[ncl-talk] Error with bin_sum
Alan Brammer
abrammer at albany.edu
Wed Nov 19 09:36:03 MST 2014
From a 2A25 perspective and this goes along with what Dennis said, know your data. Extensive Documentation here: http://www.eorc.jaxa.jp/TRMM/documents/PR_algorithm_product_information/pr_manual/PR_Instruction_Manual_V7_L1.pdf
Relevant info on page 118. Which details how to implement the scaling, what the extra dimension is. (It’s vertical so summing may not be what you what.)
And details about the two types of missing data values. They leave out the along track dimension btw.
For the archive in-case that link dies.
I’m in now way affiliated with this data, have just used it and had the documentation on hand.
rain
float32 rain[49][80];
REAL*4 rain(80,49)
Internally this quantity is stored as
int16 rain_scale[49][80] after multiplied (scaled) by 100.
rainfall rate in mm/h
49 elements in the 2-D array correspond to the angle bins and 80 elements
(first argument in FORTRAN convention) in the 2-D array correspond to the range
bins.
If the estimated Z-factor is below 0 dBZ, the rain rate is always set to 0.
If the input radar reflectivity factor Zm is below the noise level,
the corresponding rain estimate is set to 0. This procedure does not cause
any serious problem except when the measured Zm becomes smaller than the noise
level by rain attenuation. In such a case, even if some heavy rain exists near
the surface,the number in this variable is 0.
To know whether such low radar reflectivity factors are caused by large
attenuation or not, look at the forth bit of 'reliab' and the forth bit of
rainFlag.
80 range bins are filled with data from top to bottom in height.
The last element corresponds to the ellipsoid height, i.e., 0 m high above
the model ellipsoid (not the actual surface). The first element corresponds to
the radar resolution cell about 20 km above in slant range along the beam
from the footprint on the ellipsoid. The range resolution is 250 m.
If the radar data is missing, MISSING value of -99.99 is stored.
This situation may happen at range bins above 15 km high because JAXA only
guarantees the data collection below 15km.
(The highest edge of the radar's receiving window comes down to nearly 15km above
the sea level near the equator.)
The bin number of the lowest range bin that contains valid rain data is
(rangeBinNum[][6] - 1 in C) or (rangeBinNum(7,.) - 1 in FORTRAN).
Below this level, CLUTTER value of -88.88 is stored.
If the estimated rainfall rate exceeds 300 mm/h, it is reset to 300 mm/h.
--
Alan Brammer
From: Dennis Shea <shea at ucar.edu>
Reply: Dennis Shea <shea at ucar.edu>>
Date: November 18, 2014 at 21:02:13
To: Ipsita Putatunda <ipsita.putatunda at gmail.com>>
Cc: ncl-talk at ucar.edu Talk <ncl-talk at ucar.edu>>
Subject: Re: [ncl-talk] Error with bin_sum
[0] There is no need to convert the file to netCDF. NCL reads the HDF and
GRIB files the same as netCDF. Please look at the webinar at:
http://www.ncl.ucar.edu/Training/Webinars/NCL_File_IO/
Webinar #1: Introduction to NCL File Input/Output: "supported" formats
[1] The most important rule in data processing is "Look at your data"
I have one day of TRMM 2A25 files.
Looking at one file using ncl_filedump
%> ncl_filedump -v rain 2A25.20140829.95606.7.HDF | less
[SNIP]
dimensions:
nray = 49
nscan = 9239 // unlimited
ncell1 = 80
[SNIP]
float Latitude ( nscan, nray ) <=== 2D
units : degrees
hdf_name : Latitude
float Longitude ( nscan, nray ) <=== 2D
units : degrees
hdf_name : Longitude
[SNIP]
short rain ( nscan, nray, ncell1 ) <=== type 'short'; 3D
scale_factor : 100
scale_factor_err : 0
add_offset : 0
add_offset_err : 0
calibrated_nt : 22
units : mm/hr
hdf_name : rain
[SNIP ... the rain values]
0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 37
62 70 53 115 77 124 123 103 168 101
147 98 -8888 -8888 -8888 -8888 -8888 -8888 -8888 -8888
-9999 -9999 -9999 -9999 -9999 -9999 -9999 -9999 -9999 -9999
-9999 -9999 -9999 -9999 -9999 -9999 -9999 -9999 -9999 -9999
-9999 -9999 -9999 -9999 0 0 0 0 0 0
=============
=============
What do you see?
Latitude(nscan,nray), Longitude((nscan,nray)
rain ( nscan, nray, ncell1 )
Fatal:bin_sum: z must be the same size as zlat and zlon
fatal:["Execute.c":7743]:Execute: Error occurred at or near line 46 in file test2_rai
n.ncl
`
Obviously 'z' (rain) is size= nscan*nray*ncell1,
Lat/LOn are size nscan*nray
That is the reason for the error messge.
================
More problematical ... WHat are the -8888, -9999?
There is no information on the file indicating what these represent.
Really, this is a poorly created file.
Also, the 'scale_factor' is 100. There is no indication of how this should be used. well, values like 115 if they were multiplied by 100 would yield 11500 mm/hr. This is ridiculous. How about 115/100.0=1,15 mm/hr .... MUCH more realistic.
Again look at your data!!
Learn to use 'conform' and 'conform_dims'
http://www.ncl.ucar.edu/Document/Functions/Built-in/conform_dims.shtml
http://www.ncl.ucar.edu/Document/Functions/Built-in/conform.shtml
See attached files.
You will have to change for your data.
On Tue, Nov 18, 2014 at 1:17 AM, Ipsita Putatunda <ipsita.putatunda at gmail.com> wrote:
NCL users,
Am trying to get an output file with gridded data from TRMM 2A25 swath data, using the attached script (I have converted files from .HDF to .nc) . But am getting error using bin_sum.
fatal:bin_sum: z must be the same size as zlat and zlon
fatal:["Execute.c":7743]:Execute: Error occurred at or near line 46 in file test2_rain.ncl
Where at line 46 of my script the lon,lat are not equal to ndtooned(lat2d),ndtooned(lon2d). But am not sure how to handle this issue.
Any help will be appreciated.
Thanks,
Ipsita
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20141119/e8c55a45/attachment.html
More information about the ncl-talk
mailing list