Re: [idvusers] Formula to compute anomalies

• To: idvusers@xxxxxxxxxxxxxxxx
• Subject: Re: [idvusers] Formula to compute anomalies
• From: Tom Whittaker <whittaker@xxxxxxxx>
• Date: Mon, 21 Jun 2010 13:47:14 -0500

I wanted to post this example (see below), which was the solution to
Joseph Maina's question as an example.  His problem was exacerbated by
the two grids having different domains -- not only in longitude, but
in latitude....one of them had finer resolution near the equator than
in other latitudes.

On Tue, May 11, 2010 at 4:03 AM, Joseph Maina <jmaina@xxxxxxx> wrote:
> Hi List, I intend to compute temperature anomalies of a long time series by
> taking the difference between each time step  in a monthly mean time series
> and the temperature anomaly (single grid). The output would be a time series
> of this difference (anomalies), which I would then compute a cumulative sum
> of only the positive anomalies, more like the current inbuilt sum over time
> steps formula but in this case only the positive anomalies would be
> summed..any help would be much appreciated. Thanks

He wanted to get 3 sets of data:  a grid of the "tos" values minus the
"mmm" (average) -- the "tos" values were in a time ordered grid, but
the averages were in a single time grid.

Secondly, he wanted to get a set of grids where the anomalies < 0 were
set to zero, and finally, a single grid which summed up the anomalies
> 0 over the time periods.

To accomplish this, I first created a common domain and then computed
the anomalies while resampling into this common domain.

def joseph(mmm, tos):

 # first, resample the mmm grid
 nd=makeDomain("(Longitude,Latitude)",0,360,361,-90,90,181)
 mmr = resample(mmm,nd)

 # now resample the tos data and compute the anomalies
 anon = tos.clone()
 n = len(anon)
 for i in xrange(n):
   anon[i] = resample(tos[i],nd) - mmr
 #return anon

 # find where anon < 0, and replace with 0.0
 for i in xrange(n):
   index = find(anon[i], "<", 0.0)
   anon[i] = replace(anon[i], index, 0.0)
 #return anon

 # create a single-time data and set values = 0
 count = replace(anon[0], 0.0)
 # now, add "1" for each non-zero entry in anon[i]
 for i in xrange(n):
   count = count + mask(anon[i],">",0.0)
 count = newUnit(count,"counts","count")
 return count

-- 
Tom Whittaker
University of Wisconsin-Madison
Space Science & Engineering Center (SSEC)
Cooperative Institute for Meteorological Satellite Studies (CIMSS)
1225 W. Dayton Street
Madison, WI  53706  USA
ph: +1 608 262 2759

• References:
  • [idvusers] Formula to compute anomalies
    • From: Joseph Maina