_hz	Hz	Drive, holding register e.g. 40102	1 min while running
Motor current	motor_a	A	Drive	1 min while running
Drive fault code	vfd_fault	code	Drive	on change	0 = healthy.
Pump energy	pump_kwh	kWh	Drive or energy meter	15 min
Discharge pressure	discharge_bar	bar	4-20 mA on AI1	1 min while running
Main line flow	flow_m3h	m³/h	Electromagnetic meter, Modbus	1 min while running
Main line volume	volume_m3	m³	Meter totaliser	15 min	Daily and per-irrigation volume.
Well level	well_level_m	m below datum	4-20 mA on AI2	15 min
Filter dirty	filter_dp_high	0/1	Differential pressure switch on DI1	on change
Fertigation running	fert_run	0/1	Dosing pump contact on DI2	on change
Battery	battery	%	LoRaWAN devices	with each uplink

Asset model

Build the tree in Sites & Assets as the farm is managed:

• Farm (Site) with location (used for sunrise and sunset), time zone, water_allocation_m3 and tariff_window.
• Field (Area) with crop, variety, planting_year and soil_type.
• Irrigation block (Area) with area_ha, emitter_flow_lph, design_flow_m3h, and the agronomist's refill_vwc and full_vwc; the block's soil probe and valve controller are placed here.
• Pump station (Area) with the pump, well and filter as Machine assets carrying rated_kw, pump_curve_ref and well_depth_m; add Feeds relations from the pump station to the blocks it serves.

Write the refill and full points as server attributes on each block's probe as well, so one alarm rule on the probe profile compares vwc_root with each block's own limit. A Rules card homed on a field can name the probes and valves of its blocks and the pump station devices only if they are placed under the field or a common parent; home pumping rules on the farm when one station feeds several fields.

Device setup

Probes and codecs

Add probes with DevEUI, JoinEUI and AppKey or in a batch, on a Soil probe profile with the codec of the model. Check the decoded keys on the device's recent frames; map them to vwc_20, vwc_40 and vwc_60 by saving a copy of the library codec and editing it, then verify it on the codec test bench with a frame from the manual. In the profile's root flow, a script block computes vwc_root from the depth readings with weights that match the crop's root distribution, and a save block stores it.

Pump station gateway

Claim the gateway by scanning its code and assign the Pump station profile. Use remote device settings to set Modbus addresses, register lists, analog scaling and intervals, and keep a baseline per station type for other farms. Report every minute while the pump runs and every 15 minutes when it stands, if the gateway model supports conditional intervals; otherwise five minutes throughout. Gateways stamp readings with their synchronised clock and keep them during outages (capacity depends on the model); late readings are evaluated as a backfill without paging.

Commands

Define the catalog per profile: pump.start and pump.stop on the gateway (method names and whether the gateway writes the drive or switches a relay depend on its firmware), and the valve controller's open-for-minutes and close frames as raw downlinks with port and hex data from its manual. Mark pump.start high impact with a confirmation for manual use. Give every queued command an expiry shorter than the irrigation window, so a late command never starts a pump at the wrong time.

Online rules and firmware

Use the Recent message online rule with the expected interval of each device kind. Gateway firmware is signed and rolled out in waves outside the irrigation window; metered links wait unless allowed.

Dashboards

Board	Audience	Widgets and what they show
Farm map	Farm manager	Map of blocks coloured by moisture status (dry, in band, wet) from a status key the probe flow writes against each block's limits; latest weather; table of blocks with root-zone moisture, last irrigation and applied mm; alarm table.
Block page	Agronomist, farm manager	Line chart of the three depths with rain and valve state over 14 days, the classic picture of how deep each irrigation reached; root-zone moisture against refill and full points as a progress card; stress hours this week; the block's Rules card with Manual and Auto.
Pump station	Irrigation technician	Gauges for discharge pressure and flow; state cards for pump, filter and fertigation; drive frequency and current over 24 hours; well level trend; specific energy as a stat with trend; command buttons for pump start and stop behind confirmation.
Water account	Owner, association	Season-to-date m³ against allocation as a progress card; bar chart of daily m³ per block; tariff share; table export to Excel.
Weather	Farm manager	Stat cards with trend for temperature, humidity, wind and rain; compass symbol for wind direction; 24-hour rain; daily ET0 where available.

Entity parameters let one block page serve every block; a map marker click opens it. The Weather stations ready solution can install a station board to adapt.

SCADA screens

Draw the irrigation network once per farm: well with its pump and drive, filter station, fertigation unit, main line with meter and pressure transmitter, and each block with its valve and emitters. The agri family of the Food pack supplies Irrigation valve, Drip line, Sprinkler, Centre pivot, Fertigation unit, Water tank, Soil moisture probe and Weather station; the Process pack supplies Electric motor, Variable-speed drive and Magnetic flow meter; pump and tank symbols come from the core set. See the symbol library.

• States. The pump shows running, stopped and fault from pump_run and vfd_fault; block valves show open or closed; the main line pipe colours while flow is above zero; the filter blinks on filter_dp_high.
• Values. Each block prints its root-zone moisture with a colour rule against its limits; the meter prints flow and today's volume; the well prints its level.
• Commands. Valve and pump symbols can send commands behind a confirmation that names the symbol, command and values; keep the pump command to supervisors.
• 3D. A 3D view made from the drawing helps new staff learn the network before walking it.

Rules and alarms

Irrigation rules

A value trigger fires only when the value crosses its line. If a block crosses its refill point at noon, outside the tariff window, a crossing rule would never start it later. Irrigation therefore starts on a time trigger and checks the current state:

1. When: every day at the start of the low-tariff window, for example 23:00.
2. Only if: the block's vwc_root is below its refill point (the rule names the block's probe and carries the value; Rules cards compare with fixed values, while profile alarm rules read the attribute), rain_24h_mm is below a skip limit, and for sprinkler blocks wind_gust_ms is below a limit.
3. Then: open the block valve; wait until the valve reports open, with a timeout that raises an alarm and stops; send pump.start; wait until flow_m3h rises above a minimum, with a two-minute timeout that sends pump.stop and raises a Major alarm; wait the run time agreed with the agronomist; send pump.stop; wait 30 seconds; close the valve; notify the farm manager.

Blocks fed by one pump run in sequence within one rule or in rules chained with the run rule action; a rule holds up to 20 actions and a wait up to 24 hours. A Decide step can choose a longer run when vwc_root is far below the refill point. Each block, field or farm has a Manual and Auto switch; in Manual, rules stop and no rule sends commands to its devices, so the grower can irrigate by hand without interference.

Alarm	Condition	Delay and hysteresis	Severity	Notified	Escalation
Pump running without flow	Rules card: pump_run = 1 while flow_m3h below 2	Stays so for 2 min	Major (raised by the rule, which also sends stop)	Technician, SMS and push	Reminder every 15 min until acknowledged
Drive fault	Profile rule: vfd_fault != 0	Debounce 1 sample	Major	Technician, SMS	Reminder every 30 min
High discharge pressure	Profile rule: discharge_bar above attribute max_bar	Debounce 3 samples; hysteresis 0.3 bar	Critical	Technician and farm manager, SMS	Reminder every 10 min
Low pressure while running	Profile rule: discharge_bar below attribute min_bar and pump_run = 1	Debounce 3 samples	Major	Technician, push	Reminder every 30 min
Filter dirty	Profile rule: filter_dp_high = 1	Debounce 5 min	Warning	Technician, push	None
Well level low	Profile rule: well_level_m above attribute max_drawdown_m	Debounce 2 samples	Major	Farm manager, push	Severity rise to Critical at a second limit
Block below refill point	Profile rule on probes: vwc_root below attribute refill_vwc	Debounce 2 samples; hysteresis 1 %	Minor	Farm manager, in-app	None
Block too wet	Profile rule on probes: vwc_60 above attribute drain_vwc	Debounce 3 samples	Warning	Agronomist, e-mail	None
Station offline	Inactivity rule, no message for 15 min		Major	Technician, push	Reminder every 60 min
Probe or valve silent	Inactivity rule, 3 × reporting interval		Warning	Technician, e-mail	None

Lifecycle. Alarms open after their delay, are acknowledged with an optional note, for example the reason a pump tripped, and resolve when the condition clears. One alarm per device and type stays open, counting repeats. A higher severity re-notifies; reminders repeat until acknowledged. Review resolved Major alarms that nobody acknowledged each morning.

Floods and maintenance. A power cut at the pump station stops the pump, drops pressure and silences the gateway together. Keep the station-offline alarm as the leading notice and keep the consequences at lower severity. Mute devices during repairs; alarms are still recorded.

Automation and integrations

• Rain over 24 hours. A flow on the weather station: delta of rain_total_mm, statistics block summing a 24-hour time window, save as rain_24h_mm.
• Evapotranspiration. Use the station's own ET0 output if it has one. Computing FAO-56 reference ET in a flow script from daily aggregates is possible, but it is your code to validate and maintain.
• Restarts after a power cut. Hermetry does not restart pumps by itself after a trip. A rule can notify the technician when the pump station comes back online, and the technician decides.
• Farm software. alarm.* and rpc.completed or rpc.failed webhooks, HMAC-signed, record pump events and irrigation commands in farm management software; a flow can post each completed irrigation (block, start, end, m³) to it over HTTP.
• Reports. Water accounts and applied depths come from history tables exported to Excel; scheduled reports are not built in today.

Engineers can open a licensed router's web panel from its device page through router remote access: outbound tunnel, time-limited and audited sessions, and a confirmation with a data cap on metered links.

Roles and access

Who	Role or group	Can
Farm manager	User + IoT operator group	See everything, acknowledge alarms, run rules now, switch blocks to Manual, send basic commands.
Irrigation technician	User + IoT operator group	Pump station and devices, acknowledge, add notes.
Agronomist	User + IoT viewer group	Read boards and history, export; refill and full points changed by an admin on request.
Irrigation engineer	Admin	Profiles, rules, alarm rules, drawings, command catalog, gateway settings and firmware.
Owner	Owner	Members, sign-in policy, high-impact and raw commands.

A cooperative or service company managing several farms can run each farm as its own organisation, isolated by row-level security, with one account holding memberships in several. Rule changes, Manual switches, high-impact commands and setting changes are written to the audit log.

Sizing and data

As an order of magnitude for a farm with 40 blocks, two pump stations and two weather stations, counting one message per device per reading as a conservative estimate:

• Soil probes: 40 × 48 a day = 1,920. Valve controllers: 40 × 96 = 3,840.
• Weather stations: 2 × 144 = 288.
• Pump station gateways: one message a minute for 10 running hours and every 15 minutes otherwise, about 650 a day each: 1,300.
• Total: roughly 7,400 messages a day, about 225,000 a month; with about 15 keys per pump station message, roughly 30,000 values a day.

This is small for the platform. Battery life of probes and valve controllers, and LoRaWAN airtime with many devices on one gateway, are the real limits on reporting intervals. Raw readings are kept for at most 90 days, so the agronomist compares seasons on hourly and daily aggregates: choose a class that covers at least two seasons (the 1,095-day class keeps three years), and export raw data you need for longer.

Rollout plan

Phase	Timing	Work	Exit criteria
0. Survey	Before the season	Choose two contrasting blocks; radio survey in leaf; pump station register maps; agree refill and full points and the tariff window.	Signed point list and limits per block.
1. Measure	First month	Probes, weather station, pump station gateway; boards; alarm rules; irrigation still by hand.	Probe readings respond to irrigation and rain as expected; pump station values match the panel.
2. Advise	Second month	Irrigation rules in notify-only form: they tell the manager which blocks to irrigate tonight.	Manager agrees with most recommendations; thresholds adjusted.
3. Act	Rest of the season	Rules open valves and start pumps in the tariff window; Manual used for exceptions.	No missed or unwanted irrigation for three weeks; dry-run test passed.
4. Scale	Next season	Remaining blocks and farms with baselines and shared profiles.	Water account and stress hours reviewed monthly.

Commissioning checklist

1. Probes installed with good soil contact at the agreed depths, in a representative spot within the wetted zone; installation photos kept with the block records.
2. Decoded depths checked against the probe's manual; vwc_root computed as intended.
3. Weather station rain checked with a known volume; wind direction aligned to north.
4. Pump station values compared with the panel at two operating points; units and scaling correct.
5. Each valve opened and closed from Hermetry; latency of class A controllers noted; command expiry set.
6. Irrigation rule run once with Run now with the pump isolated; the run log shows each step.
7. Dry-run alarm tested by starting the pump with the block valve closed only if the drive's own protection is active; otherwise simulate the flow value.
8. Uplink removed at the pump station for 20 minutes: offline alarm; buffered readings fill the gap.
9. Drawing bound with an empty Problems tab.

Pitfalls

• Crossing triggers for scheduling. A block that crosses its refill point at noon never starts at night. Start on time, check the state.
• Probes in the wrong place. A probe outside the wetted bulb of a drip line reads dry forever. Install in the wetted zone and check the response to the first irrigation.
• One threshold for all soils. Refill points depend on soil and crop. Keep them as block attributes set with the agronomist.
• Relying on Hermetry for pump protection. The rule that stops a pump without flow is a backup. Dry-run and overpressure protection belong in the drive and panel.
• Late commands. A queued start delivered after the tariff window, or a class A close delivered late, costs money and water. Use expiries and timed opens.
• Summer coverage. A survey in winter overestimates range through orchards. Survey in leaf.
• Ignoring the deep sensor. If 60 cm rises after every irrigation, runs are too long, whatever the surface sensor says.

What Hermetry does not replace

Hermetry measures, decides and supervises. Pump protection (dry-run, overload, overpressure), pressure relief, backflow prevention, fertigation dosing interlocks and the pump controller's own safe states stay in the field and must work when the link or the platform is down. A command from Hermetry is a request that a controller can fail to carry out, so every valve and pump state is read back. Agronomic decisions remain the agronomist's; abstraction metering for regulators and farm accounts stay in their systems of record.