Time series

Setting and querying time stamps

Meta-data:

• r.timestamp - set and retrieve
• r.support - add further metadata

Raster map time series analysis

• r.series - module to analyse series of raster maps, considering the stack of input maps along the time axis

Common legends for many raster maps

The general idea is to:

• Calculate the overall min/max covering all maps with "r.info -r"
• Create a full-scale color table with "r.colors rules=" on the first map
• Copy that color table to other maps in the series with "r.colors rast="
• Draw legend to color table extent with "d.legend range=min,max"

Example using Pathfinder AVHRR SST data:

 MAP_PATTERN='*.s0451pfv50-sst*degC.q4'
 
 # find overall min/max
 ALL_MIN=999999
 ALL_MAX=-999999
 
 for MAP in `g.mlist rast pattern=$MAP_PATTERN` ; do
   eval `r.info -r $MAP`
   ALL_MAX=`echo "$max $ALL_MAX" | awk '{if ($1 > $2) print $1; else print $2}'`
   ALL_MIN=`echo "$min $ALL_MIN" | awk '{if ($1 < $2) print $1; else print $2}'`
 done
 
 echo "all_min=$ALL_MIN   all_max=$ALL_MAX"
 ONETHIRD=`echo $ALL_MIN $ALL_MAX | awk '{print ($2 - $1)/3.0 + $1}'`
 TWOTHIRD=`echo $ALL_MIN $ALL_MAX | awk '{print 2*($2 - $1)/3.0 + $1}'` 
 
 # create full-scale color table for first map
 FIRST_MAP=`g.mlist rast pattern=$MAP_PATTERN | head -n 1`
 
 r.colors $FIRST_MAP color=rules << EOF
  $ALL_MIN blue
  $ONETHIRD cyan
  $TWOTHIRD yellow
  $ALL_MAX red
 EOF
 
 # apply color table to all other maps 
 for MAP in `g.mlist rast pattern=$MAP_PATTERN` ; do
    r.colors $MAP rast=$FIRST_MAP
 done
 
 # draw
 for MAP in `g.mlist rast pattern=$MAP_PATTERN` ; do
    d.rast -o $MAP
    d.legend $MAP range=$ALL_MIN,$ALL_MAX
 done

• For ps.map use the colortable instruction's "range" parameter.

Animation

• see the Movies wiki page

Time series in GRASS GIS 7

• A huge set of new modules is available: http://grass.osgeo.org/grass70/manuals/temporal.html
• A presentation and support material is available in the attachment section of the lecture: http://www.geostat-course.org/Topic_Gebbert
• Older brainstorming: SQL database support for a raster map timeseries is discussed in the Time series development wiki page.