Reports

In preparation for the advection setup for summer 2022, I tested several VWC sensors in the lab using a plastic bucket of soil. I tested five sensors, two digital Campbell sensors and three Delta-T sensors. 

The soil was collected from old soil samples that Joe had dumped together to re-use the plastic containers they came in. The soil is pale brown when dry and dark brown when wet. Joe thinks it may be soil collected from Tennessee Valley.

Soil sensors were wired to a CR1000 which was powered with continuous wall power. The sensors were measured every 1min and 10min averages were recorded. The Campbell probes were powered continuously and factory-calibrated VWC, electrical conductivity, soil temperature, dielectric permittivity, period, and voltage ratio were saved. The Delta-T probes were powered by SW12V (they were programmed with a 1s warmup period before being read) and raw mV and factory-calibrated VWC were saved.

To measure oven-dried VWC, I took soil cores using either used a 30ml plastic syringe with its tip cut off, or a metal ring that is commonly used for taking soil samples. I usually did 3 cores using the plastic syringe and 1 core using the metal ring for each point and averaged the results.  I emptied the cores into pre-weighed small aluminum pans and dried them at 105 degC until the dry weight was constant over half a day.

Bulk density[g/cm3] = Dry soil weight[g] / soil volume[cm3]

Volumetric water content[cm3/cm3] = Volume of water[cm3] / Volume of soil[cm3]

= ((Wet soil weight[g] - Dry soil weight[g]) * Bulk density of water[g/cm3]) / (Dry soil weight[g] * Bulk density of soil [g/cm3])

Summary

Figure 1. Oven-dried VWC compared to the VWC measured by the sensors (both rounds combined, excluding cores with BD > 1.4). Average bulk density of the soil cores is plotted in purple stars along the right-hand y-axis. Sensors are relatively consistent with each other, if not consistent with the oven-dried VWC. Maybe the Campbell CS655 sensors are more similar to each other at VWC<0.25 than at VWC >0.25.

Conclusion: It's really hard to calibrate soil moisture sensors, especially in the lab. I would not use these calibration equations to correct field data.

Next steps: There might be more information looking at the raw readings. We could also try using field soil in the lab and using a larger core (larger than a 30ml syringe) to take samples to reduce compaction. The 1/3-1/3-1/3 method I figured out in the second round of testing worked best for distributing the water homogeneously in the soil. The best calibration check would probably an in situ field calibration where we take core samples near the soil sensors across a variety of VWC conditions.

Jackisch et al., (2020 Earth Sys. Sci Data) compared 50+ soil moisture sensors in an agricultural field of sandy loam. The sensors were installed within a 14mx4m area and data was recorded through the growing season from April through May 2016. The analysis focused on response to several rain events over this period. They found (emphasis added):

Overall, the data raise substantial questions about the data quality of state-of-the-art measurement systems of soil water content based on relative electrical permittivity of the bulk soil without specific, in situ calibration. Despite delivering plausible signals, neither the absolute values nor the relative reactions to events appear to be very accurate… Although several studies evaluated soil moisture sensing systems (e.g. Walker et al., 2004; Mittelbach et al., 2011; Chow et al., 2009) and sensor calibration is known to be an issue (e.g. Rowlandson et al., 2013; Bogena et al., 2017; Rosenbaum et al., 2011), the scientific application still lacks a common procedure for evaluation and calibration of such data. When data from different sites and sensor systems are combined (e.g. Dorigo et al., 2011), our findings should raise awareness that deviations are not always a matter of soil heterogeneity.

References:

Jackisch, C., Germer, K., Graeff, T., Andrä, I., Schulz, K., Schiedung, M., ... & Durner, W. (2020). Soil moisture and matric potential–an open field comparison of sensor systems. Earth System Science Data, 12(1), 683-697.

Rowlandson, T. L., Berg, A. A., Bullock, P. R., Ojo, E. R., McNairn, H., Wiseman, G., & Cosh, M. H. (2013). Evaluation of several calibration procedures for a portable soil moisture sensor. Journal of hydrology, 498, 335-344.

I did two rounds of testing.

First round

In the first round, I let the soil equilibrate for several days after adding water before taking the soil core. After this round, I realized it doesn't actually take that long for VWC to equilibrate.

To estimate how much water to add to get an approximate VWC, I calculated the approximate volume of soil.

Soil measurements:

Height = 15.2cm

Top diameter of bucket = 28.3cm

Bottom diameter of bucket = 25.5cm

Average diameter of bucket = (28.3+25.5)/2 = 26.9cm

Volume = pi*r2*h=3.14*(26.9/2)^2*15.2=8634cm3

From this calculation, 863 ml of water should increase the soil VWC by about 10%.

For the this first point, I dumped water on top of the soil and mixed with a drill and large drill bit. Checked it with a putty knife and it seemed thoroughly mixed to us. However, upon emptying most of the soil on 2022-03-04, there was a lot of bone-dry soil on the bottom. The water didn't redistribute as much as I expected. The soil was packed tightly from the top by hand. I covered the sensors with plastic wrap to prevent evaporation.

To take the soil cores, I either used a 30ml plastic syringe with its tip cut off, or a metal ring that is commonly used for taking soil samples. I emptied the cores into pre-weighed small aluminum pans and dried them at 105 degC until the dry weight was constant over half a day.

a - Plastic syringe with tip cut off, pushed in by hand

b - Metal ring, pounded in with weight

*Much harder to push this core down than sample 2 and 3 (soil packed more densely in this area)

For the second point, I dumped water on top and mixed with a drill. I then dumped most of it out into another plastic container (yellow toolbox lined with saranwrap) so I could pack it in 3 sections. I tried to distribute the muddy/sodden clumps between the 3 sections. Each section was tamped down with a loose mallet head and the surface was lightly scratched with the putty knife before putting the next section in. When it came time to core, it was difficult to push in the syringe/metal ring by hand. I had to use the loose mallet head to hammer the syringe/ring in. The bulk density (avg of 1.4 g/cm3) was way too high for Delta soils. The good thing about this method was that when I unpacked this bucket on 2022-03-10, the soil moisture seemed homogeneous throughout the bucket, so packing the soil in several sections worked well for distributing soil moisture.

Joe found the ML2x sensor at this point, which I inserted into the soil after taking the cores to get a reading. I covered the soil and sensor loosely with saran wrap.

c - Plastic syringe with tip cut off, pounded in with weight

**Soil in bucket was packed much more densely than previous time on 2022-02-25

For the third point, I dumped most of the soil out into the yellow plastic container. A lot of clods to break up by hand. Put in 1/3 of soil, dumped in 1/3 of water, mixed with a putty knife. Tamped down by hand and scarified the surface. Repeat for the middle layer except I forgot to scarify the surface. Repeat for the top layer. After getting the CS655 and ML3 readings, I inserted the ML2X sensor to get a reading before taking the soil cores.

d - Plastic syringe with tip cut off, twisted in

Figure 2.Oven-dried VWC compared to the VWC measured by the sensors. Average bulk density of the soil cores in plotted in purple stars along the right-hand y-axis. Sensors are relatively consistent with each other, if not consistent with the oven-dried VWC.

I tested 3 points, which I expected to be at 10%, 20%, and 25% VWC based on the volume of water I added. The oven-dried VWC I collected with the cores were at 15%, 27%, and 26% VWC. I'm not sure why the third point had such a low oven-dried VWC given that I added 400+ml of water between the second and third points. I didn't see any visual evidence of evaporation between the points and it's hard to imagine 400+ml of water evaporating from soil kept at lab temperature over a few days.

I think this weirdness has to do with the inconsistent bulk density produced by how I packed the soil into the bucket and how I took cores: bulk density jumped from ~1 to 1.4 g/cm3 between the second and third points.

Second round

For the second round of testing, I decided to take the soil core samples soon after the VWC measurement had stabilized and I tried to be more consistent in my packing and coring methods.

For the first point, I took the initial soil moisture measurement without adding any additional water. There was still some residual moisture from the first round of testing about a month prior; it was not quite air dry. I spent some time breaking down the clods that had formed in the soil (using the manual flour sifter Joe found in the street). Ants had laid eggs under some of the clods and they dispersed in a hurry after I started agitating the soil.

I first measured the soil's bulk density with no compaction and without adding water. I measured the dimensions of the soil in the bucket and weighed the soil on a scale (subtracting the bucket weight).

I calculated the soil has a bulk density of 1.12 g/cm3.

Soil measurements:

Height = 17.0cm

Top diameter of bucket = 28.3cm

Bottom diameter of bucket = 25.5cm

Average diameter of bucket = (28.3+25.5)/2 = 26.9cm

Volume = pi*r2*h=3.14*(26.9/2)^2*17=9661.5cm3

Total mass (weighed soil on scale) = 10.852kg (slightly moist soil)

Bulk density is 10852g/9661.5cm3=1.12 g/cm3

All these samples were taken on 2022-04-12.

e - Plastic syringe with tip cut off, pushed in by hand to "15ml" mark and then plunged down to "10ml" mark to make up for some compaction and to get a clean edge

For the second point, I added 450ml of water total. I dumped about 1/3 of the soil to the bucket, added about 1/3 of the water into the soil, and mixed. I smoothed the surface before repeating twice. The final height was about 16-17cm, so the bulk density should be about the same as the first point. There are definitely more clods after adding this much water, so there might be some more air holes in the soil.

This should have added ~5% VWC to the soil. Oven-dried VWC increased from 15% to 23%, which is higher than I expected but sure I can believe that.

***Some soil fell out so packed some more in by hand

For the third point, I added another 450ml of water, using the same 1/3-1/3-1/3 method as above.  The final height of the soil was ~14cm, so the bulk density will definitely be higher in this sample. More clods than the previous point, but not sure there's anything to be done about that. I don't want to increase the bulk density even more by compacting the soil.

Again, this amount of water should increase the VWC by ~5%. Instead, the oven-dried VWC increased from 23% to 52%. There is definitely something wrong with my packing or coring technique. I would not use this data when trying to find a calibration coefficient for the sensors.

f - Plastic syringe with tip cut off, pushed in by hand to "15ml" mark and then plunged down to "5ml" mark to make up for some compaction and to get a clean edge

Figure 3.Oven-dried VWC compared to the VWC measured by the sensors (second round of testing). Average bulk density of the soil cores in plotted in purple stars along the right-hand y-axis. Again, sensors are relatively consistent with each other, if not consistent with the oven-dried VWC.

For this second round, I was expecting the VWC to be about 15%, 20%, and 25% based on the amount of water I added to the soil. Instead, the oven-dried VWC was 15%, 23%, and 52%. Again, the last point is the most inconsistent point.

There was a lot of compaction when taking the soil core sample, which made it hard to know what volume of soil was really sampled.