steveMIfarmer - 1/7/2010 10:47 1. Calling point samples grid. Point sampling and grid sampling are one and the same. Just because I don't show the grid boundaries on the map doesn't mean that it wasn't gridded.
What you are doing is taking point samples not grid. In my mind a grid is a area you sample not a point. The grid is used to determine the location of the point sample. When using interpolation, it is very important to take your samples in the proper location.
Composite samples are better than point in my mind. You need to check your premises.There is often a huge difference between what "seems" right and what is right.
I would challenge you to do your own analysis. Take a few of your fields and do some basic analysis such as I did. Take your predetermined zones and split them up into smaller units. Pull samples from those smaller units and compare to the overall average of your original zone. I think you'll be surprised. You seem to be saying that the zone or grid is variable so why even try to come up with an average of the zone or grid and just pick a spot and go with it. I can pick zones out by yield that are similar and then take a composite sample. Is there variability in the zone, sure there is, but not as much as between the points on a point map that is choosen by random. First of all, with point/grid sampling, the sample locations are not chosen by random. And second, I have no doubt that you can pick zones by a yield map. But again I'll challenge you to look at the zones in more detail. I think you'll find that the level of fertility is much more variable than you thought.
2. Correlation of yield to P&K levels. To start out who knows where it is at but If you don't know whether or not yields correlate with P&K levels, than you are making a big mistake to start out with. This will lead to a very inaccurate fertility program. As an example; let's say that I want to develop a variable rate nitrogen program for my fields. Since I already have soil tests, I decide to base my nitrogen rates on the level of P and/or K within the field. According to my hypothetical way of thinking, if P&K levels are high, N levels should be high also. Would this type of program make sense? No it wouldn't, because P&K levels have no correlation to the amount of N needed in order to maximize yield. But in effect, this is what you are doing by letting yield dictate where you take your samples. There is often very little correlation, so why would you use yield as your determining factor? It doesn't make sense and it leads to a wide margin of error.
if you get it leveled out the P&K have a direct correlation IMO and many others. In my experience, you'll never get those areas evened out. LHaag referred to the early days of grid sampling. I remember those days well. One of the expectations when I first started with VRT was that we were going to even out the nutrient levels across the field. Boy was I disappointed when I started getting results back from retests after an intensive VRT program. The fields were not even! Imagine having to explain that to your customer! Today we know that there are many factors that affect nutrient levels. It's not as simple as knowing the soil type and the previous yield. It's much more complex, and it's a mistake to let these factors be the sole influence on your soil sampling and fertility program.
Removal rates are pretty well documented by many. I base almost everything off yield to make zones. Soil types show up on the yield maps. We do look at the soil type map as a reference but not really using it to make the zones. Again, I would urge you to check your premises. I think you'll be surprised at what you find.
| 
grid sampling ? again
