KNMI measures soil temperature at four locations in the Netherlands. Measurements are carried out with standard electrical sensors (PT-500 clamped in stainless steel shaft) at five different depths. These depths are 5, 10, 20, 50 and 100 cm below soil surface level.
At regular time intervals the sensors need calibration. In 2001 at some locations the validity of calibration was out of date and at one location even some sensors did not function anymore. Replacement of these sensors was therefore necessary. However the usual method of installation needed improvement. With this method the sensors were laid down in a dug pit and covered by loose soil. This soil needed a long time of settlement before representative measurements could take place.
To avoid this problem a new way of installation was applied. In this method the sensors are placed in pre-drilled holes in one of the walls of a dug pit of sufficient depth. To assure the right depths a mould was constructed with small holes at distances from the top of the mould corresponding to the required depths of measurements (see appendix B). Together with the mould a long drill was constructed of which the diameter is slightly smaller than the shaft of the sensor. In the pit the mould is strict vertically hanged down with its flange on the surface. Through the holes in the mould holes are drilled in the side wall of the pit. After removal of the mould the sensors are placed in these pre-drilled holes. The slightly smaller diameter of the hole compared to the diameter of the sensor shaft assures close contact of the sensor to the soil. Moreover the soil surrounding the sensor is hardly disturbed.
Because of the presence of a second measuring site at Wilhelminadorp and De Bilt, the results of the measurements of the newly installed sensors could be compared. During the digging of the pit the soil type was described and classified according to the usual system in The Netherlands (De Bakker en Schelling, 1966).
In this report two methods are explained to assess the results of the measurements. The first method uses a simple model to describe the changes in soil temperature, if certain soil specific parameters (heat conductivity and heat storage coefficient) are known. Both parameters determine the so-called extinction depth, which in turn can be calculated from the amplitude or phase shift at the three top depths in soil temperature. During June 2001 daily values for the extinction coefficient for both soil temperature measuring sites were determined. The results showed only small daily variations at the site with the newly installed sensors. Also there was a high correspondence in the extinction depths calculated from the amplitude and the phase shift at this site. The extinction depth was calculated from values of the heat conductivity and heat storage coefficient depending on soil moisture content for several soil types found in literature. The calculated values for the extinction depth at the site with the newly installed sensors showed agreement with the calculated values from literature for comparable soil types.
The results calculated from the measurements at the old site showed less satisfactory results. Up to a depth of 50 cm the soil is of the same type on both sites at The Bilt. Assuming that the extinction depth at both sites is equal, the most probable measuring depths at the old site for the three upper sensors were calculated. These calculations showed deviations up 7 cm from the required depth. During the excavation of these sensors a year later the actual depths were measured. The correspondence between the actual and calculated depths appeared to be remarkable close.
The second method, called the δ-method, calculates a parameter (δ), which describes the correspondence between two sites on the same location. Calculations were carried out on measurements before and after the installation of new sensors at one site. On the same time the old sensors at the same site were excavated and the actual depths were measured. Especially the sensors at the three top depths showed large deviations from the required depths. The jump in the value of δ on the installation date is largely explained by those deviations. The variation of δ, being the same before and after installation, shows that a time for settlement of the soil is unnecessary and that representativety of the measurements is guaranteed from the day of installation onwards.
A.N. Mazee en R. Jilderda. Verbetering bodemtemperatuurmetingen
KNMI number: TR-280, Year: 2005, Pages: 23