Looking for a proof that soil moisture sensors can be reliable

Using Arduino Project Guidance
1

Hi!

My dissertation is about comparing the accuracy of multiple, relatively cheap soil moisture sensors (RD-SOIL 3, SA-27) and I'm asking for your advice. As I will have to defend my dissertation later, I would need to have a stable answer to certain questions. I tested both of these sensors in chernozem soil and I got a really significant difference between the measurements. The cheaper (SA-27) gave 8% VWC, while the RD-SOIL3 gave 70% VWC and the real VWC (after the gravimetric measuring) was appr. 40%. I know that every soil moisture sensor needs calibration regardless of how much did they cost. So if I would get a question like 'Don't you think after these inaccurate measurements that these sensors are useless?' then how should I prove that they are not? How Im gonna defend this part of the dissertation if they will say that these sensors are just simply not reliable?

2

@chris1996

Needs ACTUAL links to the sensors as people may not know which is which without reading the tech specs.

Most of the cheaper ones (super cheap) are not long term reliable due to the galvanic effect.
Materials used come into play as does voltage swapping to negate galvanic responses.
Capacitive sensors are usually better than restive ones.

Use of the forum search will also give you some great results too. (that magnifying glass thingy upper right of your screen)

Your topic was Moved to it's current location / section as it is more suitable.

Could you also take a few moments to Learn How To Use The Forum.
Other general help and troubleshooting advice can be found here.
It will help you get the best out of the forum in the future.

EDIT....Notice I am not actually writing your dissertation just giving you pointers that you should already be aware of.

3

chris1996:
So if I would get a question like 'Don't you think after these inaccurate measurements that these sensors are useless?' then how should I prove that they are not?

I would expect that if you are expecting any actual indication of "VWC", then all these sensors with PCB prongs stuck into the soil will be, essentially, useless.

It is absurd to suggest that such a "probe" could make an accurate assessment of moisture content. As Bob points out, only the single-prong capacitive version could possibly be durable and thus repeatable so may perhaps be calibrated to within say, 30%.

Essentially these are intended - presumably - to give an indication of "Water me now!"-"I'll survive :grinning: "-"Drowning!"

4

chris1996:
So if I would get a question like 'Don't you think after these inaccurate measurements that these sensors are useless?' then how should I prove that they are not? How Im gonna defend this part of the dissertation if they will say that these sensors are just simply not reliable?

Well, isn't it a fair question? Your measurements show that they aren't accurate. If I were your examiner, I would wonder if you used "due diligence" and measured the error, determined the sources of error, tried calibration and found the effective range of calibration, and so on. It's unscientific to be invested in a positive result, and disingenious to gaslight a serious problem.

If you want to impress these people, be equivocal and thorough in your investigation. Every scientist or engineer will encounter ideas that are not what they are cracked up to be, and problems that can't be solved conveniently. The measure of their skill is not the ability to sell the idea, or solve the problem, but to provide an accurate view of the actual state of affairs so that alternative ideas or designs have a solid foundation upon which to build.

I don't think they would blame you for not knowing the outcome, at the beginning of your project. It happens frequently.

5

If you are serious about this project, consult agricultural experts who use professional methods, including soil moisture sensors, to grow commercial crops.

6

I built an arduino based soil moisture meter a while ago because I was dissatisfied with available units.
Even expensive instruments use a probe that measures the resistance or capacitance more or less at a point.
Unfortunately the soil moisture level can vary greatly even for a plant in a pot.

My device uses two identical copper probes to measure the spreading resistance.
Because the probes are a fixed distance apart, and pressed a fixed distance into the ground, the system can be calibrated.

Here is my page Experiments with an Arduino microcontroller - Soil moisture meter

Its robust and reliable - but not intended for permanent installation as the copper would corrode over time.

7

So if I would get a question like 'Don't you think after these inaccurate measurements that these sensors are useless?' then how should I prove that they are not?

There is a hidden assumption in that question, which is that in order to be useful they have to have some unspecified level of accuracy. So to even begin to provide a meaningful answer you have to know what the proposed application is and what level of accuracy is acceptable for the application. That question specifies neither an application nor an acceptable level of accuracy for the application, so it is not the sensors that are useless, it is the question. Your answer should cover at least these points.

As a general rule I suggest you need to learn to spot hidden assumptions in questions, when you spot the hidden assumptions you can deal more effectively with the question. As an exercise in doing so (which you can do or ignore as you please) what are the hidden assumptions in this question: Who are the best drivers, men or women?

Good luck with your course.

8

chris1996:
So if I would get a question like 'Don't you think after these inaccurate measurements that these sensors are useless?' then how should I prove that they are not?

What would be wrong with the reply - "Yes, that's correct, I was not able to get reliable readings from any of them"

You could then go on to explain the things that you did to get the best possible reliability from the sensors - none of which was able to make them reliable.

...R

9

PerryBebbington:
Who are the best drivers, men or women?

It depends... It is said that men have more accidents. It is also said that women SEE more accidents. :wink:

10

using the current acceptable scientific method, you need to write your conclusion and then support it by selecting all results that are in favor of it.

in the past, science used to be sample and test, gather results then align the conclusion to whatever the facts revealed.

to get a passing grade, you need to figure out what the professor expects and then do that.

In the past, when it was gather data first, then reach a conclusion, one would find the most accurate thing, and as close to a physical world thing. example, we do not use any instruments, we boil water knowing that at certain conditions it will boil. that is exactly100 deg C

in your case, a fixed pot of sand, after a few hours in an oven will be 0% moisture so will weigh X
filled, then drained, it would be fully saturated and weigh X+Y and Y= maximum moisture
repeat with clay, loam, etc

You could weigh a pot and insert a sensor and monitor it as it dries.

I would try to beg-borrow the top of the line. possibly on-loan from a manufacture or go to a lab.
and get multiples of the 'cheap' ones make multiple runs and log the results.

As a note, there is some possibility that large bags of water that are moved close to some capacitance sensors will squeeze the ground, or just the proximity will alter the readings. in other words, when you walk and move around.

My experience has been less that stellar. I get a range from about 460 to 525 on one of the cheap-cheap capacatance probes.

readings each minute
Temp_HR: 556 Soil_HR: 518
temperature RH pres reading tet # last minute average, change from last reading current reading
TA: 66.8 RH: 43.1 PR: 30.06 0 baseSoil: 508 SoilAvg: 0 Soil: 521 // 0 after pressure = first reading
TA: 67.1 RH: 42.9 PR: 30.06 1 baseSoil: 508 SoilAvg: 13 Soil: 513
TA: 67.3 RH: 42.6 PR: 30.06 2 baseSoil: 508 SoilAvg: 18 Soil: 511
TA: 67.3 RH: 42.5 PR: 30.06 3 baseSoil: 508 SoilAvg: 21 Soil: 528
TA: 67.4 RH: 42.5 PR: 30.06 4 baseSoil: 508 SoilAvg: 41 Soil: 529
TA: 67.3 RH: 42.4 PR: 30.06 5 baseSoil: 508 SoilAvg: 62 Soil: 529
TA: 67.4 RH: 42.4 PR: 30.06 6 baseSoil: 508 SoilAvg: 83 Soil: 528
TA: 67.5 RH: 42.4 PR: 30.06 7 baseSoil: 508 SoilAvg: 103 Soil: 523
TA: 67.5 RH: 42.4 PR: 30.06 8 baseSoil: 508 SoilAvg: 118 Soil: 533
TA: 67.4 RH: 42.4 PR: 30.06 9 baseSoil: 508 SoilAvg: 143 Soil: 523 // poured water into pot
TA: 67.5 RH: 42.4 PR: 30.06 10 baseSoil: 508 SoilAvg: 158 Soil: 466
TA: 67.3 RH: 42.4 PR: 30.06 11 baseSoil: 508 SoilAvg: 116 Soil: 474
TA: 67.4 RH: 42.4 PR: 30.06 12 baseSoil: 508 SoilAvg: 82 Soil: 474
TA: 67.6 RH: 42.2 PR: 30.06 13 baseSoil: 508 SoilAvg: 48 Soil: 475
TA: 67.8 RH: 41.9 PR: 30.06 14 baseSoil: 508 SoilAvg: 15 Soil: 477
TA: 68.0 RH: 41.6 PR: 30.06 15 baseSoil: 508 SoilAvg: -16 Soil: 481
TA: 68.1 RH: 41.3 PR: 30.06 16 baseSoil: 508 SoilAvg: -43 Soil: 499
TA: 68.2 RH: 41.2 PR: 30.06 17 baseSoil: 508 SoilAvg: -52 Soil: 476
TA: 68.2 RH: 41.5 PR: 30.06 18 baseSoil: 508 SoilAvg: -84 Soil: 482
TA: 67.9 RH: 42.1 PR: 30.06 19 baseSoil: 508 SoilAvg: -110 Soil: 466
TA: 68.0 RH: 41.7 PR: 30.06 20 baseSoil: 508 SoilAvg: -152 Soil: 493

take 60 readings, take the average
60 more, new average, repeat.

IMHO, the window is too small and shifts around in my (non-professional) testing.
the numbers are not repeatable between waterings.

however, you can get a too dry and too wet and have some idea of surface level water.

as another note, there are people who want to measure water at depth in the soil, the surface, at the start of the root zone and the bottom, etc.

I will wish you the best and request that you share your final result. many of us would love to know what we are doing wrong.

11

dave-in-nj:
using the current acceptable scientific method, you need to write your conclusion and then support it by selecting all results that are in favor of it.

in the past, science used to be sample and test, gather results then align the conclusion to whatever the facts revealed.

I must be old school because if I started with a conclusion, for example "Looking for a proof that soil moisture sensors can be reliable", then search for evidence to "prove" the hypothesis, I would get a big FAIL.

That is simply not how science works.