This article from the WSU (Washington State University) "TIME TO COMPARE SOIL MOISTURE SENSORS" and was published in the following newletter:
"TIME" TO COMPARE SOIL MOISTURE SENSORS
Speaking of time, I am reminded of a proverb that states, "A man with two watches never knows what time it is." I felt much the same way as I compared eight different soil moisture sensors during the 1998 growing season. First of all, the sensors provide output in different units: centibars, inches per foot, and percent by volume. Even the sensors that use similar units rarely give the same reading. Most often, scientists strive for absolute accuracy, but I wanted to know whether relative accuracy was sufficient for irrigation scheduling. In other words, do the measured changes in soil moisture match the expected changes caused by crop water use and irrigation regardless of whether sensor readings are identical or not?
The sensors used in this trial are all being marketed and supported by companies working in Washington. The neutron probe, tensiometers and watermarks are fairly common to irrigation scheduling. However, there are many new instruments coming to the market that measure the capacitance/dielectric constant of the soil. These include EnviroScan, Troxler Sentry, AquaTel and AquaFlex.
The sensors were installed in two alfalfa plots of Warden silt loam at the WSU-Prosser Research Station. In order for results to be similar to what an irrigator would encounter in the field, no special effort was made to calibrate the sensors to this location. The companies’ calibration procedures were followed or the companies’ calibration curves were utilized.
The sensor measurements are graphed below and similar trends between sensors are obvious. The alfalfa plots were irrigated four times and these peaks are very distinct. Three deep valleys are also very distinct indicating when the alfalfa was allowed to dry out over several weeks. The neutron probe data (calibrated to gravimetric samples) was graphed with individual sensors to facilitate comparison. Only the graphs of the east plot are included due to space limitations
For each sensor, soil moisture measurements were recorded before and after irrigation and the change in soil moisture was divided by the amount of water caught in the rain gages to calculate an Irrigation Ratio shown in the first column of Table 1. In almost every case, the Irrigation Ratio was less than 1.0 which means that the measured increase in soil moisture was less than the amount of water applied. This discrepancy could be partly due to water evaporating off soil and plant surfaces before entering the soil.
Table 1: RELATIVE ACCURACY OF SOIL MOISTURE SENSORS Table 1: RELATIVE ACCURACY OF SOIL MOISTURE SENSORS
IRRIG.
RATIO
PAWS ET
RATIO
Correlatation
Value
NEUTRON PROBE EAST
0.86
0.86
1
WEST
0.78
0.87
1
ENVIROSCAN - EAST
1.13
1.14
0.97
WEST
0.88
0.94
0.96
TROXLER SENTRY - EAST
0.58
0.67
0.82
WEST
0.57
0.63
0.82
AQUATEL - EAST
0.71
0.81
0.79
WEST
0.45
0.43
0.65
AQUAFLEX - EAST
0.77
0.45
0.71
WEST
0.38
0.23
0.65
TENSIOMETER - EAST
0.4
0.42
0.86
WEST
0.39
0.34
0.68
WATERMARK - EAST
0.28
0.59
0.52
WEST
0.1
0.6
0.12
In between irrigation events, alfalfa evapotranspiration (ET) was estimated from a nearby Public Agriculture Weather System (PAWS) station while declining soil moisture was being measured twice per week. This drop in soil moisture was divided by PAWS ET to calculate an ET Ratio. The average ET Ratio for each sensor is shown in the second column of Table 1. Most of the ET Ratios are less than 1.0 similar to the Irrigation Ratios. Perhaps PAWS is slightly over estimating alfalfa ET while some of the sensors are under estimating ET.
Finally, each sensor was correlated against the neutron probe. Correlation values (R square) are shown in column 3 of Table 1. As the correlation value gets closer to 1.0, a sensor has more potential to be recalibrated to match the neutron probe results.
Soil moisture sensors that have an Irrigation Ratio, PAWS ET Ratio, and Correlation near 1.0 have relative accuracy and could possibly have a high degree of absolute accuracy. These sensors could be used in a predictive type of irrigation scheduling to determine when and how much to irrigate via direct measurement.
However, all of the sensors (as calibrated during the trial) followed soil moisture trends in a fairly stable manner and they could be useful in a reactive type of irrigation scheduling where the sensor acts as a marker i.e. start irrigation when a sensor reaches a certain mark and stop irrigating at another mark. In this scenario, soil moisture does not have to be quantified and experience in reacting to the sensors is the key factor. Most of the sensors correlate with the neutron probe and site calibration would improve their accuracy.
In addition to sensor accuracy, the costs in capitol, labor, training, maintenance, data availability, and crop compatibility should be weighed against the benefits of conserving water, saving energy, increasing yield, improving yield quality, and reducing non-point pollution. Whichever sensor you chose, time must be invested to become proficient with a new tool. Soil moisture sensors will continue to be tested by Washington State University via financial support from the Northwest Energy Efficiency Alliance
Brian G. Leib, WSU - Extension Irrigation Specialist
© Bruce Metelerkamp
www.sowacs.com/comparisons/lieb1.html
last update : 29 April 1999