From: gpetty@rain.atms.purdue.edu (Grant W. Petty)
Newsgroups: sci.geo.meteorology
Subject: Re: filtering radiosonde data
Date: 6 Mar 1998 23:39:38 GMT

In article <Pine.SOL.3.96.980306115202.8543C-100000@sol.sun.csd.unb.ca>,
Paul Collins  <k4eo@unb.ca> wrote:

>I was wondering if there is any standard rejection criteria or algorithms
>to scan radiosonde soundings.

Unfortunately, none that I know of (except one, see below).  I have
developed some of my own, but they're probably not useful for your
particular problem, since errors in upper levels of the sounding don't
affect my application as much.

> Is it possible to scan the data based on
>limits of temperature change with height? What is the theoretical maximum
>and minimum temperature change with height? Any help would be gratefully
>appreciated.

Yes: one test that is fairly foolproof is to look for superadiabatic
lapse rates; i.e., decreases in temperature exceeding 9.8 deg. C per
km ascent.  While these can occur over shallow layers in valid soundings,
especially in the first above the surface or (briefly) at
cloud top (due to evaporation of dampness from the temperature
sensor), they are suspect at higher levels, especially if the
temperature drop is very sharp and/or occurs over a fairly deep layer.

It is pretty difficult to put an erroneous temperature spike in a
sounding at a single level without also creating a superadiabatic
lapse rate at the top or bottom of the spike.  Of course this test may
not catch more subtle errors, like temperature biases over a deep
layer.  In the latter case, one might be better off looking at
continuity in time for a single station.

	Grant




--
Grant W. Petty                        |Assoc. Prof.,  Atmospheric Science
Dept. of Earth & Atmospheric Sciences |Voice:  (765)-494-2544
Purdue University, 1397 CIVL Bldg.    |Fax:    (765)-496-1210
West Lafayette, IN 47907-1397, USA    |Email: gpetty@purdue.edu

From: gpetty@rain.atms.purdue.edu (Grant W. Petty)
Newsgroups: sci.geo.meteorology
Subject: Re: filtering radiosonde data
Date: 11 Mar 1998 15:20:01 GMT

In article <6e675u$r5h$2@gte1.gte.net>,
Charles Samuels  <arktika@alaska.net> wrote:
>Radiosonde data is some of the best data around.  The instruments are
>calibrated before each flight and sensors are usually stable.  There are
>rare cases where the pressure cell begins to leak and causes large
>errors in height.  These are usually caught and the data discarded.

Everything is relative, of course.  But as one who has worked with a
lot of radiosonde soundings from around the world (there are nearly
60,000 in the data base that I'm using right now for validating
satellite water vapor estimates), I have certainly had my share of QC
headaches with raobs.  Most seem to be operator-related rather than
instrument-related.  Some particular raob stations seem to produce a
much higher percentage of erroneous reports than others, especially
at some stations in non-industrial countries where equipment
and training may not be quite up to snuff.  Once I detect a station
that seems to submit a lot of erroneous reports, I generally stop
using their raobs at all.

And ship raobs are particularly troublesome. I found that some
shipboard observers chronically forget to use the correct sign on the
reported latitude and/or longitude, so you could never be sure which
hemisphere they were really in unless you tracked them individually
through your data base and caught them traversing the Sahara Desert
for example.  Another ship I followed never reported negative
longitudes, so every time it hit the date line -- apparently coming
from the west -- it appeared to "reflect" back toward the west again.
I never did figure out where that ship was really located, because I
never caught it in a clearly landlocked location.

Finally, it is well known that raob humidity observations are prone to
all kinds of systematic biases under various conditions.
Unfortunately, there doesn't seem to be any kind of a reliable
reference standard against which to compare radiosondes in order to
characterize and correct the errors more accurately.

	Grant

--
Grant W. Petty                        |Assoc. Prof.,  Atmospheric Science
Dept. of Earth & Atmospheric Sciences |Voice:  (765)-494-2544
Purdue University, 1397 CIVL Bldg.    |Fax:    (765)-496-1210
West Lafayette, IN 47907-1397, USA    |Email: gpetty@purdue.edu