The Minimum Population Mean Difference parameter of Cayula-Cornillon front detection tools should accept values in degrees C, not unscaled integers

[jjr8]

Currently that parameter accepts values in the units of the unscaled integers, following Cayula's original Fortran code. But that code did not even expose this as a configurable parameter. It just hard-coded it to the value 3, corresponding to 0.45 deg C for the data that they were working with.

In MGET, we expose this as a configurable parameter. The user can configure it to higher values and force the algorithm to ignore "weak" fronts. But because the parameter must be specified in the units of the unscaled integers, it is confusing to use. For example, for the NODC 4km AVHRR Pathfinder, the user must specify a parameter value of 13.333333 when they want the mean temperature difference to be 1.0 deg C.

For the product-specific versions of this tool (available under the MGET Data Products toolset in ArcGIS), we know how to convert the unscaled integers to degrees C. Therefore, to make this parameter easier to use, we should accept values of it in degrees C and convert to integers internally (the algorithm itself operates on the integers).

We should also require the user to make a deliberate choice about what value to use, rather than picking a small value that "finds all fronts". And we should elevate the importance of this parameter, so the user can actually discover it. To achieve these last two aims, the parameter should be moved out of the Cayula-Cornillon Algorithm Parameters options group, moved to the front of the tools' lists of parameters, and not be given a default value.

[jjr8]

Implemented in r953, released in MGET 0.8a38.