Pktools - Open Source Geospatial Tools: Applications in Earth Observation - page 191

-nodata|--nodata
No data value(s) in input or reference dataset are ignored.
-m|--mask
Use the first band of the specified file as a validity mask. Nodata values can be
set with the option msknodata.
-msknodata|--msknodata
Mask value(s) where image is invalid. Use negative value for valid data (example:
use -t -1: if only -1 is valid value) (default: 0).
-o|--output
Output dataset (optional).
-f|--f
OGR format for output vector (for vector reference datasets only) (default:
SQLite).
-lc|--lclass
Attribute name of the classified label (for vector reference datasets only) (default:
class).
-ct|--ct
Color table in ASCII format having 5 columns: id R G B ALFA (0: transparent,
255: solid).
-co|--co
Creation option for output file. Multiple options can be specified.

We will assess the accuracy of the land cover map created in Chap . 17 . The name of
the classified land cover map is LC82070232013_fmap_masked.tif .Itis
a forest, non-forest mapwith forest labeled as 1 and non-forest labeled as 2. The name
of the vector datasetwith reference labels is reference.sqlite (layer name is
“valid”). Pixels in the land cover map with value 0 are ignored for the validation
using the option -nodata 0 .

pkdiff -i LC82070232013_fmap_masked.tif -ref reference.sqlite
→ -ln valid -cm -nodata 0

0...10...20...30...40...50...60...70...80...90...100 - done.
12
1 1 3
2 30 821

class #samples userAcc prodAcc

1 54 58 (9.1) 76 (5.9)

2 851 98 (3.4) 96 (0.6)

Kappa: 0.63
Overall Accuracy: 95.3 (0.7)

The produced confusion matrix represents each reference sampling in C rows and
columns, with C the number of classes (2 in this case). The reference class labels
are shown in the first column. The first row corresponds to the predicted class labels
that have been read from the input raster dataset. A cell value at row j and column