ET vs Moisture sensor

[Bizarroterl]

Hi,
It seems to me that the use of a moisture sensor would be the best way to go. I don't see how ET could be near as accurate as it doesn't take into account the type of soil, plants, shading, etc.

Is moisture sensing the "preferred" method? If not, why?

[hult]

Hi,
It seems to me that the use of a moisture sensor would be the best way to go. I don't see how ET could be near as accurate as it doesn't take into account the type of soil, plants, shading, etc.

Is moisture sensing the "preferred" method? If not, why?

What do you mean by "way to go".

What are you trying to do?

Where do you live?

I see endless needless confusion of folks trying to automate watering their lawn.

Is that your objective?

When actually measured, not calculated or otherwise estimated, measuring Evapotranspiration (ET) has historically been obtained by monitoring the change in weight of large flat contaiiner with plants growing in it ( a lysimeter). And this_does_ "take into account the type of soil, plants, shading, etc." You may be referring to Potential Evapotranspiration (PET) which is a 'nuther kettle of fish.


... Marc

[Michael McSharry]

While measured is one approach is has been problematic in the past for both the big players in the industry and mcsSprinklers specifically. Soil will typically channel water so if you sensor is in a channel it will react quickly while if not then it takes much longer for it to register. The moisure channels also change over time as well as the head distribution also fluctuates so using a point (or a few points) estimate for a large area will typically give poor results.

The ET calculation, or more correctly as pointed out PET has yielded good results in letting your irrigation schedule follow the general needs of the foilage. It is true that each local site needs to assess what their nominal watering needs are and the rate at which soil is moistened. There is some trial and error associated with it, but with a few measurements you can come pretty close and then not worry about it again.

[hult]

At the risk of making the answer more complicated than what the OP asked, let me expand on what Michael has written in part because there are some solutions to the complicating/confounding issues raised.

(More detailed information on where the OP lives and his objectives are needed before a meaningful, site-specific answer can be offered to him , IME.)

With respect to general "home automation" garden sprinkler system, there are two basic "events" that need to be characterized, namely when to start ("ON") and when to stop ("OFF") irrigating/sprinkling.

As fundamental this two-event model might seem, a common source of confusion is to try to measure/determine both events from a single soil moisture sensor at the same single point even though the ideal depth for a sensor depends on which event you are trying to characterize.

(Depth of sensor placement is the 1-D, vertical problem. 2-D, areal variability is another, different matter. "Channeling" of water through macropores is a yet another real, common phenomenon but depending on what you are trying to measure can be mostly unimportant , and depending on the soil type and the level of effort expended in sensor installation, avoided. )

To generalize, the "ON" decision/event/condition depends on the amount of soil moisture available in the root zone. The general objective is to apply water (irrigate) substantially before the plants' wilting point is reached. So the sensor should be within, or just above, the root zone. (Easier wriiten than done with physically large sensors and shallow-rooted lawn grasses ;-)

The "OFF" decision is more loosey-goosey, and is more dependent on climate and cost/availability of water. In general, both in order to reduce liklihood of encouraging development of a shallow root system (which in turn increases the need for frequent irrigation), and preventing the build-up of dissolved solids (particularly in arid climates), the root zone should be flushed. So turning off the water when the soil moisture becomes adequate in the root zone may not be the right "event". Significant flushing of water _past_ the root zone may be the better/needed/critical "event" to trigger"Off".

In arid climates and in the absence of significant antecedent irrigation, a calculated ET/PET rather than measurements of soil moisture may be the most dependable and pragmatic solution as Michael suggests.

In more humid areas and(or) with significant antecedent irrigation, the calculated PET method may much less useful. For example, this summer in Kentucky we are in a drought. Clayey soils, drought-intolerant plants, and much irrigation during the last 60 days make PET calculations mostly unhelpful on our property.

Somewhere I have a writeup I did years ago that should be helpful. I've been using the commonly used (now Irrometer) Watermark soil moisure sensors since before/just as they became commercially available. There are interesting developments in improved sensors with other technologies that may reach affordability soon, especially for the DIYer. (My research areas in the 1980's included work on development of sensors to measure water/air/oil saturation and I'm tempted to indulge again ;-)

... Marc
hult at hydrologist dot com
www.ECOntrol.org

[Bizarroterl]

What do you mean by "way to go".

What are you trying to do?

There are 2 purposes to irrigation control systems - reduce water usage and the labor involved in irrigation. Almost all irrigation control systems do the latter. Therefore the focus of my interest is to reduce water usage. If the use of a moisture sensor results in less water usage (while still maintaining plant health) then it would be the "way to go".

Where do you live?

San Jose, CA. Soil type run the full gamut here and although the climate is mediterranean, the area is actually a large mixture of micro climates. Temps can easily vary 15degrees in 10 miles.

I see endless needless confusion of folks trying to automate watering their lawn.

Is that your objective?

When actually measured, not calculated or otherwise estimated, measuring Evapotranspiration (ET) has historically been obtained by monitoring the change in weight of large flat contaiiner with plants growing in it ( a lysimeter). And this_does_ "take into account the type of soil, plants, shading, etc." You may be referring to Potential Evapotranspiration (PET) which is a 'nuther kettle of fish.


... Marc

Yes, I was referring to PET. Although measuring the weight may work for a turf grass, I forsee some difficulties in using the same procedure for a boxwood hedge or a 50' Magnolia.

[Bizarroterl]

While measured is one approach is has been problematic in the past for both the big players in the industry and mcsSprinklers specifically. Soil will typically channel water so if you sensor is in a channel it will react quickly while if not then it takes much longer for it to register. The moisure channels also change over time as well as the head distribution also fluctuates so using a point (or a few points) estimate for a large area will typically give poor results.

The ET calculation, or more correctly as pointed out PET has yielded good results in letting your irrigation schedule follow the general needs of the foilage. It is true that each local site needs to assess what their nominal watering needs are and the rate at which soil is moistened. There is some trial and error associated with it, but with a few measurements you can come pretty close and then not worry about it again.

That was one of the problems I anticipated - not getting a true representative reading with only a few points of measurement. I can see that this could be overcome by adding more sensors but that runs up costs and at some point becomes senseless.

I dd come across a moisture sensor that appears to help overcome the typical limitations of point measurement. It is a volumetric sensor that is 3m long. http://www.mea.com.au/products/aquaflex-sensor/1/

Wouldn't this help overcome the channeling issue and give a more representative reading for a larger area (IE turf grass)?

I don't have pricing and I havent researched their controls enough to know if it can be interfaced to but the technology (if it works as advertised) sounds promising.

I did look at the Irrometer Watermark sensors, but they are point type sensors and it would get expensive quick trying to accurately monitor anything other than a small area.

[hult]

There are 2 purposes to irrigation control systems - reduce water usage and the labor involved in irrigation. Almost all irrigation control systems do the latter. Therefore the focus of my interest is to reduce water usage. If the use of a moisture sensor results in less water usage (while still maintaining plant health) then it would be the "way to go".

Thanks for the additional info. So as I now understand it, and with respect to the two-event model I suggested in my previous post, your efforts/challenge focus on the second event, namely when to stop watering. (This steps lightly around the relationship of Total = Frequency x Intensity at least for now.)

San Jose, CA. Soil type run the full gamut here and although the climate is mediterranean, the area is actually a large mixture of micro climates. Temps can easily vary 15degrees in 10 miles.

Yes. Quite different from using HA to water a lawn used for Recreational Mowing in western Nebraska ... ;-)

I grew up partly in a village in the mountains on an island in the Mediterranean and have visited/worked in the Bay Area and understand some of the differences. Historically, a very significant one in this context has been that in the Mediterranean, folks largely limited plantings to acclimated plants whereas all manner of exotics are prevalent in the BA. Depending one what you have, therein may be the root of the principal challenge (s). Surrounding our village are olives trees planted hundreds of years ago that have never seen a drop of irrigation water. It is also true that the tomato plants for which the area was also famous wouldn't make it through a fortnight in mid-summer without irrigation. There, one solution to not wasting water through over-irrigation is/was that applied water that exceeds field capacity in one terrace percolates to the terrace below, so it isn't lost or 'wasted'.

In the late 80's I visited research plots at a nature preserve in your county/area that a colleague had instrumented with soil-moisture sensors based on Time-Domain Reflectometry (TDR). He was a co-author of the original seminal paper in _Science_ on the use of TDR in soil science and hydrology. The plots were purposely chosen to reflect the local microclimates. Much of their work revolved around reconciling measured and computed values from different methods. They had the advantage of comprehensive local weather/energy data and the ability to use neutron probes and other methods that more directly measure water content. Getting calculated "ET" methods to agree with reality was still a daunting task. That hasn't changed fundamentally, but better moisture sensors may soon be a reality for DIY HA.

Yes, I was referring to PET. Although measuring the weight may work for a turf grass, I forsee some difficulties in using the same procedure for a boxwood hedge or a 50' Magnolia.

By coincidence, a sprinkler is operating even as I write within the drip zone of a 45-ft Southern Magnolia in our back yard. Maggy and I communicate regularly about her needs with nary a word about PET or HA ;-)

Are all your plants 'in the ground' and not in planters ? ( I tell my wife that she's "on her own" with respect to most of the planters she uses . )

... Marc
hult at hydrologist dot com
www.ECOntrol.org

[hult]

That was one of the problems I anticipated - not getting a true representative reading with only a few points of measurement. I can see that this could be overcome by adding more sensors but that runs up costs and at some point becomes senseless.

I dd come across a moisture sensor that appears to help overcome the typical limitations of point measurement. It is a volumetric sensor that is 3m long. http://www.mea.com.au/products/aquaflex-sensor/1/

Wouldn't this help overcome the channeling issue and give a more representative reading for a larger area (IE turf grass)?

I don't have pricing and I havent researched their controls enough to know if it can be interfaced to but the technology (if it works as advertised) sounds promising.

I did look at the Irrometer Watermark sensors, but they are point type sensors and it would get expensive quick trying to accurately monitor anything other than a small area.

This is Michael's forum, so it wouldn't behoove me to butt in (yet again ;-) on a question that you ask him ... (if I'm outa line, I ask Michael to let me know...)

But you ( Bizarroterl ) have zoomed in on a fundamental point that can be understood by groking the concept of "Representative Elemental Volume". All measurements average over some volume and 'none' provide truly point values. The question is: Over _what_ volume do you want to average? -- the answer to which depends on which question you are asking. I would instrument and calculate quite differently for our 45-foot magnolia tree than for our bluegrass lawn in full sun because the volumes and geometry that matter are different.

The description of the Aquaflex sensor makes a virtue out of the physics that necessarily pertain. By their nature, soil moisture TDR probes are "volumetric probes" and "average over their entire [effective] length". And the details of physical installation have always been important and often overlooked. One of the virtues of TDR has been that one could install the probes into essentially _undisturbed_ soil by pushing them into place horizontally from inside a trench. The Aquaflex 'method' negates that virtue by requiring installation _inside_ the trench and so being necessarily placed in recently disturbed porous media/soil. In a recently plowed field that may be inconsequential. Where No-Till is practiced, or in natural environments, it might not be/isn't.

The volume over which a sensor averages can be manipulated by how it it is installed. The Watermark effectively creates an artificial 'porous medium' with a volume of a few cc's (the sensor body itself) and measures the properties of that. One can increase the volume (somewhat, within limits) by creating another, larger artificial porous medium within the soil and instrumenting it. In particular, the issue of 'channeling' at the level of lawn grass roots can be effectively eliminated (at least temporarily) in this way. This merges conceptually with 'installation methods' depending on how and what one uses to 'backfill'.

The Aquaflex sensor helps with areal averaging (what I characterized as the "2-D" problem). It does not intrinsically help with the "1-D" issue of characterizing moisture content with depth. -- on the contrary, it negates one of the virtues of TDR for this use because of how it is installed as I previously asserted.

With respect to the 'channeling' issue, whether the use of a purposely anisotropic sensor (i.e., measures differently in one direction than in another ) installed horizontally (such as the Aquaflex) helps or hurts depends in part on where the root zone of interest is with respect to the sensor. It will help a bit if the channeling via macropores (aka secondary porosity or permeability) is due to (eg) shallow earth worms and old grass roots. It won't do nuthin for macropores from (eg) moles or decayed tree roots that extend below the volume measured. But _shallow_ 'channeling' is a relatively minor issue.

The hang-up with TDR for the two decades since it was first applied to soil moisture measurementhas been cost of the electronics -- not the sensors themselves. Note that both TDRs and Watermark sensor can be multiplexed to a single set of measurement electronics. TDR probes have the virtue of not being patented (Last I Knew! ) and are readily fabricated by the DIYer out of stainless steel rods and (eg) Lexan, and connected to multiplexor and subsequent electronics by inexpensive RF cable.

All-in-all, horizontally-averaged measurements such as those provided by the Aquaflex are helpful for many common applications. The fundamental method used (TDR) can be/is more repeatable and accurate than Watermark sensors.
I suspect that the Aquaflex is 'expensive' but don't know.

There are may also be capacitative soil-moisure methods affordable/in the offing for DIY (Methinks ;-)

HTH ... Marc
hult at hydrologist dot com
www.ECOntrol.org

[hult]

The ET calculation, or more correctly as pointed out PET has yielded good results in letting your irrigation schedule follow the general needs of the foilage. It is true that each local site needs to assess what their nominal watering needs are and the rate at which soil is moistened. There is some trial and error associated with it, but with a few measurements you can come pretty close and then not worry about it again.

Michael has created a remarkable product with heuristics that are practical, physically sound, and readily understood and implemented. My intent here is jist to be useful and to see if I can in helping to expand the range of application a bit.

Sometimes simple volumetric calculations are all that are required -- and are generally a good " reality check" in any case.

The "50-foot magnolia tree" that the OP introduced and that I coincidentally am actually husnanding through a drought by watering is a good example. One can readily calculate that the volumetric soil moisture to a depth of three feet inside the tree drip line (radius 25 feet) will increment by 0.10 if one applies water at 3 gal/minute for 24 hours. To move from a risky to nicely saturated condition might take a couple of days.

That's a purty long time, but if there is a significant soil-moisture deficit as there is/was this drought year, watering for only a couple of hours will benefit the grass, and -- very destructively! -- the moles more than the tree ... The fact that the moles tore up a large part of my yard in another area indicated to me that I wasn't watering 'deeply' enough there.

If there isn't a major, long-term moisture deficit, I don't need to water the tree regardless of how well the _lawn_ is or isn't faring. Other plants are typically intermediate in value. Frankly, there's much more at risk with the tree. If the lawn goes dormant because it is dry, it's no big deal. If we were to lose the signature tree of the yard, I'd be crushed. Young and fruit-bearing trees are particularly vulnerable and susceptible to growing of too-shallow roots because of too-shallow watering.

A better weather-based approach than short-term PET for a tree-watering decision/event that also avoids the pitfalls of DIY moisture sensors might be to calculate the cumulative departure of precipitation from Normal. The cumulative departure tells one how far one is 'behind' in precipitation in a way that is quantitatively simpler to deal with than a drought index. It is easier numerically in part because it has the same dimension, L, and units as the amount of water applied (inches of water).

The cumulative departure approach would work with no other input than precipitation history up until the first time one irrigates. After that, one would have to account for the volume of water added. The "ON" trigger value might be reset once or more per year, and perhaps modified by the local/regional value of the downloaded (eg) Palmer Drought Severity Index or Crop Moisture Index (CMI). www.drought.noaa.gov/palmer.html. The "OFF" value might be calculated volumetrically as I showed and the volume added used to adjut the cummulative departure.

IOW, for a relatively deep plant/soil system such as an young and(or) large-but-vulnerable trees, one could control that/those particular irrigation zone(s)entirely with data on the amount of applied water and local rainfall data. For this purpose, it would/could be better than either the typical DIY soil-moisture measurement or short-term PET calcs, IMJ.

Michael has created a remarkable product with heuristics that are practical, physically sound, and readily understood and implemented. My intent here is jist to be useful and to see if I can in helping to expand the range of application a bit.

HTH ... Marc

[Bizarroterl]

Hult,
You've brought up some good points, especially in drought situations. I can see where the use of PET and even moisture sensors could result in the loss of a tree since (for many) they tend to have deeper roots than turf grasses. As you pointed out, the health of the turf grasses doesn't necessarily indicate that the tree is getting sufficient irrigation.

Extending on that, I can see where a tree that has been getting its water from surface irrigation since it was planted would be less prone to drought problems as its roots would be closer to the surface.

If the above supposition is valid, that may explain why my magnolia is now wrecking the grass above by gradually filling the area with roots. As the years have rolled by it has become harder and harder to maintain good turf under the tree so that now I'm to the point where it is a struggle to maintain any turf under it. Alas, the tree's leaf drop is so messy I'm seriously thinking of removing it.

From Michael's and your posts it's clear to me that there is no "magic bullet" that eliminates the requirement of monitoring the irrigation area and making any needed tweaks as the local conditions change. PET would then be the preferred solution as it doesn't require the extra costs and complexity of moisture sensors yet does adapt to changes in environmental conditions. Methods other than PET may be preferred under special circumstances. IE extra irrigation to carry deeper rooted plants through a drought.

[Michael McSharry]

Marc has much more applicable experience and background on the subject than me. My experience base is initially what I could learn from various internet sources and talking with those in the industry followed by my real-world experiences and from working with others via mcsSprinklers evolution.

mcsSprinklers does not provide an answer to irrigation needs, but it is a tool that tries to make it easier for a user to achieve their irrigation goals.

I have worked with two users that have used the Davis Vantage Pro with the quad moisture probes and the results were similiar with both. The effectiveness was very dependent on the probe position and some areas just could not be measured well. There was also the failure modes of low battery where the expected moisture change was not being delivered so irrigation continued to the timeout.

Another problem was the slowness of response. The irrigation cycle would timeout and then sometime later the probe would report the moisture change. This scenario can probably be improved with the delays and round robin scheduling available that are now available with mcsSprinklers, but where not a few years ago.

Working with these two I came away with the impression that this approach required too much maintenance of the probes with results that were no better than the more general ET approach. I also attempted to use low cost gypsum sensor at my location as an engineering experiment. When the sensor was burried a couple inches below the surface it was affected more by sunlight heating the surface than actual moisture in the soil. When I bury it lower in my very rocky soil then it changed very little even after drowning the area where it was burried. I suspect there was just not enough soil to hold the moisture around it and the water just went down trough the rocks.

The aquaflex, as well as devices made by other companies, may be more effective than the Davis probes and may work very well. What does not exist, however is the field experience with mcsSprinklers to properly characterize their behavior for a variety of user's condidtions and just as importantly to recognize the pattern of faulty behavior so mcsSprinklers is able to take protective action. I'm willing to work with anyone who want to go down that road, but my experience shows that the effort one is trying to save with direct measurement will be consumed by the engineering and maintenance to get the system to function as desired.

When I started with mcsSprinklers I was trying to come up with an algorithm that would allow the OFF event, per Marc's description, to be automated. This did not seem to be in the cards so I left the OFF control to be deterministic with the responsibility left with the user to determine how much water needs to be added at any given cycle.

The ON event is the one that is modulated based upon the environmental conditions, measurements, or time. The user is still responsible for the definition of both the ON and the OFF events where the OFF event is fixed at a time after the ON event and the ON event is characterized based upon soil conditions and weather.

I suspect that most irrigate their lawns in a manner that encourages shallow roots which then makes the lawn more sensitive to short term weather conditions. This is something that mcsSprinklers cannot prevent. It is still the user's decision.

I had not considered the longer cycle of control that Marc describes for drought. Draught is not in the Seattle area vocabulary. While an automated control based upon this may not be appropriate, it may be useful for this status information to be made available to the user so they are aware of the longer term patterns.

For fruit trees a technique that I have employed is to run a pipe into the ground next to the trunk of the tree and apply water through this pipe. This introduces the water at a lower level and encourages the roots to seek it there. At my current location I could not do this because the rock content is so high that either it acks like a swimming pool where there is no drainage or it acts likes a hole to the center of the earth as the water goes through the crevices in the rock structure.