Q2 Workshop

1
Q2 Workshop
Multi-sensor QPE Current Operational
Capabilities, Issues and Perspectives from ABRFC
view
William E. Lawrence Development and Operations
Hydrologist Arkansas-Red Basin River Forecast
Center Tulsa, OK
June 28, 2005
2
ABRFC QPE History

May 1992 HAS software delivered to ABRFC March
1993 Formal HAS operations begin August 1993
StageIII software operational, ABRFC switches to
using MAPX vs old MAP June 1994 Start of ABRFC
online archive of QPEs, largest in NWS June 1996
ABRFC investigates COE precipitation software,
and starts porting code for our use July 1996
ABRFC finishes P1 performs testing November
1996 ABRFC switches to using P1 almost
exclusively
3
Why P1?

The P1 code, using a simple scientific algorithm,
allowed ABRFC forecasters an easier faster method
to come up with reliable QPEs Outperformed
StageIII both in speed, ease of use, and quality
of product Did not have severe underestimation of
rainfall in cooler season months Had added
functionality such as easy ways to make snow,
remove AP, save changes, etc. Code and results
were forwarded to OHD
4
LMRFC StageIII Study -33 error

• (Stellman, 2000)

5
ABRFC Process1 Study 2 error
6
Satellite Data Do we use it?
ABRFC receives satellite data every
hour Experience has shown its better than
nothing, but often misplaced and amounts are
unreliable ABRFC forecaster may use when all else
fails, but this occurs only very rarely P3 has
ability to draw in polygon area to swap to
satellite data anywhere in basin Overall, we much
prefer radar/gage mix of data
7
Satellite Data Do we use it?
8
Satellite Data Do we use it?
9
Gage Data How do we use it?
Approximately 1200 automated gages in ABRFC
area At any given time, approximately 400 in bad
gage list Only ASOS gages are heated, and even
they have trouble with frozen precipitation Oklaho
ma mesonet data is extremely valuable, well-run
mesonet GOES data also valuable, but maintenance
sometimes lacks Schoolnet, Alert systems least
desirable as little maintenance is performed
after initial setup. Overall, gage data is
EXTREMELY important to ABRFC in coming up with
acceptable QPEs
10
Gage Data How do we use it?
ABRFC feels that human interaction is necessary
for sufficient QC of gage data. What appears
good at first look may turn out to be poor data,
ie under-reporting, reporting at wrong hour, etc.
Bad data isnt always obvious until you start
adding up many hours Gages constantly being moved
into and out of bad gage list Most important
HAS function at ABRFC is QCing hourly QPEs
11
Current Issues
Continue to have serious deficiencies/difficulties
in producing accurate QPEs in winter
events Continue to have serious deficiencies with
radar estimates in mountainous areas luckily we
have few areas like this Radar estimates of QPE
at ranges greater than 75 miles continue to be
problematic during cool season List of problems
almost exactly the same as described in ABRFC
memo to OHD in 1993
12
Current Perspective
ABRFC feels that well known issues should be
addressed before we dive into probabilistic QPEs.
If we cant even produce a good deterministic
QPE year round, then why are we attempting to
develop a probabilistic QPE?
13
Why not MPE?
MPE is great improvement over StageIII Output is
roughly equivalent to P3 Older versions were not
user friendly in frozen precipitation events, nor
did they remember changes made Still problems
with performance of MPE vs P3 Radar artifacts are
still part of MPE ABRFC continues to work with
OHD so that we may adapt using a national
software
14
MPE Artifact
15
MPE Artifact