proudly sponsored by
|
archiving of SOWACS proudly sponsored by |
|
NOTE: To get off this list, send email to majordomo@aqua.ccwr.ac.za with the body of the message containing the line: unsubscribe sowacs Dear Sowacs members, First, I thank a lot all comments from the participants in this forum about the differences in measurements on field and laboratory conditions on the relation between volumetric water content and pressure head. In addition, to go on with this very stimulating discussion. I have summarised some points that I have been listing to explain these differences: I - Soil variability It was discussed by Steve, and it is often found in the literature the natural soil variability of water holding capacity. Although it is the reality with we need deal with. Power estimation of appropriate sample number, geostatistics and specially the scaling technique have been improving the sampling schemes and the mathematical analysis of the data. Further to complicate, these properties are not only spatial variable but also time variable (i.g. Reichardt et al., 1993). I believe that until now has not agreement specialists about how to deal with the problem of the soil variability in a feasible way when the objective is to built a predictive model and not only a mathematical description of a specific situation. II) Differences in the volume measured from different methods and its calibration These factors, however in the most part of the further studied was not controlled, nowadays is possible to assessed the volume evaluated by TDR (Knight, 1992) and new techniques combining both sensor for water content and pressure head, will reduce this problem in the future. I believe that the key factor is the estimation of the REV, as also stressed by Parkes, for each specific soil and if it is assessed by a sensor it should be not a serious problem when working with not exactly the same volume. The effect of length of cup of tensiometer have been also showed be important not only to assesses small gradients into the soil (Luxmoore et al., 1992) but it is also play a role on the time of response (Hendrickx et al.1994). In our experiment we used tensiometer of about 6cm length and 2 cm diameter and our TDR probes had two transmission lines of 100 mm length, a diameter of 2 mm and a distance of 16 between the lines and they were vertically installed. Calibration of the instruments are an essential factor, in our experiment we developed a specific calibration for the TDR readings and check all tensiometers before they are installed in the field. We used water-mercury manometers to improve the precision of tensiometers readings. III) Specific problems with tensiometer readings a ) Time response and accuracy. The accuracy of tensiometer is about 0.2 kPa (at least for the good ones or with water-manometer gauge) and the pressure equilibration is 60 kPa/h. It can be specifically calculated using the equation to calculate the time of response proposed by Klute and Gardner (1962). That e may be used for some corrections in some specific question when the changes in the pressure head is very fast. In the IPM at lest for "our" soil changes in the pressure head, after the first hours, was very slowly to this problem be significant. Steve wrote that: " we know that the rapidity of equilibration will decrease as the soil dries and conductivities in the soil and the porous cup of the tensiometer decrease". Could you explain a bit more this Steve? Is it happens only when the conductivity is "soil limited"? b) Effect of temperature This effect was pointed by Cliff and Parkes. This factor probably did not played a big role in our measurements, because we insulated the tensiometer and the readings were carried out at standard conditions (early morning). In the standard normal laboratory the change in the temperatures was more easily controlled. Is a variation of maximal 5 C during the evaluation a measurable problem? IV - Hysteresis This was pointed by Steve and Parkes. Sure that hysteresis can play a "big role" in this evaluation. However, how as I wrote in my question, we had a good control of the boundary conditions (in the IPM method) which could circumvented this problem. However, repetition of IPM experiment at the same site has been showing that the time variability can play a role in some situations (see Reichardt et al., 1993). I would like to know if someone has more experience about the different hysteresis effects on the soil water curves on different soil classes. Probably this question was answered many years ago, and however, Kutilek and Nielsen, 1994 discussed the dependence of the specific water capacity on matric suction. I still don't know, how important is this effect or in which soils the hysteric behaviour should be more problematic? Especially when we are trying to simulate both situation of wetting and drying in a continuous approach using only the dry brunches. The hysteric approach is already an option (of course with some simplifications) for some models (e.g. the HYDRUS). If someone has more information or specific publications about the errors in mathematical calculation when hysteresis is neglected and I'll appreciated it. V) Problems arriving from sampling procedures Steve, Cliff and Dalton have pointed out this fact and I believe that ( i.e., the compaction of the sample) it is one explanation for some studies that show higher water content (at the same potential) when compared with field data (Luxmoore et al., 1991; Dane, 1980 and some layers of our data set). Changing, hammering for jacking the samples into the soil will be a partial solution for this problem. However, the appropriate height and diameter of the soil core is a motive of some controversies and conflicting criteria. Alternatively, someone has some ideas about the ideal soil core? Perhaps, work with clods (when it is possible, of course) will be a solution. However, clods have so many problems to be manipulated, and some differences between clods and field data arrived from samples from depth layers. VI) Air entrapment The difference in the air entrapment between field and laboratory conditions is probably one of the most important factor to explain these differences. Some comments about it are given below from Corey, (1992). In addition, the "fluffy" effect of soil cores in relation field conditions was pointed out by Bouma 1997. Does someone have experience with this problem and how to deal or control it ? VII) The errors arriving from the laboratory method itself I haven't found some criticism to methods of evaluation of the soil water retention curve in laboratory. Early, Hillel and Mottes, 1966 shows the effect of plate impedance on soil moisture retention. At this point, I have some doubts that perhaps someone that has more experience and tell us: Are the results from suction and tensions, (e.g. until 300 cm of water) similar? (I.e. are the results from sandbox and the pressure-plate similar?) Campbell, 1988 shows some results comparing desorption methods with pyschrometry: "Question have been raised about equilibration of samples at low water potential (< - 500 j/Kg). Madsen et al. (1986) show expected potential of samples equilibrated on a pressure plate to be about twice the water potentials of the samples measured with a thermocouple pyschrometer, indicating that, at low potencial, the samples never reached equilibrium. Lack of equilibrium is often the result of poor contact between the sample and the porous plate, but the low hydraulic conductivity of the soil itself can cause problems, even under ideal condititons. Time of equilibrium is a point that has been neglected in some studies and it is one of the conflicting criterions about the sample height? Moreover, the errors propagation from the use of disturbed samples evaluated at very low water potentials (15000 cm water) enhance the uncertainties of these data.. Further, the slow approach to equilibrium for samples in the dry range has reduced its utility (Gardner ,1988) Above I transcribed some comments give by Corey, (1992) about measurement at low and high tensions: "As well as theory, indicating that the relationship (between pore-size and soil water suctions) does not apply for water contents greater than about 85 % of the pore space. The reason is that the air is not free to enter all parts of the porous medium. Regions of the matrix containing some of the larger pores may be entirely isolated from the air phase surrounding water-filled pores". Further, he concludes that: It is probably better guess the pores-size distribution in the higher water content range than that provided by actual water data. Moreover, he also give some comments about the low tensions: "one would not expected capillaric models to apply where water retention is primarily absorbed films of water or between lattices of clay". I believe that these two statements should be carefully further discussed and perhaps we need to use different theories, instead the capillary one, to work with both low tensions and high tensions. Interesting results have recently been published about soil water retention at lower potentials (Logsdon et al., 1993 and others). Where positive air entry values were evaluated ! and It has been related not only in the laboratory conditions but also in field conditions (McCoy et al., 1994). We also observed this phenomenon in our determination in field conditions using tension infiltrometer. It shows that the capillary equation does not apply to soils with macropores, or at least it is not valid in the macropore range. The actual models to describe SWRC do not consider it and I believe that it could be in many situations it is the key factor to describe situations near saturation. Does someone have experienced similar situations ? I still have some provocative questions to this forum. Is worthwhile to simulate fluxes with determinist models using SWRC evaluated in laboratory when it does not represent field conditions? Does it makes sense to coupled SWRC, that nowadays can be very well fitted using different approaches and computer programs (see, e.g. Durner, 1994, Kosugi, 1998, RETC manual), with prediction of hydraulic properties based in statistical pore-size distributions models (e.g. Mualen's model) when the SWRC itself is not a reliable measurement of the pore-size distributions? Because of numerical problems, models of water fluxes simulations, normally need a the fucntion of the pressure head x volumetric water content in a range large than that evaluated by tensiometer. How could be combined different methods to amplify the range of measurements evaluated in field conditions (gypsum blocks, water marker sensors, psychrometers or equipotensiometers) ? If the principal transport processes occur in the wet range, could we simplest guess some values for the dry range ? Recognising the importance of the soil variability, should we abandon the use of deterministic models and use stochastic models for instance with scaling factors as parameter to takes in account the soil variability? Finally, and specifically responding to Clinton. We carried out our measurements, on a clayey Xanthic Ferralsol in Manaus - Brazil. This soil shows clearly bimodal pore-size distributions from inter and intraaggregate. It is a "unusual" soil with a hybrid character, showing properties of sand soils near the saturation (very high infiltration rates) but it holds water and unsaturated hydraulic conductivity values are typical for clay soil (in the dry range!). Best regards from Bavaria Wenceslau Teixeira _____________________________________________________________ Wenceslau Geraldes Teixeira Embrapa - AmazŁnia Ocidental - Brasil Universitt Bayreuth Bodenkunde - Abt. Bodenphysik D - 95440 - Bayreuth - Germany Fax: +49 (0)921 552246