r.quantile returns wrong values

[jarekj71]

For spearfish database: g.region res=30

r.info -r -s map=elevation.dem@PERMANENT min=1066 max=1840 nsres=30 ewres=30

r.quantile input=elevation.dem@PERMANENT percentiles=0.001,0.01,0.1,0.25,0.50,0.75,.90,0.99,0.999 bins=1000000 returns: 0:0.001000:1067.000000 1:0.010000:1070.000000 2:0.100000:1080.000000 3:0.250000:1086.000000 4:0.500000:1091.000000 5:0.750000:1095.000000 6:0.900000:1098.000000 7:0.990000:1099.000000 8:0.999000:1099.000000

While equivalent command in R:

library(spgrass6) readRAST6("elevation.dem", plugin=FALSE, useGDAL=TRUE)

summary(r@data$elevation.dem)

Min. 1st Qu. Median Mean 3rd Qu. Max. NA's 1066 1200 1316 1354 1488 1840 3294

quantile(na.omit(r@data$elevation.dem),c(0.001,0.01,0.1,0.25,0.50,0.75,.90,0.99,0.999))

returns:

0.1% 1% 10% 25% 50% 75% 90% 99%

1080.000 1099.000 1156.000 1200.000 1316.000 1488.000 1621.000 1745.000

99.9%

1815.684

[glynn]

Replying to jarekj71:

For spearfish database: g.region res=30

r.quantile input=elevation.dem@PERMANENT percentiles=0.001,0.01,0.1,0.25,0.50,0.75,.90,0.99,0.999 bins=1000000 returns:

The arguments to the percentiles= option are, well, percentiles, i.e. the values given are divided by 100 to get fractions. So 0.99 is 0.99% not 99%.

Using correct values gives results which essentially agree with R:

$ r.quantile input=elevation.dem@PERMANENT percentiles=0.1,1,10,25,50,75,90,99,99.9 bins=1000000
Computing histogram
Computing bins
Binning data
Sorting bins
Computing quantiles
0:0.100000:1080.000000
1:1.000000:1099.000000
2:10.000000:1156.000000
3:25.000000:1200.000000
4:50.000000:1316.000000
5:75.000000:1488.000000
6:90.000000:1621.000000
7:99.000000:1745.000000
8:99.900000:1816.000000