Facing an unseen problem on ArcPro: Classified pixels being lost while using Con tool

[AdityaSharma1]

Facing an unseen problem on ArcPro: I have a land cover raster with 24 classes, made up of square pixels with pixel side length being 313 metres. I downloaded the attribute table of this raster dataset and noted total of 752,076 pixels. Then I went about re-working the 24 classes into 4 major classes by using the 'Con' tool. When I see the attribute table of this new raster dataset with 4 classes, the number of pixels add to 704,523. This means a difference of 47,553 classified pixels emerges between the 2 datasets. Could somebody please help me understand that how are classified pixels getting lost in the process of combining the classes into 4 major ones using the 'Con' tool? Thanks.

[DanPatterson]

Your con statement and classification would help to make sure that all conditions were covered

[AdityaSharma1]

Hi Dan.. sorry but I didn't get you. Are you asking me to share the con statement. In that case, I am attaching the 2 statements that I have exported from ArcPro. I am also attaching the PDF of the help sheet that accompanied the raster. The raster is too big to be attached here.

So I am following the steps of 'Con' implementation: -

1) Group all the Natural Land Cover classes together in the original raster and classify the grouping as Value = 1 in the new raster. In this process I combined 16 Land classes into 1 category.

2) In the raster acquired from Step 1, Group all the Human affected Land Cover together and classify it as Value = 2. In this process I combined 6 Land classes into 1 category.

3) Finally, the raster output from Step 2 had 4 classes (16 sub classes= 1, 6 sub classes = 2, two sub classes from the original raster).

Now when I sum the pixels of these 4 raster classes, they don't add up to the sum of original 24 sub classes as in the original dataset.

That's where I was wondering what could've gone wrong in a seemingly straightforward 'Con' geo-processing tool?

Thanks 

Aditya