proudly sponsored by
|
archiving of SOWACS proudly sponsored by |
|
>Cliff Hignett gave an interesting discussion on calibration in soils with >depth and location variable clay contents. Cliff, do you have any papers >out on this since the Greacen et al. book? What do you mean in your >recommendations when you say to "always pair tubes". I was lost at this >point. >-------- >Steve Evett srevett@ag.gov http://www.cprl.ars.usda.gov/ USDA-ARS, Reply from Cliff Hignett to Steve Evett No we have not published any more work in this line (apart from the reference I gave - which actually came after the book) This sort of work is not well supported financially and the reality of reasearch in the 1990's is ' no sponsor - no work' . I am working in other fields and Bill Greacen passed away last year. I have an unpublished paper using the techniques I am talking about which I can send you if you contact me privately. "always pair tubes" ? When a field calibration for a NMM is planned, it is necessary to decide how many access tubes will be needed for sacrificial sampling. ie how many tubes will be measured with the NMM and then destroyed by soil sampling to provide calibration points. The number of these tubes must be decided based on past experience, the site situation, plot layout, field variability etc. Half these will be sacrificed dry, the other half in the wet condition to get maximum spread of calibration points. In a conventional field calibration these tubes are randomly scattered around the field, with random selection for wet and dry sampling. Therefore, the error between any wet point and any dry point will include the field error. All I mean by 'pairing' is that you set out half the tubes randomly (or in any other pattern, decided on the basis of experimental design considerations and experience) THEN the other half of the tubes MUST be situated as close as possible to the existing tubes to form pairs of tubes One of the tubes in each pair must sampled in the wet condition and one in the dry. Data from each pair must be kept together. ie if you decide that 10 tubes are needed for the calibration, then select 5 sites and place two tubes at each site with the tubes of the pair as close together as possible without compromising the readings. I place mine 2m apart and put any excavations during the first tube sampling on the side furthest away from the second tube. The components of the soil causing field error are (in theory) randomly distributed across the field. But variation in soil texture (clay content) is the main contributor to error and (in most fields) the further apart you go, the larger the differences in the soil texture are likely to be. If the wet and dry point are measured in close proximity, then it is possible to get a 'mini calibration' for that specific site which is relatively free of field error. When you have several such 'mini calibrations' you may have an interesting problem if they are clearly different in slope or intercept. (as in my example data to trevor Finch where the slopes were the same, but the intercepts different) If they all define the same line - then my comments do not apply - your field is not variable in this way. I strongly recommend that you plot the example points I gave and do the regressions I suggest so you can see what I am talking about. First plot all the points and calculate the regression slope and intercept in the conventional manner as if these tubes were randomly distributed. Then draw the lines defined by the wet and dry points at each pair site - you will then have three separate mini calibrations, one for the site of each pair of tubes. You will see that the slope of the joint calibration (ie the conventional analysis) is quite different to the slope of the pairs of points defining the minicalibrations for each individual site. Bill Greacen and I nearly drove ourselves insane trying to think our way through this stuff, but in the end, the solution was quite simple. :- At any given state of dryness of the field, if the clay content varies across the field , then so will the water content (just because the clay will have higher wilting point water) . The water content of the field IS subject to this error. It IS real. There is NO way to avoid it. BUT the NMM is a non destructive sampling technique - you can go back to the SAME point many times. A single measuring point has the same clay content EVERY time it is measured, and the NMM can measure that point with a VERY high level of precision. ( NMM error is poisson distributed ie standard error is the square root of the number of counts - your precision is any level you want, just by counting for longer) The error comes when you want to average that value across a field. In particular, in defining a calibration for the field and when you want to get averages from your field data ussing it. As I said above, The water content of the field IS subject to this error. There IS no way to avoid it. If you want the water CONTENT of a field you MUST use a conventional calibration analysis and you must tolerate the error. But the water content CHANGE at any site in the field is NOT subject to the error caused by clay variation and can be measured with considerable precision by taking the change in count rate over time at each site and multiplying by the average of the slopes of the mini calibrations for the field. But to have this option, of course you must have the slopes of the mini calibrations. hence my advice to 'always use a paired tube calibration design' It costs no more than a conventional calibration but gives much more flexibility in analysing the data if you have the sort of field variability I am talking about. I am trying to make two points 1) It is possible to increase the precision of the NMM in measuring water content CHANGE in a field - FOR MINIMAL EFFORT. 2) In variable clay content soils the slope of a conventional calibration may be WRONG for calculating CHANGE in water content but okay for calculating water CONTENT itself. Cliff Hignett Cliff Hignett CPSS CPAg CSIRO Land and Water PMB 2 Glen Osmond South Australia 5064 ph (08)8303 8459 fx (08) 8303 8551 ah(08) 8276 7706 email cliff.hignett@adl.clw.csiro.au