Subject: Ladscape Patterns
Date: SUN JAN 17,1999 17:02:29 EST
posted by: [EMAIL PROTECTED] (MidNight Mapper)
Soil sampling via any method you identified remains a problem when stratified to NRCS soil mapping unit or management unit reveals as much variation within as between mapping units. The causes of this are many. Essentially nature’s soil chemistry has been so disrupted and historically mis-matched by man’s attentions that all that may remain are the grossest effects of soil formation – topography, parent materials, and conditions of erosion.
GPS located soil sampling is a method to explore “spatial” variation in soil fertility/chemistry. As before composite versus point soil sampling gives you different certainly of what you believe is there versus what is there. One odd core can skew any composite.
The use of diamond, staggered start, and random sampling methods are techniques to reduce sampling bias. Grids are simply a way to help visualize and “sell” what will be done. Grid size is an index of sampling intensity. How the sampling is done once intensity is agree to is another issue. It is point or composite? Is it random or patterned?
Continuos surface estimation from sparse data is problematic. Inverse distance is clever, easy, and tends to hold the value of the centroid sample point. Kreiging on the other hand needs continuos variation in spatial distances between samples to gain the best result.
Centroid sampling gives any surface engine stepped distance between samples in multiples of the grid size and their diagonals. A random pattern will provide many fold these point-to-point distances.
On-the-go soil sensing is a no-brainer. A problem remains in that they have many proxy effects – temperature, texture, pH, salts, and etc. Identical sensor readings at two points in the field may indicate a pH issue at one and a texture at the other. To fully rely on the on-the-go soil reading with out some companion point soil sampling is a foolish step in my mind.
FWIW
MidNight Mapper
Date: SUN JAN 17,1999 17:02:29 EST
posted by: [EMAIL PROTECTED] (MidNight Mapper)
Soil sampling via any method you identified remains a problem when stratified to NRCS soil mapping unit or management unit reveals as much variation within as between mapping units. The causes of this are many. Essentially nature’s soil chemistry has been so disrupted and historically mis-matched by man’s attentions that all that may remain are the grossest effects of soil formation – topography, parent materials, and conditions of erosion.
GPS located soil sampling is a method to explore “spatial” variation in soil fertility/chemistry. As before composite versus point soil sampling gives you different certainly of what you believe is there versus what is there. One odd core can skew any composite.
The use of diamond, staggered start, and random sampling methods are techniques to reduce sampling bias. Grids are simply a way to help visualize and “sell” what will be done. Grid size is an index of sampling intensity. How the sampling is done once intensity is agree to is another issue. It is point or composite? Is it random or patterned?
Continuos surface estimation from sparse data is problematic. Inverse distance is clever, easy, and tends to hold the value of the centroid sample point. Kreiging on the other hand needs continuos variation in spatial distances between samples to gain the best result.
Centroid sampling gives any surface engine stepped distance between samples in multiples of the grid size and their diagonals. A random pattern will provide many fold these point-to-point distances.
On-the-go soil sensing is a no-brainer. A problem remains in that they have many proxy effects – temperature, texture, pH, salts, and etc. Identical sensor readings at two points in the field may indicate a pH issue at one and a texture at the other. To fully rely on the on-the-go soil reading with out some companion point soil sampling is a foolish step in my mind.
FWIW
MidNight Mapper
