How Often Should a Soil Sensor Log Data? Intervals, Averaging and Data Quality

The interval is part of the measurement. A soil sensor does not become useful merely because it can produce a reading every few seconds. The buyer must decide how often the instrument measures, how often a value is written to memory, how often a packet is transmitted, and how the controller turns those values into an irrigation action. These are different intervals, and treating them as one setting can waste battery, hide short wetting events or create a dashboard full of noisy values that no operator can interpret.

This guide owns the data-logging decision: choose a practical measurement and recording cadence for the decision being made, then specify averaging, event capture, timestamps, missing-data handling and retention. The same sensor can be configured differently for different use cases, but the sampling rule must be written down before a logger is purchased.

Four intervals buyers should not confuse

Term	What it controls	Why it matters
Measurement or sampling interval	How often the sensor is read by its electronics.	A short interval can capture a rapid wetting front or control event, but may increase sensor duty cycle and power use.
Recording interval	How often a value is stored locally or in a database.	A logger may measure more frequently than it records. The stored series is the evidence available for later analysis.
Transmission interval	How often data is sent over RS485, LoRaWAN, cellular, Wi-Fi or another link.	Transmission can be less frequent than measurement. It also introduces packet loss, queueing and connectivity costs.
Control or decision interval	How often software evaluates a threshold or issues an irrigation command.	A controller may use a rolling average, hysteresis or a delay rather than reacting to every incoming point.

Greenhouse measurement guidance distinguishes measurement frequency from recording frequency, and extension guidance describes data loggers as systems that store readings at regular intervals [E01][E02]. The specification should therefore name all four intervals instead of accepting a supplier's single "update rate" as if it described the whole system.

Start with the time scale of the decision

Irrigation scheduling in an open field

For a field operator, the useful question is usually whether the root zone is approaching a refill or stress threshold, not whether the soil changed during every second of a pump cycle. A starting cadence of 15 to 60 minutes is common in connected irrigation-monitoring systems, but it is a starting point rather than a universal requirement [E03]. The final interval should reflect the speed of depletion, the irrigation event duration, the acceptable delay before action and the cost of a false trigger.

Greenhouse or fertigation events

A greenhouse may need a shorter measurement cadence around a short irrigation or fertigation pulse because the wetting front, drainage response and controller decision can occur within minutes. One practical design is to measure more frequently than the system records during normal conditions, then retain a shorter event window when a valve opens, a threshold is crossed or a rain event is detected. That design preserves useful event shape without forcing the entire season to be stored at the fastest rate. The exact values must be tested against the crop, substrate, emitter flow and sensor response.

Research and water-balance work

Research projects need a cadence that preserves the phenomenon being studied and a record of how data were integrated. A daily average can be adequate for a seasonal water-balance question but can erase the timing of infiltration, drainage or hysteresis. A project studying wetting fronts, fertigation pulses or sensor response may need sub-hour observations during the event and a slower background interval between events. The protocol should state the time base, averaging rule, missing-value rule and whether values are instantaneous, interval averages, minimums, maximums or medians.

A practical cadence matrix

Buyer task	Reasonable starting point	Do not omit
Manual scouting or low-frequency management	Record when visiting or at a declared daily/weekly check; use the same locations and depths.	Location, depth, soil condition, time and whether the reading is a spot value or a stabilized value.
Open-field irrigation threshold	15–60 minute logging is a defensible pilot range for connected systems; shorten it if the allowed delay is shorter than the interval.	Threshold hysteresis, timestamp alignment with irrigation/rainfall, and an alarm when data are stale.
Greenhouse irrigation or fertigation event	Use a faster measurement cadence around the event and a slower background record; confirm against actual wetting and drainage response.	Valve state, flow or applied volume, substrate depth, event marker and post-event equilibration.
Research or model input	Choose the interval from the model time step and the fastest process of interest, not from a default dashboard setting.	Raw data, aggregation method, calibration version, sensor status and a documented retention period.

The word "real time" is not a specification. A dashboard that refreshes every five minutes may still display a value measured ten minutes earlier, averaged over an hour or delayed by a disconnected gateway. Ask the supplier to define the age of the newest value under normal connectivity and after an outage.

Averaging, events and the meaning of a stored value

Averaging is a choice, not a neutral default. A 15-minute average can hide a brief over-wet period in a substrate, and an instantaneous reading can look erratic when the true soil state is stable. Decide whether each stored record is an instantaneous sample, a mean over the interval, a minimum, a maximum or a median, and keep that rule in the data dictionary. When both trend control and event detection matter, store a fast measurement series for the event window and a slower aggregated series for routine reporting.

Event capture deserves an explicit rule. Record what triggers it (valve state, rainfall, a threshold crossing or a manual marker), how long the elevated cadence continues, and how the event data are tagged so they can be separated from background records. Without tags, a burst of high-frequency records looks like a logging fault rather than a real wetting event.

Missing data and stale data are part of the interval decision

A system that normally records every 15 minutes but loses a gateway for six hours has a data-quality problem even if the last visible point looks plausible. Define the expected timestamp spacing, the maximum acceptable gap, the behavior during a gap and the behavior after reconnection. A controller should not interpret a repeated last value as a fresh soil measurement unless that behavior is explicitly intended and visibly labelled.

Questions that decide the logging specification

• Does the device measure at the same rate that it transmits, or can those intervals be configured independently?
• Does a record represent an instantaneous reading, an average, a minimum, a maximum or a median? Over what window?
• Can the logger retain raw data locally during a network outage, and how are queued records timestamped after reconnection?
• What is the sensor warm-up or settling time after power-up, sleep, wetting or a change of measurement range?
• What are the missing-value code, stale-data alarm, duplicate-packet rule and clock-drift behavior?
• Can the buyer export the raw time series, configuration version, battery voltage, signal strength and firmware version?
• Does the control rule use hysteresis or consecutive readings, and can the buyer inspect the values that triggered the action?

Two sourcing candidates with different roles

These are sourcing candidates for a logging workflow, not certified recommendations. Marketplace fields are stated information and must be confirmed against the quoted configuration. Supplier identity, exact wireless band, firmware behavior, calibration documentation, data retention and affiliate tracking remain separate checks.

Candidate — HONDETEC online monitoring data logger

Candidate — GemHo CHTR-7IN1-01 wireless soil sensor

HONDETEC is the clearer lead when the buyer needs a central field logger and several sensor nodes. GemHo is a lead when the buyer already has a logger or cloud path and needs a configurable wireless measurement node. Neither listing proves that the displayed interval, accuracy, server retention or radio performance will hold in the buyer's field.

Commission a logging system in three passes

First, run a short high-resolution commissioning window. Record raw readings more frequently than the planned production interval while the sensor is dry, wetted, exposed to an irrigation event and recovering. This reveals contact problems, response lag, duplicate packets, clock offsets and unexpected smoothing. Preserve the configuration used for the test.

Second, compare the logged series with independent event records. Align sensor timestamps with valve state, applied water, rainfall, crop stage and, where relevant, a manual or gravimetric reference. Check whether a change appears at the expected depth and delay. A sensor can be repeatable yet poorly placed; a logger can be reliable yet record the wrong channel.

Third, test the failure path. Disconnect the gateway or network, allow the logger to miss several transmissions, restore the connection and verify that records are neither silently lost nor assigned misleading timestamps. Then test a stale-data alarm and confirm that the controller's action is safe when the newest valid reading is older than the declared limit. Only after this pass should the production interval be accepted.

Bottom line

Do not buy a logger because its dashboard says real time. Specify the measurement, recording, transmission and control intervals separately. Use a shorter interval when the decision depends on a fast wetting or drainage event; use a slower interval when the decision is seasonal or threshold-based and the added data would only increase noise, storage and power cost. Keep raw commissioning data, define stale-data behavior and verify the complete event-to-action chain. HONDETEC is the clearer candidate for a central multi-node logging workflow; GemHo is a candidate field node for a buyer who can verify the backhaul, firmware and storage path. Both remain sourcing candidates until the quoted configuration passes the buyer's pilot.

Evidence and source notes

• E01 — Virginia Cooperative Extension, soil moisture sensors and data loggers: https://www.pubs.ext.vt.edu/BSE/BSE-338/BSE-338.html
• E02 — Plant Methods, guidelines for measuring and reporting environmental parameters in greenhouse experiments: https://link.springer.com/article/10.1186/s13007-015-0083-5
• E03 — NSW Department of Primary Industries and Regional Development, soil moisture probes and logger intervals: https://www.agtech.dpi.nsw.gov.au/blogs/blogs/soil-moisture-probes-explained
• E04 — Alibaba listing, HONDETEC online monitoring data logger, Product ID 1600294788246: https://www.alibaba.com/product-introduction/ONLINE-MONITORING-DATA-LOGGER-LORA-LORAWAN_1600294788246.html
• E05 — Alibaba listing, GemHo wireless 7-in-1 soil sensor, Product ID 1601094183062: https://www.alibaba.com/product-detail/RS485-4-20mA-Wireless-Soil-Temperature_1601094183062.html
• E06 — Journal of King Saud University, variable sampling interval for energy-efficient agricultural wireless sensor networks: https://doi.org/10.1016/j.jksuci.2018.04.010
• E07 — University of Minnesota Extension, soil sensor placement and irrigation scheduling context: https://extension.umn.edu/natural-resources/conservation/agricultural-soil-and-water/irrigation/soil-moisture-sensors-for-irrigation-scheduling