Every so often (well, quite often actually) someone will talk about the hard “red clay” soil they have in their yard. Recently a poster described the soil as “Georgia Red Clay”. Now, usually we say: “it’s probably just hard, not actually clay”. But this time I decided to go looking for the clay – at least a little.

The USDA National Resources Conservation Service has a neat online tool for exploring a lot of soil survey data. It’s called the Web Soil Survey. There are lots of things to look at, but I specifically looked at the soil survey for Houston and Peaches Counties, which are near the center of the state. I looked at all the texture analysis for all the distinct mapped soils to a depth of 6 inches and plotted them on a USDA soil texture diagram. The area of the bubble is relative to the percent area of the survey district covered by that soil family. In several cases you can’t really see the bubble because that soil doesn’t occur very frequently.

There really isn’t much clay in Houston and Peaches Counties, Georgia. If we group the soil classes together and plot the percentage area of the survey area with those soils (to a depth of 6 inches), you can easily see that most of the soils in these two counties are a mix of sand and loam.

Now, a soil survey can’t tell us what kind of soil is in a particular yard – you need to test that for yourself. And it’s probable that in many urban and suburban areas, the surface layers have been disrupted by construction. A more detailed analysis of this soil survey shows that the clay-fraction of the sub soil is generally higher than the surface soil, so as top-soil is removed, what is left will be somewhat more “clay-like”, though usually still towards a loam.

Comments, suggestions and critiques always welcomed.

I often get a notion that I can’t get rid of – like cleaning the garage. But it’s usually not nearly so constructive.

I’ve been wondering how, exactly, different organic applications look on the ground at the specified rate. It really doesn’t matter – the application rate isn’t critical, and after a couple of pseudo-calibrated applications, you can get close enough by inspection and feel. But I wanted to take some photos of measured amounts.

The three grains I usually are corn meal, cracked corn and alfalfa pellets. I applied each to a 1-foot square (1 sq-ft) area, each at a rate of 20 lbs/1000 sq-ft, though much higher rates might be used in practice.

So here’s what they look like at 20 lbs/1000 sq-ft:

.
Corn Meal:

.
Cracked Corn:

.
Alfalfa Pellets:

.

And now you know.

Here on the BestLawn forums, we have a special tool we use to determine if your soil is compacted or how well it is being watered.  It’s called a screwdriver:

If you can push the screwdriver into the lawn easily (especially when the soil is properly watered), then your soil tilth is good and you don’t suffer from compaction. Obviously it’s a subjective test since everyone presses down on the screwdrivers with a different force, screwdrivers are different lengths, etc.

So what’s a guy to do who likes objective tests where results can be measured, graphed and otherwise over-analyzed?

You make a dynamic cone penetrometer, that’s what you do. Or at least, that’s what I did with some found objects and help from a bored machinist (total cost ~ $6 for the shaft collars).

(Click for a larger image)

The idea is that the drop weight (the cylinder in the middle) falls a consistent distance, specified by the location of the top shaft collar, which acts as a stop when the weight is raised. The lower shaft collars are fixed in place and acts as the anvil for the drop weight. The tip of the shaft is machined to a 30-deg angle. Using the drop distance, the mass of the drop weight assembly and the diameter of the shaft, the physical forces applied to the soil can be calculated. The soil resistance is essentially measured by how far the shaft penetrates the soil over a number of strikes.

While the physics is fun and may come into play at a later time, I wanted to do a quick ‘proof of concept’ test. I tested a patch of ‘dry’ soil in my backyard with an 8-inch screwdriver. I could only push it in about 4″ (and I’m a big guy). With my homemade penetrometer I dropped the dropweight 50 times, measuring the total penetration distance every 5 strikes – I did this twice to a distace of about 4″. Using an established measurement method, the moisture content of the top 4 inches of soil was 12%.
I then well-watered the test area (twice) and was easily able to drive the 8-inch screwdriver in to the hilt. I repeated the penetrometer test once and measured the soil moisture in the top 4 inches to be 23% (though the penetrometer shaft went in more than 10″).

[For the record, my soil is on the edge of sandy-clay-loam and clay-loam, with ~5% organic matter.]

.
Well what do you know – it’s easier to push a rod into wet soil than into dry soil.

Did I need the penetrometer to tell me that the soil was moist and not compacted? No – the screwdriver could tell me that. But maybe with some more testing and characterization I can use the penetrometer to give good estimates of soil moisture. If I can, I’ll have a consistent and objective tool for testing various ‘soil amendment’ products and concoctions next season. Or maybe not. Either way I’ve finally made the device and published a result. Maybe I can stop thinking about it for the winter.

Soil solarization is a simple, two-step process: (1) Take energy from the sun ( sunlight) to warm the soil, and (2), keep it there.  Step (1) is generally the easier, though there are many variations.  Step (2) requires more ingenuity and sensitivity to local conditions.  This usually involves covering the soil with thin sheets of plastic.  When done properly, solarization can raise soil temperatures near the surface by 10ºC (18ºF) over bare soil.

Soil temperature results at 10, 20 and 30 cm for PE covered soil compared to bare soil. [Katan, p 29]

There are many reasons to use soil solarization – kill pathogens (bacterial, fungal, nematodes, etc), sterilize weed seeds, possibly change the soil structure and chemistry, and other approaches which are not always well understood or documented.  For the purposes of this discussion, we will treat solarization in a general sense.

Without any help at all, soil temperatures go through diurnal and seasonal fluctuations with the rising/setting of the sun and the seasonal altitude of the sun above the horizon.  Other than site selection and shade removal, most soil solarization methods do nothing to intensify that solar flux.  Instead methods are devised to keep the thermal energy collected during the day in the soil, with cumulative effects increasing the soil temperature over several days or weeks.

Consider a fairly simple thermal model of solar heating of the soil (with no plastic covering in place).

There is only one source of energy into the system: q_solar, which heats the top layer of soil.  From there, the heat is transferred through conduction vertically down into the soil (q_cond,vert) and laterally across the soil (q_cond,lat – though in a one-dimensional model, lateral conduction is zero).  In this example, q_conv and q_cond represent thermal losses into the atmosphere.  Typically in an open field, q_cond will be negligible and q_conv will be the significant term.  If there is moisture in the soil, some of it will evaporate.  The latent energy required for evaporation is represented by q_latent,evap.  Finally, any object which has thermal energy (the earth) emits thermal radiation.  The thermal radiation (energy) emitted is q_thermal.

To increase the degree and depth of soil heating through solarization, we want to maximize q_cond,vert, while minimizing all of the thermal losses from the soil.

The solar radiation which reaches the earth’s surface is a function of the sun’s temperature (and thus spectral irradiance) and the gases in the earth’s atmosphere which absorb or scatter that radiation.  The wavelength of solar radiation is between 300 and 2000 nm, and the wavelength of thermal radiation from the heated soil starts around 3000 nm (or 3 μm) – this is in the infrared, or IR, portion of the spectrum.

Comparison of the solar spectrum and thermal spectrum.  Variations in the terrestrial solar spectrum from the theoretical curve are largely due to absorption and scattering by atmospheric water vapor and gases.

When designing a solarization or greenhouse system, the goal is to let the energy/heat in (the solar flux), but prevent the thermal radiation from escaping.  Glass does this very well – it is transparent in the solar spectrum (especially the visible range), but it is nearly opaque to the thermal IR spectrum.  A hot car on a summer day is an example of this –solar energy is transmitted through the glass windows and heats the car interior (seats, dashboards, etc), but the glass is not transparent to the wavelengths at which those items re-radiate in the thermal spectrum and the energy is trapped.  The result is a very hot car.  There are special thin-film applications and additives for glass sheets which are used to address thermal issues in automobiles and architecture.

For obvious reasons, glass is not suited for large-scale agricultural/horticultural applications.  Thin, pliable polyethylene (PE) and polyvinyl chloride (PVC) films are used instead.  They are reasonably transparent through the solar spectrum and provide some barrier to thermal radiation.  Certain additives are used by manufacturers specifically to reduce the IR transmission of thin PE films for solarization in agricultural use.  Mulching with thin PE sheets can reduce the thermal losses to 30-80% of bare soil, depending on the film.

Comparison of PE and PE IR-treated films for transmission properties in the thermal spectrum.  [Adapted from Katan, p 145]

Adding the thin film over the soil also helps reduce the conductive and convective losses.  If the plastic is tightly sealed around the edges and there are no gaps or tears in the film, the airspace is still and convection is minimal.  Still air also provides some thermal insulation, but where the plastic touches the soil, heat conducts into the plastic and is lost through convection on the atmospheric side of the cover.  In some applications, a second sheet of PE is suspended a short distance above the ground layer, providing insulation from these conductive losses.

Conductive losses also occur laterally through the soil at the edges of the plastic, since the soil is cooler outside the covering.  Thus soil temperature under the mulch is at a maximum in the middle and a minimum at the edge.  There are few practical, large-scale methods to reduce these losses.  Use a plastic covering that extends beyond the solarization zone, and keep the soil around the perimeter of the plastic dry.  Dry soil is less conductive and will help reduce lateral losses.

One step in soil solarization which is often promoted is deep watering before applying the plastic.  There are several reasons for this:

1)      Pathogens (viral, bacterial, fungal, nematodes, etc.) are often more susceptible to thermal damage when moist and active.  For weed control, seed coverings are softened and they may even germinate if damp.  They too will be more susceptible to damage, leading to a higher proportion of control.

2)      Wet soil has a higher thermal mass.  Overnight, when the solar energy is absent, the heated water in the soil retains thermal energy.  Since pathogen and weed control is often a function of temperature and time, longer heat equals better control.  It can also leave the soil at a higher temperature in the morning – thus an increased starting point for a new thermal cycle.

3)      Wet soil has increased thermal conductivity.  Solar energy heating the surface can penetrate deeper into the soil, increasing the temperature at greater depths.  Wet soil has a thermal conductivity about three times greater than dry soil.

4)      Moisture in the soil allows for evaporation.  Evaporation requires energy which would otherwise heat the soil, but the water vapor is trapped in the airspace and serves as a greenhouse gas.  Water vapor is largely transparent to solar radiation but opaque to thermal radiation, so it helps warm the air space immediately above the soil.  When the water vapor condenses on the plastic, it releases energy, and the latent heat is returned to the system (albeit on the plastic instead of the soil).  Too much condensation, however, can reduce the transmission of solar energy.  Some films have anti-condensations properties to reduce this loss.

One oft-debated decision in soil solarization is the selection of plastic: clear or black.  Either way, solar energy has to get into the soil.  With clear plastic, the energy gets there directly through the covering.  With black plastic, the covering heats up significantly then re-radiates into the soil or conducts heat into the soil wherever contact is made.  The black plastic is generally no better at retaining thermal radiation (without special additives) than the clear PE, but both reduce convective losses and trap moisture in the air.  The literature is not definitive on which is the obvious choice – probably due to the large number of variables in field trials.  Most studies show that the difference between clear and black films is in the speed and duration of heating, not the actual increase in soil temperature.

Mean values for soil temperatures comparing bare soil to clear and black PE films.  These means represent data between 800 and 1200 hours of measurements.  In terms of average soil temperature, there appears to be little difference.  Hasing did show that the minimum continuous soil temperature was higher for clear plastic than black.  [Hasing, p 45]

There are a number of books, published articles and an untold number of thesis papers written on the subject of soil solarization – for weed suppression, pathogen suppression, increased growth response, microbial population management, etc.  Plus, many state extension agencies  have pamphlets describing details of how to solarize soil.

Rather than include endnotes in this document, here a few research sources:

* Katan, Jaacov and James DeVay, Soil Solarization, CRC Press, 1991.

* Solarization Informational Website, Kearney Agricultural Center, University of California

Hendrickx, Jan, et al, “Worldwide Distribution of Soil Dielectric and Thermal Properties”

* Hale, G.M. and M.R. Querry, “Optical Constants of water in the 200 nm to 200 μm wavelength region”, Appl. Opt., 12, pp 555-563 (1973).  Raw data available at: http://omlc.ogi.edu/spectra/water/data/hale73.dat.

* Zhu, S., et al, “Modeling the thermal characteristics of greenhouse pond systems”, Aquacultural Engineering, 18, pp 201-217 (1998).

* Jaffer, Aubrey, “FreeSnell: Polyethylene”.

* Giacomelli, Gene and William Roberts, “Greenhouse Covering Systems”, Department of Bioresource Engineering, Rutgers University.

* Okur, N. et al. “Effects of soil solarization on the microbial population and activity in the greenhouse”, Cahiers Options Mediterrannees, 31, pp 407-411.

* Hasing, J.E., “Agroeconomic effect of soil solarization on fall-planted lettuce”, Masters Thesis, Louisiana State University, 2002.

* Rubin, B. et al, “Soil Solarization: an environmentally-friendly alternative”, Technical Meeting on Non-Chemical Alternatives for Soil-Borne Pest Control, Hungary, 2007.

* American Society for Testing and Materials, “Terrestrial Reference Spectra for Photovoltaic Performance Evaluation”, ASTM G-173-03.

Now’s the time that no one’s been waiting for – the data.  I’m only presenting the analysis for a few standard soil analysis methods.  After looking at the data, and understanding the estimates and assumptions involved, I’m not sure it makes too much sense to try and characterize every test this way.

To recap:  The following standard deviations (σ, usually expressed as a percentage of the mean) describe the distribution of test results of a single sample tested many times by different soil laboratories.  The ‘minimum significant difference’ represents the minimum difference between two results which is statistically meaningful at the 90% confidence level.  This number may be useful for seasonal soil testing to determine the effectiveness of certain soil amendments.  More details about the method and rationale are posted in earlier blog entries.

The standard deviation (or repeatability) from a single soil lab from year to year is likely smaller than the results reported here, but I did not have access to that data – it is, understandably, private. I did contact Susan Shaner from Logan Labs, and she recommended that to be significant, measurement differences should exceed 15-20%. This is comparable to the standard deviations I have calculated, but less than my more stringent definition of ‘minimum significant difference’. Ms. Shaner also ascribed most of the difference in soil results to sampling techniques. Using the NAPT data in this analysis removes that variable, and good sampling practices by the homeowner should reduce that error in practice.

Has this effort proven useful? I don’t know. If you have yearly results from a lab using these methods, I’d be interested in your feedback. The methods I’ve used were described earlier, and there are several assumptions and estimated used in these calculations which can justifiably be questioned. As always, questions, critiques, comments or concerns are always welcome.

I know that I’m doing no favors to the ‘bestlawn.info’ brand by adding overly-complicated blog entries, but I wanted an addendum to the last posting  – this time with a few graphics.  Please remember that I’m not a statistician in real-life, I only play one on the internet.

In order to accurately determine a soil property though analysis, we need many measurements from which to calculate the average (mean) value.  But we don’t – we take a single sample once or twice a year, it gets tested and we then claim to know that particular soil property.  But we really don’t.

However, from the methods described in earlier posts, we can estimate what a normal distribution of multiple samples would have looked like if we did test 10, 20 or 100 times.  This measure of the distribution is the standard deviation of the method.  It tells us nothing about the mean value, but we can put limits and ranges on statistically significant values.

For a given standard deviation (σ), 90% of all measurements would lie within 1.645 x σ of the mean value.  (The multiplier goes up for increased confidence intervals, but 90% seems reasonable given all the other estimates and assumptions which have gone into this process.)  In other words, there is a 90% chance that the ‘true’ (mean) value of the measurement is within 1.645 x σ of any reported value.

Here is an example of 30 samples, normally distributed with a mean pH of 6.4.  Ninety percent of the measurements lie within the stated confidence interval, but there is no way to know which one of those results was the one reported by the soil testing lab as your soil result.  So a soil which has a pH of 6.4 is 90% likely to be reported between 6.2 and 6.6, and any single result is 90% likely to be within 0.2 units of the ‘true’ pH.

Often we are concerned with changes in soil parameters from year to year, or between applications of soil amendments or micronutrients.  We can make some estimates on the significance of changes as well from the standard deviation without knowing the mean values.

Assume we have a soil measurement from 2009 (point ‘1’) in the figure.  Knowing the standard deviation of the test method, we can calculate the width of the 90% confidence interval.  Let point ‘2’ represent the measurement from 2010.  The difference from the 2009 measurement is not greater than the 90% confidence interval.  In other words, there is a 90% chance that the samples come from a distribution which shares the same mean – there is no statistically-meaningful difference between the values.  If the 2010 measurement gave a result at point ‘3’, it lies outside of the 90% confidence interval and there is a statistically-meaningful difference in the measurements.  Note, however, that it doesn’t say anything about what the new ‘true’ mean value of the soil parameter is.  Using a similar rational, upper and lower limits can be put on the shift in mean value – this is left as an exercise for the reader.

Note that several times I have said that we can use the standard deviation of the test methodology independently of the mean value – this is true to an extent.  However, in many cases I will be presenting the standard deviations as percentages of the mean.  In these cases it will suffice to scale the standard deviation using the single-point reference value returned by the testing laboratory.

Questions, critiques, comments and corrections are always welcomed.

Four times a year, the North American Proficiency Testing program (NAPT) sends identical samples of 5 soils to participating labs.  Each sample is subjected to 80-90 different tests, though not all labs perform all of the tests.  I have been looking at data between 2007 and 2009 – a total of 60 soil samples (5 per quarter).

The NATP data gives 3 metrics for each soil sample tested (for each analysis): number of results (per analysis), the mean test value, and the Median Absolute Deviation (MAD).  Unfortunately, the MAD is an ordinal metric and doesn’t give any description of the distribution of the residuals/deviations.  I propose that the MAD numbers can be used to estimate the precision of the soil analysis methods.

Here is what I propose, along with some assumptions:

1)      Assume that the deviations about the mean (from which the MAD is determined) were normally distributed and that σ ≈ 1.4826 X MAD for each sample, where σ is the standard deviation.

2)      The test mean is a characteristic of the specific soil sample presented for analysis.  I propose that the MAD (and thus standard deviation) is a characteristic of the analysis method and can be used to characterize the precision of the specific test and method (i.e., pH (1:1 water), CEC (cation displacement), boron (hot-water extraction), etc.)  It is reasonable to assume that the same lab retesting the sample n times would result in a lower MAD than n labs each testing the sample once – but that’s not the data we have.

3)      I will check for a correlation between MAD and the test mean.

1. If there is no correlation, calculate the mean of the MAD values for the analysis method.
2. If there is a linear correlation, use least-squares to determine the best-fit ratio between MAD and the test mean.
3. If there is a non-linear correlation  – WALK AWAY.
4. The mean MAD or best-fit MAD/mean ratio relates to the precision of the test.

4)      Check for normality in the residuals using the Shapiro-Wilk test.  My confidence interval will be 90%.  If the residuals are not normal within this confidence – WALK AWAY.

5)      Calculate the standard deviation of the residuals – this should  be the same as the standard deviation of the test statistic itself (MAD or MAD/mean).

6)      The standard deviation of the residuals represents the uncertainty in the mean MAD, and thus the uncertainty of the test precision.  While this is an interesting number, we will ignore it for now in this simple analysis.

The standard deviation of the test will be defined as follows (the average MAD could also be the best-fit MAD/mean value, expressed as a percentage of the mean test value): σ(test) ≈ 1.4826 X MAD

So at a 90% confidence level, the difference between two soil measurements must be within 2 x 1.645 x σ(test).  The factor of two is required for single-point measurements because one measurement (i.e., the 2009 soil test) could have been at the lower end of the confidence level, and the 2010 test could be at the upper end of the confidence level – both having the same mean value.  If you assume the 2009 result was an accurate estimate of the mean value, then the factor of 2 can be removed.

I do not plan on performing this analysis for all of the tests methods reported in the NAPT reports.  I expect  to start with the test methods used by the Logan Labs and UMass soil testing labs.

If you have any comments on the rationale behind the method of using MAD as a test characteristic or the value of such a metric, I always welcome comments, corrections or sundry observations.

On Bestlawn.info, a lot of stock put into soil testing and the art and science of interpreting those results.  But when we see a shift in a nutrient level from one year to the next, where did that shift come from, and how significant is the result?

There are several reasons the soil results may vary from year to year:

(1) Mother Nature (nutrient leeching, mineral weathering, acid rain, etc.)

Soil systems are not static, but they do tend to move slowly.  If you understand the characteristics of your soil, you can anticipate the natural cycle of the soil chemistry.

(2)  Soil Management (purposeful amendment of the soil)

We judiciously apply fertilizers, micronutrients, organic matter, etc. to optimize the soil conditions for healthy turf.  These applications can have predictable effects on the soil chemistry.

(3)  Sampling Distribution (soil characteristics are not uniform across the lawn)

It is reasonable to expect soil conditions across a small lawn to be uniform, but there’s no guarantee of it.  And with a large lawn comes a greater likelihood of variation.  By sampling at many locations uniformly throughout the turf (all from the same depth), localized soil variations average out.

(4)  Test Methodology (different labs use different methods)

For most of the standard soil tests, there may be 3-5 ‘standard’ measurement methods – and they often don’t give results which are directly comparable to one another.  The choice of methodology may be driven by cost, equipment, expertise, or perhaps by regional consensus.  The methods used for soil testing should be obtained from the soil lab.

(5)  Laboratory Error (swapped samples, hung-over grad students)

It happens.  Most labs hold samples for a number of weeks after analysis, so if you suspect a true error, you can often contact the lab for re-analysis without submitting a new sample.

(6)  Test Variance (same sample, same test, different results)

Physical measurements are imprecise – measure twenty times, you’ll get twenty different results, and how those results differ from one another is a measure of the variance, or precision, of the test.  If a method is imprecise (i.e., results vary widely), multiple results may give the appearance of a change when none exists.

I’ll be addressing Reason 6 (‘Test Variance’) in a day or so with some real-world examples, and hopefully some guidelines.  Are there any other reasons I left out, or comments on this list?

I like graphs.  That’s probably why I like data – so it can be put into some form of informational graphic.  I don’t claim to have any proficiency in it, but playing with graphs gives me something to do – and I can always find data to play with on the internet.

One source is the Soil Science Society of America and the North American Proficiency Testing Program.  Each calendar quarter, soil samples are sent to participating soil testing labs for a suite of measurements.  Each lab receives samples from the same representative soils, so common measurements are replicated up to 30 – 70 times, depending on the procedures and capabilities of the test labs.   The quarterly NAPT results are available online – each test result represented by a median value and a median absolute deviation (MAD) from all of the laboratory submissions.

There are a lot of interesting things to look at in the data – like considering the significance of reported changes in micronutrient levels (i.e., is a 0.1 ppm shift in boron levels real, or within statistical variance?).  Or you can compare the results for the same micronutrient from different extractant methods.  But first I wanted to explore Cation Exchange Capacity (CEC) as a function of soil texture.

The CEC is a measure of how much ‘stuff’ the soil can hold – good stuff like calcium, magnesium and potassium ions, and not-so-good stuff like hydrogen and aluminum ions.  Organic matter and clay particles have high binding capacity, so soils with high clay and OM fractions will have a high CEC.  Sand particles do not hold ions well, and so sandy soils have a lower CEC.  Soil pH also has an effect.

From the 2007 – 2009 NAPT results, there are 60 soil samples with results including pH, soil organic matter, soil CEC and particle size analysis (percentage sand, silt and clay).  Most of the samples are silt loams, sandy loams and loams.

Soil Texture Frequencies

It is frighteningly easy to use regression analysis in Excel to come up with a 5-variable model (pH, %OM, %sand, %clay and %silt) to represent the data .  Even though the 5 variables aren’t independent, there are published examples of researchers doing just this.  I am more comfortable with a 3-variable model using pH, %OM and %clay, but even that’s iffy given my lack of subject knowledge.  There is also the question of which type of pH measurement to use, and should I use total soil carbon or just a loss-on-ignition measurement for organic matter?

I don’t have the background to make reasonable determinations, so I’ll stick with a simple model borrowed from the University of Minnesota (among others).  Taking the average clay particle to have a CEC of 50 meq/100g and the average organic particle to have a CEC of 200 meq/100g, I get this:

    CEC = %OM x 200 meq/100g + %Clay x 50 meq/100g

For the %OM, I’ve used the soil organic matter as determined by loss-on-ignition.  On average, this model did as well as any of the multi-variable regression models for the NAPT data set.  Even so, the utility of the model as anything more than a rough estimate for medium soils is questionable.

CEC Model Results

I’d like to be able to plot the samples on a USDA Soil Texture Triangle, with bubble sizes to indicate the soil CEC (or maybe the residuals from the CEC model).  However, Excel doesn’t do ternary diagrams, and the program I have that can generate ternary plots doesn’t support bubble plots.  But a soil diagram is still in two dimensions, so a little creativity with a 2D bubble chart in Excel produces acceptable results.  Here the bubble size is a function of the measured CEC.

CEC Plotted on Soil Texture Triangle

After all of that, the graph isn’t nearly as interesting as I thought it might be, but it shows what we already know:  sandy soils have low CEC (small bubbles), and the CEC generally increases with clay content.  It would be nice if there were more clay loam and sandy clay loam samples in the data – I wonder why they’re not well represented?

So I didn’t get an interesting graph out of this exercise, but I learned a bit about modeling soil CEC.  More importantly I have a few years of NAPT data formatted into a spreadsheet, a soil texture lookup table using the USDA guidelines for classifying texture based on mineral fractions and a soil texture triangle to put bubbles on.  That’ll do for now.

This is Cactus’ first blog post – ever. Don’t get too excited, he’s still trying to think of a good initial revelatory pronouncement.