What Soil Moisture Sensor Accuracy Claims Really Mean

Accuracy is only one part of a usable measurement. A soil moisture sensor can show an attractive accuracy number and still be a poor fit for a control or monitoring job. Accuracy describes the difference from a reference under stated conditions. Resolution describes the smallest displayed or reported increment. Repeatability describes how closely repeated readings agree when the same condition is measured again. None of these terms, by itself, proves that the sensor will remain useful after it is installed in a different soil, salinity range, temperature range or communication system.

This guide gives a procurement method for comparing those claims before purchase. It focuses on the evidence behind the specification, the conditions attached to the number and the acceptance test that should be written into the RFQ.

Start with the decision the reading must support

A buyer who needs an irrigation start signal has a different requirement from a researcher who needs a defensible VWC time series. The first buyer may accept a trend signal if the trigger is stable and the site has been validated. The second needs a defined measurement unit, a reference method, a documented calibration range and a repeatability result that can be reproduced.

Buyer task	Evidence that matters first	Typical acceptance question
Irrigation control	Stable response around the dry and wet operating band; known unit and scaling; fail-safe behavior.	Does the same wetting or drying event produce the same control decision at this station?
Field monitoring	Accuracy across the expected VWC range, soil-specific effects, installation footprint and drift information.	Can the readings be compared across stations without hiding soil or installation differences?
Research or model input	Reference method, calibration equation, uncertainty or error metric, time response and metadata.	Can another operator reproduce the measurement and its conversion to VWC?
Greenhouse substrate	Evidence for rockwool, cocopeat or the named medium; electrical conductivity and temperature limits.	Was the stated accuracy measured in the actual substrate and EC range?

This first cut prevents a common procurement error: treating the largest percentage printed on a listing as the most important specification. The useful specification is the one that controls the risk in the buyer task.

Separate accuracy, resolution and repeatability

Accuracy and repeatability answer different questions. A sensor may repeat a biased value very consistently, or it may be close to a reference on average while showing too much short-term scatter for a switching threshold. Resolution is even narrower: a 0.1% display increment does not mean the measurement is accurate to 0.1%. The displayed step can be smaller than the sensor error, the calibration uncertainty or the variation caused by the installation.

Term	What it can tell you	What it cannot tell you
Accuracy	How close a reading is to a stated reference under stated test conditions.	How the sensor behaves in a different soil, depth, temperature or salinity range.
Resolution	The smallest output increment or reporting step shown by the instrument or interface.	The true uncertainty or the ability to resolve a real change in the soil.
Repeatability	How closely repeated readings agree when the condition and procedure are held constant.	Whether the repeated value is close to the reference or remains stable after installation.
Response time	How quickly the output reacts to a change at the sensing volume.	Whether the reported change represents the whole root zone or only the local probe volume.
Drift	How the output changes over time when the measured condition is held or periodically checked.	Whether a single factory calibration will remain valid for the buyer's full service life.

Ask the supplier to define the reference, range, temperature, soil or substrate, sample preparation, number of repeats and calculation method behind each number. "±3% accuracy" without those conditions is a discovery lead, not a complete performance specification.

Read the unit before reading the percentage

Soil moisture may be reported as volumetric water content, gravimetric water content, water potential, relative humidity inside a probe, a normalized percentage or an arbitrary register value. These are different measurands. A listing that says "0–100% moisture" does not establish that the output is 0–100% VWC. The controller or logger may also apply a scale factor, offset or lookup table before the value reaches the dashboard.

The RFQ should require the supplier to provide the exact output unit, register map or analog scaling, conversion equation, default calibration and any user-editable coefficients. If the unit is not stated, keep the candidate as a trend or switching input only until the supplier documents the conversion. Do not compare a 0–100% vendor display directly with a VWC specification from another sensor.

Check the range and the conditions attached to the claim

Accuracy is meaningful only inside the range used to establish it. A specification may be tighter at low VWC and wider near saturation, or may exclude high salinity and temperature extremes. A sensor that is suitable for mineral soil may require a different calibration in cocopeat or rockwool. The measurement principle also matters: FDR and capacitive sensors infer water content from dielectric behavior, so the relationship depends on the surrounding material and the installation.

Turn the claimed range into a test matrix rather than accepting one pass point. Cover at least the lower, middle and upper moisture conditions that matter to the application, and add the temperature and EC conditions that the installation will actually see. For each point, record the reference method, sensor output, repeats, mean error and spread.

Test factor	Minimum question	Why it changes the buying decision
Moisture range	Is the stated accuracy valid at the dry trigger, normal operating point and wet limit?	A sensor that performs well only in the middle may fail at the control boundaries.
Soil or substrate	Was the test performed in the named soil, rockwool, cocopeat or a documented equivalent?	The same sensor output can map to different water contents in different media.
Temperature	What temperature range and compensation method were used?	Temperature response can look like moisture change if it is not compensated or validated.
Electrical conductivity	What EC or salinity conditions were included?	Ionic effects can change the inferred dielectric or electrical response.
Installation	What insertion depth, contact, orientation and sensing volume were used?	Air gaps and local wetting create errors that a bench specification may not expose.
Time	How long was the test run and how was drift checked?	A short acceptance test cannot establish stability over a season or service interval.

Use repeatability to judge a control threshold

Repeatability is especially important when a controller acts on a narrow band. A reading that moves by more than the deadband during a stable condition can cause chatter, repeated starts or false alarms. The test must state whether the sensor was removed and reinserted, left in place while readings were repeated, or moved between samples. Those procedures measure different risks.

• Hold the reference condition steady and collect repeated readings without moving the probe. This checks electronic and short-term output scatter.
• Remove and reinstall the probe using the intended field procedure. This checks contact sensitivity and operator-dependent variation.
• Repeat the test at a dry, middle and wet condition. A single repeatability value can hide a problem near the operating limit.
• Run the same input through the complete analog or Modbus path and compare the logger value with the local instrument value. This catches scaling and rounding errors.
• Use a deadband larger than the observed noise only after confirming that the remaining resolution is adequate for the control objective.

A 2022 comparison study of low-cost resistive and capacitive soil-moisture sensors used gravimetric and volumetric calibration plus response tests after water additions [E04]. That type of paired reference and response testing is more informative than a product title that simply says "high accuracy."

Do not confuse factory calibration with field validation

A factory calibration may establish a useful starting curve, but it does not automatically transfer across soil texture, bulk density, salinity, temperature, probe placement or substrate. The University of Minnesota notes that soil moisture sensors require site interpretation and that field-capacity observations depend on the installed soil and drainage conditions [E05]. The practical procurement question is therefore not "Is it calibrated?" but "For which material and conditions was it calibrated, and how will we validate it here?"

Supplier evidence	How to classify it	Buyer action
Accuracy number with reference, range and test method	Useful stated performance claim	Request the report or datasheet revision and reproduce at key operating points.
Accuracy number without reference or conditions	Incomplete marketplace claim	Do not use it as a cross-product ranking; ask for the missing test definition.
Resolution listed in the output table	Interface or display characteristic	Confirm it is finer than the required control step, but do not treat it as accuracy.
Calibration curve or coefficients supplied for a named medium	Potentially transferable within a defined scope	Verify the medium, range, temperature and installation match the project.
"High precision" or "high sensitivity" with no metric	Marketing language	Leave the field unverified and require a measurable acceptance test.

Two candidates with different evidence profiles

The two offers below are sourcing candidates for different integration roles. They are not ranked recommendations. Their stated numbers are attributed to the supplier or marketplace listing, and they remain subject to RFQ documentation and a site acceptance test.

Candidate — Renke RS-SD-*-TR-1-T soil moisture sensor

Candidate — RIKA RK520-01 soil moisture and temperature sensor

Write an accuracy requirement the supplier can answer

A usable RFQ does not ask for "the most accurate sensor." It specifies the measurand, operating range and test conditions. For example: "Report moisture in VWC for the 15–35% range in our named substrate, with the reference method, temperature and EC conditions, repeatability after three reinsertion cycles, output scaling, response time and calibration coefficients. State the error metric and the number of samples."

Also ask which parts of the result are guaranteed and which are typical. A certificate or a marketplace badge does not transfer performance from a different model, medium or factory batch. Require the supplier to identify the exact part number, probe length, cable, connector, output and firmware or register revision that will ship.

Bottom line

Choose the sensor whose evidence matches the decision, then test the quoted configuration in the material and range where it will operate. Accuracy must have a reference and conditions. Resolution must be separated from uncertainty. Repeatability must be tested with the real installation and signal path. Until those links are documented, a marketplace accuracy number is a sourcing lead rather than a performance guarantee. Renke is a multi-output PLC/logger candidate with a documented test-data request; RIKA is an FDR probe candidate whose m³/m³ claim needs the calibration scope confirmed.

Evidence and source notes

• E01 — RIKA Sensors, RK520-01 soil moisture and temperature sensor product information: https://www.rikasensor.com/rk520-01-soil-moisture-temperature-sensor.html
• E02 — Alibaba marketplace listing, Renke soil moisture sensor 4–20 mA industrial IoT sensor for agriculture, Product ID 60583870325: https://www.alibaba.com/product-detail/2019-RS-WS-N01-TR-485_60583870325.html
• E03 — Chowdhury, Sen and Janardhanan, comparative analysis and calibration of low-cost resistive and capacitive soil moisture sensors, arXiv:2210.03019: https://arxiv.org/abs/2210.03019
• E04 — University of Minnesota Extension, soil moisture sensors for irrigation scheduling: https://extension.umn.edu/natural-resources/conservation/agricultural-soil-and-water/irrigation/soil-moisture-sensors-for-irrigation-scheduling
• E05 — RIKA Sensors, RK520-01 product information and specification basis: https://www.rikasensor.com/rk520-01-soil-moisture-temperature-sensor.html
• E06 — Alibaba marketplace listing, Renke RS-WS-N01-TR soil moisture sensor, Product ID 60583870325: https://www.alibaba.com/product-detail/2019-RS-WS-N01-TR-485_60583870325.html