Effect of removing seasonal oscillation caused by Temperature from the data

It is well known that many sensors used to measure water content are affected by temperature. This effect can express itself as changes through the day as the temperature rises and falls, but also as changes through the year as the average daily temperature rises and falls. The daily fluctuation of temperature usually affects soil temperature to depths of less than about 30 cm, so it is not observed here because the shallowest measurements are at 40 cm. The seasonal temperature oscillation, on the other hand, can affect soil temperatures to depths of 200 cm or more, and so is important in this project.

Obvious seasonal oscillations existed in the uncorrected water content measurements at most sites. They are most noticeable during periods when there is little change in water content, ie. in dry years, and at the deeper depths. The graph below shows water content data from the relatively dry period from January 2003 to June 2005 for Paddock 2 at Traralgon, before and after the correction was made for the seasonal oscillation. [NB. the 40 cm data is omitted for clarity].

A seasonal oscillation can be seen in the uncorrected data clearly at the 120 and 180 cm depths, and less so at the 80 cm depth, with the water content reaching a maximum in late summer, and a minimum in late winter. When the data is corrected, the water content at 120 and 180 cm appears not to have changed very much at all, and this is expected from the rainfall patterns and experience elsewhere. At 80 cm there is some evidence of wetting after the correction, most noticably in spring 2003, when quite good rains were received. It can be seen that this water was quickly extracted by the crop. Further gradual wetting occurred at 80 cm for the next 12 months, with the crop once again extracting that water before the end of the year. These patterns, particularly the accumulation and extraction of water at 80 cm in 2004, are largely obscured by the temperature effect before it is removed.

Removal of the seasonal oscillation caused by temperature also improves the interpretation of available water for the paddock, as shown in the graph below.

Prior to correction, available water in the first half of 2003 appeared to increase in January and early February despite there being no rain until 21 February [see note below], and reach a minimum at the end of June, clearly showing the seasonal temperature characteristics. After correction, available water is more or less steady during this period. Similar patterns can be seen in 2004 and 2005. In addition, the maximum amount of available water is larger after correction, by almost 20 mm. This is because available water peaks in winter, when the temperature effect acts to reduce the uncorrected water contents and therefore the available water. This is a dramatic demonstration of the magnitude of the effect of the temperature on water content measurements - up to +/- 20 mm of soil water storage - a 20 mm overestimate in summer and a 20 mm underestimate in winter.

Note: Despite being substantial, the rainfall on 21 February does not show up as a spike in the available soil water. This is attributed to the fact that (a) the rainfall was very intense and no doubt resulted in a lot of runoff, and (b) most of the rain would have been absorbed by the soil above the shallowest measurement depth at 40 cm.