[GRASS-user] Estimating Albedo from Landsat
nikos.alexandris at felis.uni-freiburg.de
Wed Sep 15 20:45:19 EDT 2010
> > > # so, here it goes... seems to run fine
> > > i.landsat.toar band_prefix=landsat_postfire_east_sterea_ellas
> > > method=uncorrected sensor=5 date=2007-09-05 solar_elevation=52.4986238
> > > product_date=2007-09-05
> > ...
> > > r.info landsat_postfire_east_sterea_ellas.toar.6 -r
> > > min=203.36601783578
> > > max=319.083636050686
> > >
> > > No way the min value is true ( 203 K = -70.15 C -- irrational for
> > > Greece - for the moment! -)
> > >
> > > The max value (319 K = 45.85 C) seems to fit for Greece but not for
> > > September! Maybe for July/August.
> > looking at the image/histogram again those min and max are outliers I
> > guess. The main body of information starts at ~280 K (6.85 C) and stops
> > at ~307 K (33.85 C).
> > I am not sure how to explain the min < 270 K (those values lie mostly
> > within water bodies), but the max > 310 K areas are burned surfaces (a
> > few days old burn scars) which will explain the 45 C (at surface) in a
> > September day in Greece.
> > So, those values seem rational after all.
> And, of course, an even better look reveals that those mins are the clouds
> # acca
> i.landsat.acca -5 -2 -s band_prefix=landsat_postfire_east_sterea_ellas.toar
> Pass one processing ...
> PRELIMINARY SCENE ANALYSIS
> Desert index: 0.975
> Snow cover : 0.001 %
> Cloud cover : 12.856 %
> Temperature of all clouds
> Maximum: 299.82 K
> Mean : 251.23 K
> Minimum: 203.37 K
> HISTOGRAM CLOUD SIGNATURE
> Histogram classes: 100
> Mean temperature: 254.32 K
> Standard deviation: 15.99
> Skewness: 3298.36
> 97.50 percentile: 292.86 K
> 83.50 percentile: 276.87 K
> Pass two processing ...
> Finds clouds, shadows but includes too many irrelevant pixels -- bad result
> here :-(
> # withous the "-s"
> i.landsat.acca -5 -2 band_prefix=landsat_postfire_east_sterea_ellas.toar
> output=landsat_postfire_east_sterea_ellas.toar.acca --o
> Superb! This detects clouds (not the shadows though) very precise I dare to
> say :D
> # with "-f"
> not sure about the real differences with the previous command. Look fine
> though :-)
OK, there is some salt and pepper but this can be removed by some filtering.
It can be considered acceptable (?). Maybe I should post a screenshot
(...later I'll try out the NC landsat5 image as well).
More information about the grass-user