Management Zones maps are generated based on the combination of Vegetation Index maps, through a process called 'cluster analysis'. It's also possible to include information from previous yield monitor maps if available.
This is a procedure relatively simple, automated and objective, that allows to compare the field production and variability in different times, in a single map. Remote sensing imagery is selected in key dates, when crops are vigorous, and Vegetation Index (NDVI) maps are generated on those dates. These layers are then normalized, and integrated in the management zone map.
Methodology - Cluster analysis
A cluster is a predefined area, for statistical analysis purposes. For instance a cluster could be the pixel of a satellite image, such as 30x30 mts for Landsat images, 15 x 15 mts for Aster images, or 10 x 10 mts for Spot images. Once the Cluster area is defined, a normalized NDVI map is generated, with a grid with the cluster size.
the following sample images and corresponding NDVI illustrate the process:
//insert images in the table
These images are integrated by analyzing each cluster, defining homogeneous production areas, and obtaining the management zones map.
Processing is carried out at the field level. The main reason is that in a given farm, fields may have at the same time different types of crops or uses. Thus, NDVIs from different fields cannot be compared. For this reason different images are selected for different fields.
As a conclusion, management zones maps should only be used to compare variability within fields, but not as a means of comparison between different fields.
Applications and Benefits
Management zones maps provide guidance to know the variability within a given field. Based on this knowledge, field visits can be carried out, making decisions regarding field management, management zone definition, selecting areas for tests, etc.
Using Management zones to define sampling points makes possible to review on site the different zones and the reasons behind the possible limitations.
This can be a starting point to define inputs and recommendations, selecting the type and quantity of inputs according to the management zones.