[GRASS-user] t.register: register 4 consecutive rasters maps per month for long time series

[Johannes Radinger]

Hi Veronica,

thank you for your input regarding the registering maps with 0.25*months
time steps.
Here the solution I came up with was to import the maps in groups of 4
bands from the netCDF file. Then I calculate average over the 4 monthly map
and register the average map as a single map for the respective months:

# Register raster maps to assign correct date
scpdsi_maps_no = range(1,2688+1) # Total number of bands/maps
scpdsi_maps_groups_month = [scpdsi_maps_no[i:i + 4] for i in xrange(0,
len(scpdsi_maps_no), 4)]
scpdsi_maps_groups_monthyear = [scpdsi_maps_groups_month[i:i + 12] for i in
xrange(0, len(scpdsi_maps_groups_month), 12)]

year=1961 # First year
for idx, i in enumerate(scpdsi_maps_groups_monthyear):
month=1 # First month
for j in i:
print(str(year) + "-" + str(month).zfill(2) + "-01")
grass.run_command("r.import",
flags="o",
overwrite=True,
input="/path/to/my/scpdsi.nc",
band=j,
extent="region",
output="scPDSI")
scpdsi_maps = ["scPDSI."+str(x) for x in j]
grass.run_command("r.series",
overwrite=True,
input=scpdsi_maps,
output="scPDSI"+"-"+str(year)+"-"+str(month).zfill(2),
method="average")
grass.run_command("g.remove",
flags="f",
type="raster",
name=scpdsi_maps)
grass.run_command("t.register",
overwrite=True,
#flags="i",
type="raster",
input="scPDSI_monthly",
maps="scPDSI"+"-"+str(year)+"-"+str(month).zfill(2),
start=str(year) + "-" + str(month).zfill(2) + "-01",
increment="1 months")
month=month+1
year=year+1

This solution works fine for me.

/Johannes

On Tue, Oct 24, 2017 at 7:04 PM, Veronica Andreo <veroandreo at gmail.com>
wrote:

> Hello Johanes,
>
> there's and example in t.register manual page with netCDF files [0] that
> might help. You would have 48 maps per year (IMHO, 0.25*month is a bit
> strange as time step, but well). I would say that you need to write a
> script to get dates for each 1/4 of a month (maybe someone else know of a
> function to do a similar task, dunno) in your time series and then make a
> file with mapname|start_time|end_time to pass to t.register.
>
> However, if you only need to estimate yearly means and long term means,
> you can use r.series [1] with every 48 maps for yearly means and, the whole
> list of maps for the long term mean.
>
> hth,
> Vero
>
> [0] https://grass.osgeo.org/grass72/manuals/t.register.html
> [1] https://grass.osgeo.org/grass72/manuals/r.series.html
>
> 2017-10-24 11:56 GMT+02:00 Johannes Radinger <johannesradinger at gmail.com>:
>
>> Hi,
>> I am very new to the temporal functionalities of GRASS. Specifically I
>> have a netCDF file that contains many layers where each layer represents a
>> raster map of a drought index (http://monitordesequia.csic.es/map/) for
>> a timepoint of a timeseries. The time series is from 1/1961 to 12/2015 with
>> 4 maps for each month. So these are not really weekly maps but maps with an
>> interval of 0.25*months. The layers of the netCDF are only numbered
>> consecutively so there is no indication of a specific date etc. (I just now
>> that they start with 1/1961 and then the interval of 0.25*months). What
>> would be the procedure to register these raster maps in GRASS so that I can
>> calculate later only e.g. yearly means and a long-time year average?
>>
>> Best regards,
>> Johannes
>>
>> _______________________________________________
>> grass-user mailing list
>> grass-user at lists.osgeo.org
>> https://lists.osgeo.org/mailman/listinfo/grass-user
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-user/attachments/20171109/5b617e14/attachment.html>