227_062  Classification Notes

Level 1 Classifiation is followed, in scheme identifed by team.

There are 5 classes
1- Forest
2- Regenerating Forest/Managed Forest/Degraded Forest
3- Agro-Pasture and Bare Soil
4- Water
5- No Information

Descriptions of each class are found in the LUA-IAM Eixo 2 Classification Scheme.

The following steps are followed to complete the Unserpvised classification for each image.

1)  Unsupervised ISODATA 100 run  (name is image ID_Uns100)
2) Classes are combined into above 5 classes (name is image ID_classl1).  This classes are combined and
file renamed (image ID_classl1_classcombine).
3) If there are no information classes these areas of the image are subset (name is image ID_classl1_noinfo#)
and a mask of no informations areas made (name is image ID_classl1_noinfo#_noinfomask)
4) Subset masks (from #3) are manipulated as necessary (clouds classed in other classes
recoded to no information class) and combined class image also manipulated to reduce error through
recoding or filling, (from #2) are mosaicked. The above images are placed in a unique folder titled Classification_Processing_Images
5) A 3x3 filter is run on (image ID_classl1_3x3majority).
6) The image is renamed and placed in its own folder titled Classified_Image (image ID_unsupclassl1) .This is ready for mosaicking with 
classified images of adjacent path_rows from the same date.An area column is added in Hectares of each class.
7)If there are no single cloud free images for one date, but there are two from that date with clouds in different locations
 both images are classified and then the final classified image is placed in a folder image path_row_year_unsupclassl1_mosaic