The Auto-load match criteria dialog controls the behavior of the automatic profile loader. Usually, the default settings work fine. However, there are situations (see below) where it can be useful to override the way that profiles are selected.
The fields in the dialog correspond to the annotations in noise profiles. For each field, there is a Policy and an Override value:
The Policy controls whether and how a particular annotation participates in the matching process.
The Override value, if set, will replace the corresponding value in the image.
The Clear button clears all override values and resets all policies to "Soft match".
The Auto Fill button attempts to initialize override values from EXIF data of the current image. It only changes fields for which it finds corresponding EXIF data.
If you work with cropped images, then you might want to change the Megapixels matching criteria. By default, if the image dimensions are smaller than the profile's Megapixels annotation, the auto loader assumes that the image is a downsampled (reduced-resolution) image, not a crop. Since downsampling reduces noise, the auto loader might pick a profile for a lower ISO in order to compensate. To work with cropped images, set the Megapixels policy below to "Ignore", or set the Override value to the uncropped image size.
One example of when it might be useful to use Override values is if you sometimes shoot under conditions or with camera settings that require distinct sets of profiles that can't be distinguished using the image EXIF data. For instance, noise for RAW files might be different from the noise in medium-quality JPEGs. You could create profile sets for both RAW and JPEG image qualities. Then you can set the Policy for Quality to 'Exact match' and set the Override value to match the image quality for any particular shoot. This will force the auto loader to consider only those profiles with the specified image quality value.
For another example, some camera sensors produce noticeably more noise in hot weather than they do in cold weather. You could create profile sets for cold, normal, and hot temperatures, and use the Temperature annotation to distinguish the different profile sets. Then you can set the Policy and Override values for the Temperature field to restrict matching to the appropriate set of profiles for a collection of images.