| Re: Find Closest Coincident Measurements In Time And Space Between Two Data Sets [message #62923 is a reply to message #62909] |
Wed, 15 October 2008 18:32  |
Jeremy Bailin
Messages: 618
Registered: April 2008
|
Senior Member |
|
|
On Oct 13, 8:14 am, "|Rob|" <r08...@gmail.com> wrote:
> I was hoping that somebody could help and apply some magic to this
> problem.
>
> I have two sets of satellite measurement data and for dataset A want
> to find the value in dataset B that are closest to it in time and
> space.
>
> If that's easy enough to speed it it'd also be useful to find ALL
> values in databaset B that were within a certain spatial and temporal
> distance of each point in dataset A (for example, 5deg lat/lon and 2
> hours).
>
> As both datasets are quite large I have a feeling that this problem is
> quite similar to the one found herehttp://www.dfanning.com/code_tips/slowloops.html
> but I don't think I understand it quite well enough to solve and
> eliminate the FOR loops.
>
> My (working but slow) code is below.
>
> Any help would be great.
>
> Cheers
>
> ;datasetA structure contains lat, long, time and values
> ;datasetB has lat, long, time and value arrays
>
> count=0L
> final_lat=fltarr(100000)
> final_lon=fltarr(100000)
> final_datasetA=fltarr(100000)
> final_datasetB=fltarr(100000)
> final_time=strarr(100000)
> final_date=strarr(100000)
>
> ;value in hms of datasetB meas. either side of datasetA meas. time
> where time is still considered coincident
>
> time_margin=010000
>
> ;value in degrees of datasetB meas. either side of datasetA meas.
> location where location is still considered coincident
>
> space_margin=2. ;degrees lat/long
>
> print, ‘Starting to loop through dataset A to find coincident dataset
> B points'
> FOR i=0L, n_elements(datasetA.lats)-1 DO BEGIN
>
> ;find measurement in datasetB that is within time and space margin of
> datasetA
> coin=where(datasetB_hms GE datasetA.time[i]-time_margin AND
> datasetB_hms LE datasetA.time[i]+time_margin AND $
> datasetB_lat GE datasetA.lats[i]-space_margin AND
> datasetB_lat LE datasetA.lats[i]+space_margin AND $
> datasetB_lon GE datasetA.lons[i]-space_margin AND
> datasetB_lon LE datasetA.lons[i]+space_margin)
>
> ;account for no coincident matches between datasets
> if coin[0] NE -1 THEN BEGIN
> final_lat[count]=datasetA.lats[i]
> final_lon[count]=datasetA.lons[i]
> final_datasetA[count]=datasetA.value[i]
>
> ;store mean of datasetB values that are coincident to datasetA
> measurments
> final_datasetB[count]=mean(datasetB_values[coin])
> final_time[count]=datasetA.time[i]
> final_date[count]=datasetA.date
> count=count+1L
> END
> ENDFOR
> print, 'Finished looping through datasetA data to find coincident
> datasetB data'
While not the full solution, you might find WITHINSPHRAD useful:
http://astroconst.org/jbiu/jbiu-doc/astro/withinsphrad.html
within JBIU:
http://web.astroconst.org/jbiu
-Jeremy.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
