| Re: Irregular grid -> 2D binned [message #49408] |
Sun, 23 July 2006 17:54  |
![will[1] is currently offline](theme/default/images/xoffline.png.pagespeed.ic.XRkd1fkXye.png "will[1] is currently offline")
will[1]
Messages: 6
Registered: July 2006
|
Junior Member |
|
|
Wow, this actually worked much, much faster than griddata (and is more
understandable than the extremely verbose, but not very informative,
griddata documentation). Thank you!
Ed Hyer wrote:
> also note that JD smith has written a routine called HIST_ND (get it
> from David Fanning's website), which will do histograms in up to eight
> dimensions, and provide reverse indices.
>
> Your example is rather interesting, in that your initial data set is
> itself a histogram...
>
> Here's how to get with HIST_ND where you're going:
>
> h2=hist_nd(transpose([[data1],[data2]]),[bin1,bin2],min=[min 1,min2],max=[max1,max2],reverse_indices=ri)
> totals2=h2*0; initialize totals
> for i=0l,n_elements(h2)-1 do if(h2[i] ne 0) then
> totals2[i]=total(value[ri[ri[i]:(ri[i+1]-1)]])
>
> I've been doing this so much it's becoming muscle memory.
>
>
>
> will wrote:
>> Excellent, thank you! (and here I was going to do it all so manually).
>>
>> will
>>
>> jgc wrote:
>>> have a look at GRIDDATA and GRID_INPUT functions, this last with
>>> DUPLICATES=Avg
>>>
>>> J.
>>>
>>> will wrote:
>>>> I've sucked it up, I think I need a push in the right direction.
>>>> Imagine that I have a long (~6 million points) list of data that
>>>> includes three fields: latitude, longitude, abundance. The lats and
>>>> lons are all over the place (i.e. no regular grid) and I'd like to bin
>>>> them into an image. Additionally I'd like to average abundance of each
>>>> bin to be the color for the image.
>>>>
>>>> I can use the reverse indices keword in histogram. I can even use mean
>>>> pretty well. I can get a binned histogram using histo_2D. But I'm
>>>> having a hard time thinking about how to go from the reverse indices of
>>>> histogram to the histo_2D case which doesn't offer the same keyword.
>>>> The only way I can think of to do this is to:
>>>>
>>>> a) do a histogram of the latitudes (using RI)
>>>> b) do a histogram of the longitudes (using RI)
>>>> c) find the intersection of indices for each bin that I want
>>>> d) "flatten" the lat, lon, indices/abund cube with matrix
>>>> multiplication
>>>>
>>>> It's the "c" part I'm sketched out on, my brain's can't think around
>>>> anything but a very evil Loop.
>>>>
>>>> Has this really easy or been answered here before? (Array decimation
>>>> was the closest thing I found).
>>>>
>>>> Thanks in advance!
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
