Open access publications in the SLU publication database

Abstract

There has long been a need for a simple, objective method of soi1 texture classification as an alternative to the expensive and often needlessly accurate mechanical analysis. In this investigation the correlation between data on two different water retention parameters and clay content is examined. The water retention data used in this context are the moisture content at 50 % relative humidity and the water content at the permanent wilting point. The intention is to investigate the possibility of estimating clay content with sufficient accuracy to enable soi1 texture to be classified from water retention data. The first part of the report consists of a 1iterature review, in which the most important factors affecting the hygroscopic water and the water content at the permanent wilting point are discussed. The second part contains an account of the investigation and of the statistical relationships obtained. The investigation was carried out on 123 soil samples of different textural classes, representing the different types of arable land in Sweden. The soi1 samples were analysed in the folloving respects; soi1 texture (mechanical analysis), content of organic matter (determined as organic carbon), hygroscopic coefficient at 50 relative humidity, loss on ignition and water content at permanent wilting point. The results of the analyses were then treated statistica11y. Multiple linear regression analyses were used to compute the statistical relationships between different variables. The relationships examined were: 1. The relationship between hygroscopic moisture, c1ay content and loss on ignition (Glf). Equation (3). 2. The relationship between water content at the permanent wilting point (wt150), clay content and loss on ignition. Equation (7). 3. Clay content as a function of hygroscopic moisture and loss on ignition. Equation (4). 4. Clay content as a function of wt150 and loss on ignition. Equation (8). Clay contents for the 123 soi1 samples were then calculated with the equations obtained and compared with the clay contents given by mechanical analysis. The equations were compared with each other in respect of determination coefficient (R2) and the residua1 variance obtained. The outcome of this comparison shows tnat equation (8), where the clay content is calculated from wt150, gives a better correlation between the calculated clay contents and the clay contents given by mechanical analysis than that given by equation (4), where the clay content is calculated from hygroscopic moisture content. Equation (8)is given below; L = -2,5 + 2,8 wt150 - l,7 Glf: Glf = loss on ignition In order to test the accuracy of the resu1ts, the clay contents for a collection of soil samples not used earlier in the investigation were ca1culated using equation (8). These values were then compaired with the clay contents given by mechanical analyses. The results show that although most of the estimated clay contents roughly coincide with the clay contents determined by mechanical analysis, some of the contents calculated diverge widely from those determined. The divergence is most extreme for soil samples with a large content of organic matter and especial1y if the percentage of organic matter exceeds the clay content. The reasens for this are discussed. In order to improve the calculation of clay contents some new equations were computed. One of these, equation (12), was obtained from 159 observations and restricted to apply only to mineral soils with a loss on ignition of less than 15 %. Equation (12) is given below: L =-0,4 + 2,9 wt150 - 1,2 Glf: To obtain an estimation of how well the clay content of an individual soil sample can be calculated with this equation, a prediction interval was computed. The prediction interval shows that the calculated clay content for an individual observation chosen at random, with 95 %probability will deviate less than 6,5 per cent from the value given by mechanical analysis. This error might be acceptable for soil samples with high clay content, makes the estimation of c1ay content very unreliab1e at low leve1s. Equations were also obtained, whereby the content of organic matter can be calculated with respect to water retention data.