New technology for measuring and advising on soil water

Author: Neal Dalgliesh and Neil Huth, CSIRO Ecosystem Sciences, Toowoomba | Date: 05 Mar 2013

Neal Dalgliesh and Neil Huth, CSIRO Ecosystem Sciences, Toowoomba

Take home message

Soil water can be effectively monitored to assist managers in crop decision support. However, using currently available technologies it is unrealistic to think that highly accurate estimates of plant available water will be possible given the inherent variability of northern cropping soils and currently available sensor technologies. Given the vagaries of the system, there are a number of technologies which will provide a level of information useful in decision support without excessively high levels of investment.

Introduction

Stored soil water is one of the most important drivers of crop production in the summer rainfall dominant, northern cropping region. Traditionally agronomists and farmers have made estimates of availability based on their intuitive experience or through the use of a push probe or soil coring. With increasing interest in obtaining real-time information on soil water status and increasingly limited labour to undertake intensive gravimetric monitoring a study was undertaken to investigate technologies with potential to improve delivery of information to farmers on the cracking clay soils.

Methodology

A replicated (4) trial was undertaken in 2010-2011 on a black vertosol soil (Mywybilla; 60-70% clay; PAWC to 180cm-wheat 250 mm, cotton 300 mm) at Norwin on the Darling Downs. A sequence of 3 crops (wheat; maize; wheat) was planted over the period with soil water monitored over both fallow and cropping phases. A suite of readily available monitoring technologies were evaluated for their ability to accurately measure soil water across the range of moisture contents, suitability for use on cracking clay soils and appropriateness for deployment in commercial agriculture. Three in-situ real-time sensing technologies (capacitance, gypsum block and time domain reflectometry-TDR), 2 mobile devices (Electromagnetic Induction-EMI, Neutron Moisture Meter-NMM) and 1 modelling technology (APSIM) were evaluated using gravimetric moisture content as the control (Table 1).

Results

There are positives and negatives associated with all of the tested devices with no one likely to meet all requirements (Table 1). By their very nature, in-situ devices or those that rely on an access tube, can only be located at one geographic point and yet are expected to adequately represent the moisture environment of a much larger area. If a high level of accuracy is required then the solution is to increase replication but this brings with it additional capital and maintenance costs. Many in-situ devices provide the capacity for continuous logging of soil water which, to some, will be the most important consideration. Others will see expediency in measurement, mobility, portability and an absence of physically located in-situ devices as being more important than continuous logging. Others will consider the simulation of water availability as being sufficiently accurate to make cropping decisions. The final decision will be up to the individual and will be based on their views of the level of accuracy required, the necessary frequency and timeliness of information delivery, soil type and salinity level and the investment and continuing commitment required.  

How accurate is soil water monitoring?

There is a perception that it is possible to monitor field-scale soil water at high levels of accuracy, the reality however, is that this is difficult using currently available point scale technologies, the inefficiencies associated with monitoring and the variable nature of Australian soils. But one could also consider whether an absolute measure of soil moisture is necessary. Is knowing the plant available water (PAW) to within +/-5 mm likely to change a management decision, or is it more likely that decisions will be based on a PAW level of +/- 25 or even 50 mm?

In many ways this decision is taken out of the operators hands anyway. The assumption on which this research was based was that gravimetrically measured soil water would act as an experimental control. However, it quickly became apparent that this was not a logical premise, given the high levels of soil variability found in the apparently homogeneous research area of 420 m2. Analysis of gravimetric sampling data taken over the life of the project (at 12 points per sampling (4 reps x 3 locations)) showed a variability around the true mean of +/-15 mm to a depth of 90 cm (at a confidence level of 95%). Such a result highlights the difficulties associated with broad scale gravimetric sampling, or in selecting a point, or a number of points to locate in-situ monitoring devices. Add issues associated with individual monitoring technologies to the mix and it becomes clear that attempting to obtain a definitive measure of soil water is very difficult.

Individual technologies

TDR

TDR sensors (Campbell 616) operate effectively between 20% volumetric water content (VWC) and 50% VWC with reduced accuracy as VWC approaches Drained Upper Limit (DUL) on clay soils. Water contents above 50% VWC are clipped until water content drops into the measurable range through crop water use, evaporation or drainage. This can be an issue in heavy clay soils where it is not uncommon for the DUL to be >50% VWC and for soil moisture levels during the season to reach this or higher levels in both irrigated and dryland systems. In the case of the Mywybilla soil on which the experiment was conducted, DUL ranged between 48 and 52% VWC in the top 90 cm of the profile. TDR sensors are more difficult to install than other sensor types due to the requirement for either a pit or large auger holes to gain access at depth. This may increase costs compared to other devices. Good sensor/soil contact is essential if quality and consistent data output are to be achieved. Whilst not as prone to the impact of soil cracking as surface installed sensors (such as many capacitance probes or NMM access tubes), it is likely that probes located near the soil surface will be affected by seasonal cracking with the potential to impact on data quality.

Given the above caveats, TDR was highly effective at capturing short-term water dynamics in the root zone of the crop. Daily time series of TDR estimates for VWC and volumetric soil water collected gravimetrically are presented in Figure 1.

Capacitance

The EnviroSMART soil water capacitance probes were logged at 10 minute intervals and daily volumetric water content (VWC) for each layer generated using a locally developed calibration derived from regression analysis of sensor output and measured VWC. As the zone of measurement surrounding the capacitance probe is quite small (80% of signal sensitivity occurs within 25mm of the outside of the casing; Paltineau and Starr 1997), installation that results in good soil/device contact, without the creation of air voids, is essential for accurate soil water measurement. The issue of good soil/device contact is further heightened when these devices are used in shrink/swell soils where cracks not only interfere with measurement accuracy but also provide a preferential pathway for water entry which may result in non-representative moisture concentrations around the device.  Capacitance probes also suffer from the ‘topping’ of data at ~50%VWC in heavy clay soils. Overall this technology did a reasonable job of measuring soil water although did show instances of data topping (Figure 1) and of water entry at depth.

Neutron probe

NMM has a long track record in soil water measurement for research and in irrigated systems. A significant advantage of NMM is the large soil area in which water measurement takes place (15-25 cm radius of the source depending on soil texture), which has the effect of improving overall measurement accuracy.  However, measurement accuracy is affected by soil cracking in the same way as other in-situ devices and by preferential water flows around the tube. While generic calibrations are provided by the manufacturer, a local calibration model derived using a regression analysis of NMM output and measured water content was used in the determination of VWC. NMM measurements were undertaken every 2 to 4 weeks and generally showed acceptable correlation with gravimetric measurements (Figure 1). NMM did an adequate job of soil water monitoring but was affected by soil cracking and preferential water entry as the soil profile dried through crop water use.

Electromagnetic Induction (EMI)

Once calibrated the EM38 can provide estimates of stored soil water comparable to those produced by NMM. The advantages of this device are its portability, ease of use and ability to provide an estimate of VWC based on a large soil volume (1 m long x 1.5 m deep). As a result of the portable nature of the device it is possible to select measurement points unaffected by cracking. Currently it is not possible to apportion soil water to particular layers within the profile although techniques are being investigated for this purpose.  Use of EMI can be strongly impacted by soil salinity.

Soil water potential (Gypsum block)

Watermark 200 soil matric potential sensors were installed vertically at 4 depths in adjacent core holes (1 sensor per hole each 20 cm apart) with sensor resistance logged at 10 minute intervals and processed to daily mean values for analysis. As this device measures soil matric potential, calibration was required to convert output to VWC. Gypsum block sensors are able to provide a reasonable estimate of VWC when soils are wet (higher matric potentials) but are not well suited to measurement at lower potentials (dry soils) making the device less suited to dryland systems. Response time to moisture change is slow relevant to other devices tested (hours versus minutes). Because the units are installed from the soil surface, this device also suffers from preferential water flow and soil/device contact issues in cracking soils.

Simulation

The APSIM and Yield Prophet simulation models are able to predict daily soil water based on local weather and soil PAWC information using the SoilWat water balance model which has been used for many years in the prediction of crop production. The soil at Norwin was characterised for PAWC and the model run for the period of the experiment. Figure 2 compares predicted daily water (mm) with gravimetric and EM38 monitoring showing a comparable estimate of soil water to the other devices.

Conclusions and recommendations

a)  Devices for soil monitoring: In-situ devices that have relatively small zones of measurement and rely on good soil/sensor contact to measure soil water are at a disadvantage in shrink/swell soils where soil movement and cracking are typical. This is more important in dryland than irrigated systems where seasonal soil water levels vary from above field capacity through to wilting point or lower. Consequently, the potentially high levels of error associated with cracking and soil movement and high levels of inherent soil variability mean increased device replication would be necessary to achieve confidence in results. This however comes at an increased capital cost. Some devices (capacitance, TDR) also have an upper measurement limit over which they are unable to accurately measure soil water. This may be an issue on high clay soils where moisture content at DUL is likely to be >50% volumetric, the common limit for these devices.

In comparison, the use of a portable EMI device to measure bulk electrical conductivity and calculate soil water has a number of advantages. EMI is quick, allowing for greater replication, measures the soil moisture of a large volume of soil (to 150 cm depth), is not affected by cracking or soil movement and does not require installation of an access tube thus making it available for use on multiple paddocks. The down-sides are that it is unsuitable for use in saline soils and does not apportion soil water to particular layers within the soil profile.

b)  Calibration of monitoring devices: Electronic monitoring tools require calibration to convert the device output signal into information easily understood by the user e.g. millivolts to volumetric soil water or plant available water. This process requires the development of a relationship between sensor output and physically measured soil moisture content at moisture levels from dry to wet. The resulting calibration is then used to convert device output signal to gravimetric or volumetric water content.

To calculate the availability of soil moisture for crop use (in millimetres of available water) requires the further processing of the data and knowledge of a soil’s PAWC. A suitable characteristic may be identified from the APSoil database or SoilMapp, or electronic sensor output used to identify the soil’s water content operating range and reasonable assumptions made on values for drained upper limit and crop lower limit. An alternative is to use Soil Water Express (Burk and Dalgliesh 2012), a tool which uses the soil’s texture, salinity and bulk density to predict PAWC and to convert electronic sensor output to meaningful soil water information (mm of available water).

c)  Modelling of soil water: Simulation of the water balance should be considered as an alternative to field based soil water monitoring. Considering the error surrounding in-field measurement and issues surrounding installation of sensing devices there is a reasonable argument that the modelling of the water balance, when initialised with accurate PAWC and daily climate information, is likely to be as accurate as direct measurement. APSIM and Yield Prophet successfully predict soil water and should be considered for both fallow and cropping situations. CliMate is a logical choice for managing fallow water (Freebairn, 2012).

So what is the farmer/consultant to do on shrink/swell soils?

Consider the simulation of soil water as a viable option when PAWC and reliable met data are available. If a field based technology is preferred, identify as many monitoring sites as can be efficiently managed (and afforded) within a particular soil type or field, selecting locations that represent the productivity of the zone (through observation or yield mapping). Identify a technology that suits your general approach and provides a level of information that suits your needs. At each site (the number will depend on the technology), monitor longitudinal changes in soil water availability and make decisions based on the information, knowing that there may be significant error around the results.

diagram showing Soil Water Content (mm/mm) or Daily Rainfall (mm) over time

Figure 1: Soil water content (mm/mm) monitored using Capacitance (dotted line), TDR (solid line), Gravimetric (solid point) and NMM (open point), and rainfall (mm) for the period March 2011-Jan 2012. Estimates (mm/mm) of saturation (SAT), drained upper limit (DUL) and crop lower limit (LL) are provided.

 

diagram showing Soil Water Content or Daily Rainfall (mm) over time

Figure 2: Soil water content (mm) modelled using APSIM (solid line) compared to gravimetric (solid point) and EM38 (triangle and square) monitoring and rainfall (mm) for the period April 10-Jan 2012.

References

L. Burk and N. Dalgliesh (2012). Soil Water Express – a system to generate approximate soil water characterisations and current soil water estimates from minimal input data. "Capturing Opportunities and Overcoming Obstacles in Australian Agronomy". Edited by I. Yunusa. Proceedings of 16th Australian Agronomy Conference 2012, 14-18 October 2012, Armidale, NSW.

N. Huth, G. Boulton, N. Dalgliesh, B. Cocks and P. Poulton (2012). Electromagnetic Induction methods for monitoring soil water in irrigated cropping systems. Edited by I. Yunusa. "Capturing Opportunities and Overcoming Obstacles in Australian Agronomy". Proceedings of 16th Australian Agronomy Conference 2012, 14-18 October 2012, Armidale, NSW.

P. Poulton and N. Huth (2012). Improved monitoring of soil water resources to benefit crop management decisions. "Capturing Opportunities and Overcoming Obstacles in Australian Agronomy". Edited by I. Yunusa. Proceedings of 16th Australian Agronomy Conference 2012, 14-18 October 2012, Armidale, NSW.

D. Freebairn and D. McClymont (2012). CliMate-a smart phone App for analysing climate data. "Capturing Opportunities and Overcoming Obstacles in Australian Agronomy". Edited by I. Yunusa. Proceedings of 16th Australian Agronomy Conference 2012, 14-18 October 2012, Armidale, NSW.

I.C. Paltineau and J.L. Starr (1997). Real-time soil water dynamics using multisensory capacitance probes: laboratory calibration. Soil Sci. Soc. Am. J. 61(6): 1576–1585.

Contact details

Neal Dalgliesh and Neil Huth, CSIRO Ecosystem Sciences
203 Tor St, Toowoomba
Ph: 07 46881200
Fax: 07 46881193
Email: Neal.Dalgliesh@csiro.au; Neil.Huth@csiro.au

Reviewed by

Dr Jeremy Whish
CSIRO, Toowoomba