Sextractor【2】MAP_WEIGHT and HOW it works.

The parameters are only breifly mentioned in the manual of Sextractor. However, you should be careful if you really want to change the paramters.

(1) Choose a right type of WEIGHT_TYPE
MAP_RMS is the most safe way to use weight image. It is the map showing the rms (you can call it error or variance.). Sextractor will select pixels with value higher than threshold*rms as a source (the value I mentioned here is background subtracted and smoothed). The error are also calculated based on the counts of object and the rms value.
MAP_WEIGHT is not the same as define in Multidrizzle or somewhere else. It can be a flat map, a exposure map. The Sextractor will convert it to rms map whatever your input is. What is most important and often missed is that it will scale the weight image according to the rms measure from the background. I think it just multiplies the rms value which is show in the output, 78.5964 in the following output.
(M+D) Background: 0          RMS: 78.5964    / Threshold: 78.5964 
So the weight image is used to tell which part of the image is more important than other parts. The arbitrary value is meaningless. The program will scale it and only use the related values!
MAP_VAR In the manual, "The data are converted to variance units (by defi- nition variance ∝ 1/weight), and scaled as for MAP VAR" However,  I have not used it! 
BACKGROUND The program will created a rms map based on the background measure my the program. This method are really not good. If you want to do real science, you should not choose it.
None Then every part of the image will be treated as the same weight, which means the regions which are not observed and are set to value 0 will also be treated as other region. The errors of objects in the corner will also be under-estimated. Notice that even you have set the MAP_WEIGHT, the overall errors are also probably under-estimated, because the rms is calculate from the image and the information of read noises and dark noises are missed.

(2) Detect and measure with different weight map.
People usually know that in dual model, we can define the MAP_TYPE to the detection image and measurement image separately. We can also set this in one image run.
"

All the weighting options listed in §7.1 can be applied separately to detection and measurement images (§3), — even if some combinations may not always make sense. For instance, the following set of configuration lines:
WEIGHT_IMAGE rms.fits,weight.fits 
WEIGHT_TYPE MAP_RMS,MAP_WEIGHT
will load the FITS file rms.fits and use it as an RMS map for adjusting the detection threshold and CLEANing, while the weight.fits weight map will only be used for scaling the error estimates on measurements. This can be done in single- as well as in dual-image mode (§3). WEIGHT IMAGEs can be ignored for BACKGROUND WEIGHT TYPEs. 
"

(3) Created you own weight and rms image.
The details will come later, you can check my other blog and also give me an email.