Reading the data

A soil moisture reading means nothing until you know three other numbers

Two farms outside Naivasha can see the same figure on the same dashboard on the same morning. One should irrigate today. The other should wait four days. Neither of them is misreading the sensor.

Manuel Lopes 10 August 2026 7 min read

A soil moisture sensor does not measure water. It measures how the soil around it responds to an electrical field, and water happens to be the thing that changes that response most. The number you see on a dashboard is the end of a chain of conversions, and somewhere in that chain the units stopped being physical.

This matters because the number on its own cannot tell you whether to irrigate. It can only tell you where you are between two limits, and those two limits are properties of your soil, not of the sensor. Most of the frustration we hear from growers who bought moisture sensors and stopped using them comes down to this one gap.

What the sensor is actually doing

Capacitive probes, which is what almost every affordable field sensor uses, exploit a large contrast in dielectric permittivity. Dry mineral soil sits somewhere around 2 to 6. Water sits near 80. Push an oscillating field through the soil, watch how the circuit responds, and the result tracks water content closely enough to be useful.

Closely enough is doing a lot of work in that sentence. The response also shifts with soil texture, bulk density, temperature, and salinity. A probe reading 62 in a sandy loam and a probe reading 62 in a heavy clay are not reporting the same amount of water, and neither of them is reporting volumetric water content unless somebody calibrated them against gravimetric samples from that specific soil.

There is a quick sanity check you can run on any dataset. Mineral soils saturate somewhere between roughly 40 and 55 percent volumetric water content, because the rest of the volume is solid particles. If your sensor is reporting values in the seventies and eighties, it is reporting percent of its own scale, not percent of soil volume. That is a legitimate way to run a sensor. It just is not the thing people assume they are looking at.

Horizontal bar divided into four zones: water locked below wilting point, a stress zone, the readily available water band, and water above field capacity that drains away under gravity.
Only the green band is water your crop can take up without working for it. Everything to the left is expensive to extract, everything to the right is on its way to the water table. Diagram by Soil Sensors Kenya. Zone definitions follow FAO Irrigation and Drainage Paper 56.

The first two numbers: field capacity and wilting point

After heavy rain or a full irrigation, soil drains for a day or two and then stops. What it holds at that point is field capacity. Water above it cannot be held against gravity and leaves, taking dissolved nitrogen with it. At the other end, there is a water content below which roots can no longer pull quickly enough to keep up with transpiration. That is the wilting point. The gap between the two is the total available water, and FAO Irrigation and Drainage Paper 56 is where the standard definitions and the arithmetic come from.

Both limits move with texture. A sandy soil might hold 80 mm of available water per metre of root zone. A silty clay loam can hold more than twice that. This is why the same reading produces different decisions on two farms a few kilometres apart, and why a threshold copied from an agronomy blog written for Nebraska will quietly mislead you in Nyandarua.

Those two limits are not fixed forever either. Compaction, organic matter loss and erosion all reduce how much water a soil can hold, which means field capacity itself drifts downward on a degraded field and last season's threshold slowly stops being right. NuaSense covers what is driving that in Kenyan conditions, from land pressure and residue removal through to the climate interaction, in declining soil health in Kenya.

The same sensor reading of 62 marked across three soil types. On sandy loam it sits above field capacity, on loam it sits comfortably in the readily available band, and on clay it sits inside the stress zone.
The dashed line is one reading. Where it lands depends entirely on which soil the probe is buried in. Diagram by Soil Sensors Kenya.

The third number: how far you are willing to let it fall

You do not irrigate at the wilting point. By then you have already lost yield. FAO-56 handles this with a depletion fraction, written as p, which is the share of total available water a crop can give up before it starts to suffer. Multiply p by total available water and you get readily available water, the working range you actually manage.

For many field crops p sits near 0.5 under moderate demand, which means you refill after roughly half the available water is gone. Shallow-rooted vegetables want a smaller number. Crops with deep roots and low sensitivity tolerate a larger one. High evaporative demand pushes p down, because the plant needs water faster than the soil can deliver it. On a hot windy afternoon in the Rift Valley, the same soil at the same moisture content is a harder place to be a plant than it was at dawn.

Once you have those three numbers, the sensor reading finally becomes a decision. Not before.

Knowing when to irrigate is only half of it. How the water is delivered decides whether the refill actually reaches the root zone or runs off the surface. The NuaSense overview of smart irrigation in Kenya works through drip and micro systems on smallholder plots, the solar-powered pumping that gets around unreliable grid access, and why adoption tends to stall on cost and local manufacturing rather than on the technology itself.

Getting field capacity without a laboratory

You do not need a lab to make this practical. You need one wet-up and some patience.

  1. Irrigate a small area past saturation, ideally after a harvest so nothing is drawing water.
  2. Cover it with plastic sheeting so evaporation cannot confuse the picture.
  3. Watch the sensor. It will spike, then fall steeply, then flatten. Sandy soils flatten within a day, clays take two or three.
  4. The value where the curve flattens is field capacity on your scale, at that depth, in that soil.

Wilting point is harder to pin down honestly, and most growers never measure it directly. A workable substitute is to note the reading at which the crop first shows afternoon stress and does not recover overnight, then treat that as the floor you never revisit. It is not the textbook value, but it is yours, and it comes from your field rather than from a table.

One caveat worth stating plainly. Low-cost capacitive sensors drift, and the published work on them is blunt about the size of the problem. A 2025 calibration study in Sensors found that soil-specific calibration was required for reliable results and that recalibration every three to four months was needed to hold accuracy. If your thresholds were set last season and nobody has looked since, they are probably no longer the thresholds you think they are.

Sanity check before you trust any threshold

Run these three tests on your own data. Range: if the maximum is exactly 100, you are looking at percent of scale. Physics: any mineral soil reporting above 55 percent volumetric water content is not reporting volumetric water content. Coverage: count readings per day against the expected interval. Averages built on days with 90 percent packet loss are not averages, they are noise.

2 to 6
permittivity, dry soil
≈ 80
permittivity, water
40 to 55 %
VWC at saturation, mineral soils

One depth will lie to you

A single probe at 15 cm is measuring the layer that responds fastest to everything and holds nothing. It spikes within hours of rain, and it dries out first under sun and wind. Read it alone and you will irrigate on days the crop did not need it, because the root zone below still had reserves.

Two depths turn that noise into information. When the shallow probe rises and the deep one does not, the water did not get past the top layer, which usually means the event was too small or the surface sealed. When both rise together, the wetting front reached the root zone. When the shallow probe falls well below the deep one during a dry stretch, the crop has moved on to drawing from depth, and you have more time than the top sensor suggests.

Line chart of a shallow sensor at 15 cm and a deep sensor at 45 cm through a rain event. The shallow line spikes within hours and later falls furthest, while the deep line rises two days later, peaks lower, and declines slowly.
The gap between the two lines carries more information than either line on its own. Diagram by Soil Sensors Kenya. Schematic of the pattern described above.

Placement matters as much as depth. Put probes where the crop is representative, not next to the tap, not in the wheel track, not in the one patch that always looks worst. Two probes in a uniform block beat six scattered across soils that behave differently, because you can average the two and know what you averaged.

What to do this week

  • Find out whether your sensor reports percent of scale or calibrated volumetric water content. If nobody can tell you, assume it is percent of scale.
  • Establish field capacity on your own soil with one wet-up and a sheet of plastic.
  • Write down the reading at which you will irrigate, and the depth it refers to. A threshold without a depth is not a threshold.
  • Add a second depth in at least one block before you add a second block.
  • Put a recalibration date in the calendar for three months out.

None of this requires better hardware. It requires the sensor to be anchored to two limits from your own field, which is roughly an afternoon of work and lasts several seasons. If you want the mechanics of the measurement itself, we cover them under soil moisture, soil temperature and electrical conductivity. If you would rather have somebody establish the baseline with you on site, that is part of what NuaSense does during installation.

Understanding the depletion fraction is also the point where soil data starts talking to weather data, because p moves with evaporative demand. That is the subject of the next piece.

Sources

  1. Chapter 8: ETc under soil water stress conditions, Irrigation and Drainage Paper 56, Food and Agriculture Organization of the United Nations. Definitions of total available water, readily available water and the depletion fraction p.
  2. Calibration of Low-Cost Capacitive Soil Moisture Sensors for Irrigation Management Applications, Sensors, 2025, via PubMed Central. Soil-specific calibration requirement and recalibration interval.
  3. Understanding Soil Water Content and Thresholds for Irrigation Management, Oklahoma State University Extension. Practical threshold setting and available water by soil texture.

Questions we get asked

Is a reading of 62 good or bad?

On its own it is neither. It becomes good or bad once you know the field capacity and wilting point of that soil at that depth. On a sandy loam, 62 on a relative scale can sit above field capacity, meaning water is draining past the roots. On a clay, the same figure can sit inside the stress range.

Do I have to convert readings to volumetric water content?

No. Working in percent of scale is fine and is what most growers end up doing. What matters is that your irrigation threshold is expressed on the same scale as your readings, and that you never present a scale reading as though it were volumetric water content.

How many sensors does one block need?

Two depths in one representative position tell you more than several single-depth probes scattered around. Add a second position once the first is calibrated and you know what normal looks like there.

How often should sensors be recalibrated?

Published work on low-cost capacitive sensors points to every three to four months to hold accuracy. Recalibration here means re-establishing your field capacity reference, not sending hardware anywhere.

Does soil temperature affect the reading?

Yes. Dielectric response shifts with temperature, which is one reason probes report soil temperature alongside moisture. Compare readings taken at similar times of day when you are setting thresholds.

Read next

Want the baseline set up properly?

We install LoRaWAN soil probes across Kenya and establish field capacity on your soil during commissioning, so the thresholds in your dashboard mean something from day one.

Talk to us