• The best first look at the raw data is through a quick unmanipulated dot map. We like to lay the various data layers out on a table as a paper GIS.
  
  

• Beware of trying to draw too many conclusions or relationships between the layers. Often the most critical information, such as soil moisture, is not collected, making any computerized layer analysis hazardous. Also, be wary of data integrity. We have often noticed that raw data may change, and not in a uniform fashion, as it is moved from one software package to another or even between versions of the same package.
  
  

• Data analysis without the knowledge and direct involvement of the farm manager is a recipe for wrong conclusions. Remote "expert" analysis is fraught with error.

Many of our initial ideas have changed, for example:

• We thought we would be able to control and change things, but now believe the better approach is to understand the resource and better manage it. For example, improving profit in a field may be best accomplished by increasing inputs in the most productive areas and leaving out the least productive areas, resulting in increasing field variability rather than decreasing it.
  
  

• Instead of looking for major gains in one or two areas (which we have not seen) we are now focusing on marginal efficiencies in many areas.
  
  

• Rather than the "best" system for each crop or application, we look for a good system that is compatible with the rest of the system. Precision agriculture needs standardization. We have seen little genuine effort in this area.
  
  

• We are now very wary of black boxes that "just need a little calibration" before they will give useful information about crop growth or soil condition. Instead we put much more effort in trying to capture farmer knowledge of the field using aids such as aerial images and yield/profit maps to help spur the thought and ranking process. As is often said, farmers are stewards of the knowledge of the land.
  
  

• We have experimented with all sorts of imagery and get the best bang for the buck from a simple color image taken from a low flying plane. We appreciate high resolution photos, quick turnaround and without cloud shadows.
  

  
  

• We have moved away from any sort of systematic "grid" sampling strategy to a directed sampling and management zone concept. In doing so we can build a better field management system that enables growers to tailor many inputs to field conditions and expected outputs. To establish these zones we consider:

  •Cropping/management history
  •Texture
  •Topography and drainage
  •Yield & profit potential
  •Nutrient & pH status
  •Earthworks, fence lines
  •General field knowledge
  

We believe that farmer ownership and involvement in this process is essential. It is an arrogant service provider that thinks they know what is best for their client without involving them in the process.

Soil property analysis should be a part of the whole management system not as a stand alone product. Within these zones, set different yield goals & management practices if appropriate (like side dress applications to the sandy ridges or fungicide rate changes as the soil becomes heavier). We have seen some soil properties remain fairly stable such as phosphorus, while others, particularly mobile ones are less predictable.

We are continually reminded that in most years soil nutrient levels are not the yield determining factor, most of the time it is water. We suggest you focus primary interest on those obvious items causing the most variation.

We have moved away from looking for electro/mechanical sensors to help automate the system in favor of devising tools to help human sensors better manage inputs. Now we use electro/mechanical sensors mostly in repetitive use applications.

Finally, we have revised our time frame from: "this will be quick and easy" to "this is going to take a generation or more".

We have confidence in recommending the following:

1. Hire the best labor you can afford.
2. Gather as much farmer knowledge of the field as you can.
3. Use aerial images of bare soil and early season crop growth to help in making management decisions.

http://nespal.cpes.peachnet.edu/pa/home/main.asp?TargetDir=30&content=0&media=