Summary of M.K. Hansen AM400 and Onset HOBO Datalogger and Watermark Sensor Demonstration and Testing near Twin Falls, Idaho during 2000

Dr. Richard G. Allen
University of Idaho,
Kimberly, Idaho

Purpose of the Study

The purpose of the study was to demonstrate and monitor two relatively low-cost and continuous datalogging systems for measuring soil water potential using Watermark sensors. The intent is to encourage adoption of an irrigation management system that can provide continuous and real-time or near real time information to the irrigation system operator regarding the current status of soil water content as well as recent trends in water content. The equipment selected for this study is so inexpensive and user friendly that it has the potential to be used on all crops, even low value crops which normally do not command a higher levels of water management attention. This information can help the operator in assessing past performance of the irrigation system, current status of the plant root zone, and decision-making in regard to future system operation.

Background and Equipment Used

The two datalogging systems demonstrated were the AM400 logger manufactured by M.K. Hansen Company of East Wenatchee, WA (http://www.mkhansen.com/) and the HOBO H8 logger manufactured by Onset Computer Corp. of Bourne, Mass (http://www.onsetcomp.com/). The AM400 logger has a real-time graphical field display that presents soil water potential information to the observer at any time. Approximately 35 days of information can be presented in the field, enabling the observer to evaluate trends in and history of water potential over time in the environment the data are taken. In addition, there is sufficient memory to record more than a season's worth of data for later download and analysis. The HOBO loggers are passive loggers, with no display and with adequate but less memory than the AM400. Data must be transported to a notebook computer or data shuttle during site visits for processing and before presentation for decision-making. Both logger types are battery operated and require no external energy source.

The cost for the HOBO H8 loggers is approximately $85 for the "indoor" type (described later) and $169 for the outdoor "industrial" unit. A HOBO "data shuttle" can be purchased for $159 for use to transfer data between multiple HOBO loggers and a computer, or a notebook computer can be attached directly to the loggers during site visits using a serial interface. The retail cost for the AM400 unit is approximately $400. Data from the AM400 can be transferred to a notebook computer in the field using a serial interface.

Soil water potential relates to the energy level (i.e., "potential energy") of water in the soil. In an unsaturated soil system, water potential has a negative value. The more negative the value, the lower the energy level relative to a saturated condition at atmospheric pressure. The more negative the potential, the more difficult it is for roots to extract water from the soil. Soil water "potential" is the opposite in sign of soil water "tension." Potential is commonly expressed in units of kiloPascals (kPa), which is equivalent to the older term of centibars (cb).

The Watermark sensor, manufactured by Irrometer Corp (http://www.irrometer.com/), is a granulated matrix of a sand-type of material mixed with gypsum and encased in a stainless steel housing. Resistance to flow of electricity is measured between two electrodes within the matrix. The resistance of the sensor changes with water content of the sensor. Because the sensor is in direct contact with the soil, the water content (actually water potential) of the Watermark changes with water content of the surrounding soil. Irrometer has established a conversion table for the Watermark model 200SS that can be used to convert from resistance readings into soil water potential. An equation was developed by Shock et al. (1998) using experimental data that can also be used to convert the resistance readings into units of potential, although it may only apply to potentials above -80 kPa. An additional set of equations for reproducing the Irrometer table was developed during this study and is described later.

Watermark Soil Moisture BlockThe Watermark sensors are about 3 inches (75 mm) in length and 7/8 inch (22 mm) in diameter. Irrometer Corp. recommends that the Watermark sensors be "read" using an alternating current (AC) to reduce the effect of electrolysis on the reading (discussed later) and/or to reduce "polarization" of the sensor caused by sustained application of a direct current. However, it appears that direct current (DC) can be used to read the sensors if applied for very short periods of time (less than 0.5 second duration) and infrequently (only a few times per day).

Installation

In late April of 2000, two sets of Watermark sensors were installed in an established alfalfa field 8 miles southwest of Twin Falls, Idaho. The cooperator's name is Jeff Ward. The irrigation system is a center pivot lateral equipped with spray nozzles. The system is supplied with water from the Salmon Falls Canal Company canal system. Two sets of Watermark sensors were placed at 6, 18, and 30 inch depths (150, 450, and 750 mm). One set of sensors was connected to an M.K.Hansen Company AM400 datalogger/display unit. The other set of sensors was connected to two Onset Computer Corp. HOBO datalogger units. The systems were operated from May 1, 2000 until October 1, 2000.

The Watermark sensors were installed using a 1.0 inch (25 mm) diameter probe/bulk density sampler coupled with a 7/8 inch (22 mm) push probe. The sensors were cemented to 3 ft. (1 m) lengths of ½ inch IPS/315 PSI/SDR 13.5/1120 PVC pipe. The outside diameter of the pipe was 7/8 inch (22 mm). Each sensor was placed with the PVC pipe extending above the ground. Sensors were pushed into a 5 ml slurry placed at the bottom of the hole. The slurry was made of local soil material and water.

The sensors were placed 30 ft (10 m) inside the outer tower of the center pivot, about 100 ft (30 m) from the outside of the field. Horizontal spacing between sensors was about 10 inches (250 mm). The order of spacing was 6, 18, 30, 6, 18, 30 inches, from north to south, parallel to the travel of the center pivot.

One set of Watermark sensors (at 6, 18, and 30 inch depths (150, 450, and 750 mm)) was attached directly to an M.K. Hansen Company AM400 datalogger . A Hansen soil temperature probe was also attached to the AM400 datalogger and was inserted to a depth of 6 inches. The AM400 unit was mounted on a 12 x 15 inch (300 x 380 mm) plywood board (1/2 inch (12 mm) thickness) that was oriented vertically and attached to a 1.5 x 1.5 inch (38 x 38 mm) wooden pole. The unit was positioned at about 30 inches (750 mm) above the ground surface and was oriented so that the display faced to the north.

The AM400 unit was shipped with a plastic cover for the display surface. However, at the request of the manufacturer, the cover was not used in order to test the durability of an unprotected AM400 unit when placed beneath a center pivot spray system. The AM400 unit was not waterproofed and had some small openings in the back of the case to provide for circulation of air and drying. The unit functioned well during the demonstration period, with no malfunctioning noted. The interior of the AM400 remained essentially dry due to the air circulation and protection of the back of the unit by the plywood mounting board.

The AM400 datalogger uses an alternating current (AC) to read the resistance of the WM sensors. Multiple sensors are read using an internal multiplexing unit. Sensors were scanned each 8 hours.

The second set of Watermark sensors was attached to two different Onset HOBO H8 "4-Channel External" dataloggers by splicing between channels of the two loggers. Each Watermark sensor was attached to the dataloggers using a 10 kOhm resistor wired in series as described in the attached appendix. Sensors were read by measuring the voltage drop across the half-bridge circuit. The HOBO dataloggers use direct current (DC) to read the WM sensors. The total reading time for each HOBO logger was about 0.7 seconds for the three Watermark sensors. Sensors were monitored every eight hours.

The first type of HOBO datalogger used was the "indoor version" H8 unit that is non weather-protected. The indoor unit excites all four channels (sensors) simultaneously and concurrently as it reads the voltage of each channel in succession. This causes some bias in readings for WM sensors due to localized electrolysis of water that is in direct contact with the WM electrodes. The electrolysis begins within a few milliseconds after initiation of excitation and creates micro-bubbles of gas and vapor that increase the effective resistance of the sensor reading . Therefore, channels two and three generally have a progressively larger bias in the resistance reading as compared to channel one. The indoor H8 unit was placed inside a sealed 4 inch (100 mm) Rubbermaid plastic container along with a large supply of desiccant in a perforated plastic bag.

The second type of HOBO datalogger was an industrial H8 unit housed in a hermetically sealed enclosure for direct placement outdoors. The particular unit used in this study had been specially retrofitted at the Onset factory in 1999 to separate the excitation of specific channels so as to minimize the degree of electrolysis and sensor bias. This particular "retrofitted" model is now produced by Onset for sale to client OEM's. It is now listed as the Model HOBO 4-Channel External IP.

The data logger clocks for the two HOBO units were set to the same date and time in the field using the Onset data shuttle during transfer of data during site visits. There did not appear to be significant conflict between the two units (connected to the same three sensors) due to the nearly identical sensor clock times, except for perhaps the 18 inch sensor when read by the "indoor" style of HOBO unit, as discussed later.

The AM400 logger is programmed at the factory to scan the WaterMark sensors once every 8 hours (three times per day). The HOBO loggers can be user programmed for a wide range of datalogging frequency. In this study the HOBO loggers were programmed to scan the WaterMark sensors every 8 hours to be consistent with the AM400 time series.

The systems were visited about once each two weeks for monitoring purposes. Data were retrieved from the dataloggers each two months.

Water Management

The 2000 irrigation season was one of low water supply for the Salmon River Canal Company. Therefore, many systems, including the center pivot system of the cooperator, were operated in a conservative and sometimes deficit mode.

The cooperator's alfalfa field was harvested four times during the 2000 season by chopping for forage. The irrigation system was shutoff two or more days prior to the harvest and was generally not turned on for several days following harvest. During this time the soil water content (i.e., soil water potential) was observed to decrease substantially.

The center pivot system of the cooperator was generally operated on a one-day rotation. Application depth per rotation was estimated to be about 1/3 inch (8 mm). The high speed rotation and deficit nature of irrigation caused the soil profile to dry at the 18 and 30 inch depths during the main portion of the irrigation season, with only the 6 inch depth responding to individual irrigations. The relatively high frequency irrigation used here may have resulted in a higher percentage of applied water evaporating from the frequently wetted soil surface as compared to a less frequent scenario (for example, a deeper and less frequent 2.5 to 3.5 days irrigation). Evaporation from the soil surface is generally a lower total percentage of total applied water for a lower frequency system due to the drying of the soil surface between irrigations and consequent deeper infiltration of larger depths of water per irrigation. This center pivot system was operated in the high frequency mode in an attempt to reduce surface runoff.

Results

AM400. Soil water potential at the 6, 18 and 30 inch (150, 450, 750 mm) depths as measured and recorded by the AM400 are plotted vs. time of the year in the following figure 1a-f for the months of May through September and October 1-9. The trend for medium dry to moist conditions at the 6 inch depth, with dry to very dry and decreasing water potential and moisture at the 18 and 30 inch depths is obvious in these figures.

The impact of individual, daily irrigations (generally occurring every three data points) are discernable at the 6 inch depth, as well as are the dry downs preceding harvest of alfalfa in late May, late June, late July, and mid September. Significant rain events (as measured at the Twin Falls Regional Airport four miles (6 km) to the east) occurred on May 7 (approx. 0.7 inches (17 mm)), May 16th (approx. 0.3 inches (8 mm)), and Aug 31-Sept 2 (0.4 inches (10 mm). The center pivot system was turned off during the mid-May period.

A shortcoming of the current version of AM400 logger is noticeable in the figures, in that the instrument is programmed to read soil water potential to only -100 kPa (kiloPascals or -100 centibars). When readings drop below -100 kPa, the AM400 readings become asymptotic to -99 kPa, beyond which the logger reports no reading. Soil water potential values of -100 kPa or higher are associated with stress points for many sensitive agricultural crops. Soil water potentials of -20 to -30 kPa are commonly associated with the upper level of long term soil water retention in the soil (i.e, field capacity) and potentials of -1000 kPa to -1500 kPa are associated with permanent wilting point (crop depth).

Under agronomic practices intended to maximize crop production, the soil water potential is usually managed to maintain levels between -30 and approximately -80 kPa within the effective rooting zone. Under water management practices where water supplies are limited, or where economics dictate deficit irrigation, or where the soil water profile is dried prior to harvest, the average soil water potential in some fields may frequently draw-down to below -100 kPa. As shown in later figures for data recorded by the HOBO loggers, the WaterMark sensors may provide consistent readings to about -600 kPa.

Based on readings taken by the AM400, the center pivot system could have been operated, given adequate water supplies, to apply more water and perhaps less frequently (i.e., at lower rotating speed of the lateral) in order to maintain more moisture at greater depth to promote root extraction and evapotranspiration while reducing the percentage of evaporation from the soil surface. The very low potential readings and decreasing water content at the 18 and 30 inch depths indicate that the alfalfa crop was "mining" water from the soil profile and was probably having difficulty in supplying the climatic demand for water. The alfalfa crop was noted to be somewhat stressed at various times of the growing season, although no direct measurements of stress were taken. The effective rooting depth of the alfalfa crop was estimated to be about 5 ft (1.5 m).

It is common for irrigators to not follow field information in the first contact season, and for many reasons. However, often, when the irrigator/farmer has time during the off-season to reflect on the impact of recorded irrigation management related data, he often dramatically changes field management the following season. The cooperator did indicate an appreciation for the AM400's display of information in the field and that the system confirmed what he already knew: that his system was not keeping up with water requirements. He noted that the failure of the AM400 logger to display data when potentials decreased to below -100 kPa was somewhat disconcerting.

 

Figure 1. Readings of soil water potential by the M.K. Hansen AM400 logging system and Watermark sensors for the 6, 18, and 30 inch depths during year 2000, by month.

HOBO loggers and Comparison with the AM400 data. The HOBO systems measured and recorded voltage drops across Watermark sensors that were placed in series with 10 kOhm resistors. The AM400 uses a built in equation to determine soil water potential. The voltage recordings by the HOBO units, after transfer into a spreadsheet program, were converted into resistances using a half-bridge transformation. The resistances were then converted into equivalent soil water potential using one of three equations, depending on the resistance level. The three equations, intended to best reproduce the calibration table from Irrometer, are as follow:

For R = 1 kOhm, a linear relationship is used:

...................................................(eqn. 1)


where
P is soil water potential in kPa
R is measured resistance in kiloOhms
T is sensor (soil) temperature in Celsius

The 0.018*(T-24) term in equation 1 represents the 1.8% shift in resistance readings per one degree C change in temperature from a 24oC base. This temperature correction is recommended by Irrometer. The value 1.8% per C is equivalent to 1% per degree F.

For 1 kOhm < R = 8 kOhm, the curvilinear equation 8 of Shock et al. (1998) is used:

.............................................................(eqn. 2)

The Shock equation uses a different means for incorporating a temperature correction. This equation was found to perform better than any other regression equations developed during this study (against the Irrometer table) for the range 1 kOhm < R = 8 kOhm. This range is equivalent to about -10 to -48 kPa potential.

For measured resistance > 8 kOhm, the following quadratic equation, developed during this study, is used:

......(eqn. 3)

The 1+0.018*(T-24)) adjusts for sensor temperature. Equation 3 was determined by least squares regression from the Irrometer table for the range of -10 kPa > P > -200 kPa. The coefficient of determination r2 = 0.9996 and standard error of estimate was 1.07 kPa.

The equation by Shock et al. was developed using soil water potential data in the range of -10 to -75 kPa, so that applying this equation beyond this range is not recommended. In addition, the equation tends to underpredict values for potentials below about -100 kPa in the Irrometer table.

Six-inch depth. Figures 2a-e show comparisons by month for sensors at the 6 inch (150 mm) depth. One sensor at the 6 inch depth was monitored by the AM400 logger and the other sensor at the 6 inch depth was monitored by the two HOBO loggers. The two sensors were separated by about 3 feet (1 m).

The figures for May, June, July, and August are similar in regard to comparisons between the logging systems. All three sets of readings tracked one another closely when readings of soil water potential were above about -80 kPa (or centibars). The readings by the AM400 were generally slightly above those by the HOBO units. The downward bias by the HOBO units may have been caused by effects of using direct current, as discussed previously, or may have been caused by differences between two Watermark sensors, either in construction or in placement and location in the soil. Another cause of difference between the AM400 and HOBO systems might be a difference in the conversion of the resistance reading into units of soil water potential. However, the AM400 system is based on Irrometer conversion table as were the equations listed above that were used to convert the HOBO resistance data.

When soil water potential readings declined to below about -80 kPa, the readings by the AM400 tended to assymptotically approach the minimum reading of -99 kPa tolerated by the logger. The readings by the two HOBO units continued to decline past -100 kPa and generally "accelerated" downward. This type of behavior is expected for a drying soil, since the slope of soil water potential / soil water content "steepens" as the soil dries.

It is unclear whether AM400 readings at potentials below -80 kPa were biased upward by the design of the logger's measurement system and whether the readings by the HOBO units were biased downward by effects of using the direct current. It is possible that both types of biases may have been occurring. No independent measurements of soil water potential were made to confirm the accuracy of either logging system.

The "indoor" HOBO datalogger, even though connected to the same Watermark sensor as the retrofitted industrial HOBO datalogger, produced readings that were below those of the industrial unit. This downward bias demonstrates the need to use the HOBO Model 4-Channel External IP retrofit if accurate readings of potential are required. Both systems tracked one another closely, although the departure of the indoor unit from the retrofitted unit increased as recorded soil water potential decreased.

The readings for the AM400 datalogger went off scale for the 6 inch depth in early September, as the upper soil water profile dried. During this period, the readings by the HOBO units were above those for the AM400 logger. It is unclear whether the soil around the second Watermark sensor was wetter than that around the sensor monitored by the AM400 logger, or if there was some type of shift in calibration of the individual sensors.

 

Figure 2. Soil water potential readings at the 6 inch (150 mm) depth by the AM400 and HOBO dataloggers during 2000 by month.

 

Eighteen-inch depth. The soil water potential readings for the two Watermark sensors placed at 18 inch (450 mm) depth are shown in figures 3a-e for the months of May - September.

The readings by the retrofitted HOBO logger tended to oscillate during periods in May, July, and August. The cause of the oscillation is unclear. It may have been caused by a loose connection or short in the cable connection to the datalogger or it may have been caused by interference between the two HOBO loggers, which were connected to the same sensors. The lower curve for the HOBO retrofit is more likely the "correct" one. The cable to the indoor HOBO logger was purposely disconnected between May 5 and May 12 (figure 3a) to determine if datalogger interference was the cause of the oscillation of the retrofitted HOBO. This did not appear to change the behavior of the retrofitted logger, so that there was probably no interference between the two systems.

The readings by all three systems followed similar trends in May, with the HOBO indoor unit producing readings that were below those of the AM400 and the retrofitted HOBO logger producing readings that were equal to or higher than the AM400 logger. The indoor HOBO datalogger produced lower readings than the retrofitted model, even though attached to the same Watermark sensor, because of the electrolysis bias effect caused by the concurrent excitation of all channels of the indoor unit and consequent longer excitation time with the direct current before a reading was made. The bias between the two HOBO units seemed to increase as soil water potential decreased.

The readings by the AM400 logger at the 18 inch depth went off scale (< -99 kPa) in late May and did not come back on scale until the eighth of June (figure 3b). The readings from the two system types (and Watermark sensors) were very different at the 18 inch depth during June and July. Much of the difference may have been caused by differences in root extraction of water during the drying of the soil profile. Some of the difference may have also been caused by the apparent tendency of the AM400 to asymptotically approach the -100 kPa reading limit, even though this asymptotic trend runs contrary to the behavior that one would expect from a drying soil. For example, in both June and July (Fig. 3b and c), the soil water potential reading by the AM400 asymptotically approached and then "lingered" at the -99 or -100 kPa reading. This behavior is highly unlikely by water stressed alfalfa vegetation, which would continue to extract water from the soil to potential levels as low as -1000 to -2000 kPa. It is possible that some of this asymptotic behavior could stem from the implementation used to convert ohms to kPa of soil water potential within the AM400 logger.

The readings by the HOBO units for the 18 inch depth fell to below -500 kPa during periods of extreme soil drying, but tended to recover and increase as the irrigation system rehydrated the soil profile during periods of vegetation regrowth following forage harvests when the ET was low.

 

Figure 3. Soil water potential readings at the 18 inch (450 mm) depth by the AM400 and HOBO dataloggers during 2000 by month.

Thirty-inch depth. The soil water potential readings for the two Watermark sensors placed at a 30 inch (750 mm) depth are shown in figures 4a-e for the months of May - September. The Watermark sensors at 30 inches rapidly declined in early May, following installation, to readings that were below -100 kPa (figure 4a), so that the AM400 logger was unable to produce any values for the 30 inch depth.

The bias in the indoor HOBO readings as compared to the readings by the retrofitted HOBO logger are substantial and tended to increase at lower readings of potential. A period of significant irrigation and possibly unrecorded precipitation around June 17th (no rainfall was noted at the Twin Falls Regional Airport four miles to the east and only 0.09 inches (2 mm) was recorded on June 12th)) appeared to increase the soil water potential readings at the 30 inch depth (figure 4b), after which time the readings began a gradual decrease. No measurements of precipitation were made at the study site.

It is interesting that the HOBO readings at the 30 inch depth increased during much of August (fig 4d) and were even above readings taken at the 18 inch depth (fig 3d). The increase in water potential at 30 inches during mid August, when water potential at the 6 inch depth was decreasing (fig 2d) and that were above those at the 18 inch depth is counterintuitive and is unexplained. It is possible that some of the increase in readings was due to an increase in soil temperature at the 30 inch depth, but no recordings of soil temperature at that depth were made. Soil temperature at the 6 inch depth (figure 5) increased from about 13 degrees Celsius (oC) in May to about 20 oC in July and then decreased beginning in mid August to about 15 oC by mid September.

 

Figure 4. Soil water potential readings at the 30 inch (750 mm) depth by the AM400 and HOBO dataloggers during 2000 by month.


Figure 5. Soil temperature at 6 inches (150 mm) depth beneath alfalfa during 2000.

Summary and Conclusions

Both data logging systems (AM400 and HOBO) functioned well during the growing season and performed as advertised by the manufacturers. Measurements by the two types of logging systems coincided well when soil water potentials were above about -80 kPa. The retrofit of the Onset HOBO logger to separate the excitation of channels appears to be important, with the bias caused by lack of separation of channels to increase with decreasing soil water potential and with the order of the channel (1, 2, 3, or 4) in the reading sequence.

The AM400 system's display of data in the field was very beneficial for irrigation decision-making, where both the current measurement of soil water potential and 35-day history showing trends were useful. The AM400 logger could be improved by allowing the logger to measured soil water potentials that are below -100 kPa.

The HOBO units, being less expensive than the AM400 unit, are useful for monitoring and storing soil water potential data for purposes of evaluation of irrigation water management periodically during the growing season or at the end of the growing season. The data collected by the HOBO units could be used for near-real-time irrigation scheduling, but only if the data are downloaded to a computer, converted to units of soil water potential, and plotted.

Both types of datalogging systems, coupled with the Watermark soil water potential sensors represent relatively low-cost and useful systems for irrigation water management. There are uncertainties in readings at soil water potentials below about -80 kPa, both for the AM400 and for the HOBO units, with no independent observations to confirm either datalogging system. The AM400 logger perhaps appropriately terminates readings at -100 kPa to avoid these uncertainties, although there is some question regarding the accuracy of readings by the AM400 in the range of -80 to < -100 kPa. Moreover, providing estimates of potential at levels below -100 kPa, as was possible with the HOBO systems, has value when the instrument is used to observe trends and patterns.

The AM400 system's display of data in the field was very beneficial for irrigation decision-making, where both the current, real-time measurement of soil water potential and 35-day history showing trends were presented in a single graphic. The AM400 logger could be improved by allowing the logger to measure soil water potentials that are below -100 kPa with sensors capable of measuring those values.

The HOBO units, being less expensive than the AM400 unit, are useful for monitoring and storing soil water potential data for purposes of evaluation of irrigation water management periodically during the growing season or at the end of the growing season. The data collected by the HOBO units could be used for near-real-time irrigation scheduling, but only if the data are downloaded to a computer, converted to units of soil water potential, and plotted.