17
Surface Ob Issues in RTMA/URMA Steve Levine Systems Research Group and NOAA/NWS/NCEP/EMC NCEP Production Suite Review 2015

Surface Ob Issues in RTMA/URMA

Embed Size (px)

DESCRIPTION

Real Time Mesoscale Analysis (RTMA) Gridded (NDFD resolution) hourly near real time surface fields for situational awareness, verification, ‘analysis of record’ Used for tuning National Blend, initial forecast grid population Use all available surface obs and near-surface satellite wind obs over water Mesonet (MADIS Integrated mesonet feed) METAR Surface Synoptic Surface Ship Buoy ASCAT winds Satellite winds over water and below 850 hPa Assimilate obs with background field (short-term forecast, RAP/NAM/HRRR) Un-Restricted mesoscale analysis (URMA) runs 6 hours later to pick up latent data (also includes min/maxT) Overlap, duplicate, and quality issues with and across all types!! Added near-surface satellite winds, and ASCAT winds

Citation preview

Page 1: Surface Ob Issues in RTMA/URMA

Surface Ob Issues in RTMA/URMA

Steve LevineSystems Research Group and NOAA/NWS/NCEP/EMC

NCEP Production Suite Review 2015

Page 2: Surface Ob Issues in RTMA/URMA

Real Time Mesoscale Analysis (RTMA)Gridded (NDFD resolution) hourly near real time surface fields for situational

awareness, verification, ‘analysis of record’Used for tuning National Blend, initial forecast grid population

Use all available surface obs and near-surface satellite wind obs over waterMesonet (MADIS Integrated mesonet feed)METARSurface SynopticSurface ShipBuoyASCAT windsSatellite winds over water and below 850 hPa

Assimilate obs with background field (short-term forecast, RAP/NAM/HRRR)Un-Restricted mesoscale analysis (URMA) runs 6 hours later to pick up latent

data (also includes min/maxT)

Overlap, duplicate, and quality issues with and across all types!!

Page 3: Surface Ob Issues in RTMA/URMA

GTS

NWSTG/TOC/

“Gateway”

NESDIS

MADIS

RadarROC

NCO/PMB

EMC/MMB

LDM

BUFRTanks

/dcom

Satellite ingest scripts“Katz” scripts/EMC

FTP

LDM

decodeGather data

tranjb

SMS

job

quer

ies

NES

DIS

ser

vers

fo

r new

dat

a

Inside NCEP

Outside NCEP

TNC

Dump Files

dum

pjb

PREPDATAPrevents

SYNDATACQCBUFRPROFCQCCQCVAD

PREPACQCACARSQC

PREPBUFR

Con

vent

iona

l da

ta

Analysis(GSI QC happens here)

Non-conventional Data

BUFR mnemonic tableParm cards

tranjb

Different tanks/dumps for different types - not cross checked

Data Processing for RTMA

*original slide from Dennis Keyser (EMC)

tcvitals

Global (not RTMA) Guess (perhaps relocated)

Station dictionaries

Page 4: Surface Ob Issues in RTMA/URMA

Observational (mostly mesonet) Data IssuesAnybody can set up a mesonet site with little training, knowledge, warning

CWOP, Weatherbug obs dominate the station countsLack of consistent siting, equipment standardsAlso many DOT, University sponsored mesonets (generally more reliable)

Many different mesonet providers, but we mostly treat them all the sameShould we?All mesonets weighted the same, less than METAR via GTS/TOC (obs weighted by type)

Sites MADIS collects overlap with other data sourcesMETAR/AWOS sites, NOS sites, some buoysDuplicate checking tough to do between ob types

What’s a bad/unrepresentative ob? On what scale? How do we find them?Who do we call when we find a bad ob?

You see obs we don’t (or don’t want to see)We see obs you don’tFor non-mesonets, station dictionaries must be updated (requires RFC)

Page 5: Surface Ob Issues in RTMA/URMA

Quality Control in RTMA/URMA right now• MADIS

– MADIS marks of reject (X) are flagged but monitored• Prepbufr

– SDMedit reject list for non-mesonet data (dynamic, updated instantly by/through SDM but requires manual intervention)

• GSI– ‘Gross error’ O-B check (relaxed by terrain, buddy check for temp)– Provider/subprovider accept lists and Station-based reject/accept lists for

mesonets (static, required RFC to change) (StationID based)– Dynamic reject list (O-B last 6 hours)– “Variational” QC (problems with 2+ obs/hour/station)

• Data could also ‘fall on floor’ during decoding or ingest– Wrong type, dictionary not updated, etc.

Page 6: Surface Ob Issues in RTMA/URMA

Data ‘Falling on the Floor’

• WFOs pulling their obs from private/individual providers on their own– e.g. We do NOT get data from Weather Underground– Obs are not rebroadcast in way NCEP can receive– Easiest solution: Put ob into CWOP (run out of MADIS)

• RTMA cutoff time is XX:30 UTC for XX UTC analysis– Generally occurs during mesonet decoding/tanking process– Many obs come in late all or some of the time– URMA largely solves this issue (cutoff time is XX+6:30 for XX UTC)

• Processing challenges/issues at provider, MesoWest or MADIS itself– Mesonets are ‘data of opportunity’

• Ob ‘occasionally’ missing– Used in hourly analyses when available, but left out of min/maxT

Page 7: Surface Ob Issues in RTMA/URMA

A Typical RTMA Query:

“Why doesn’t the RTMA/URMA match the value of my favorite ob? How do you expect us to use this when it doesn’t match our obs (like MatchObsAll)?” (sometimes: “Don’t you know we maintain this ASOS site?!”)

Usual Answers:● We didn’t get your ob (or it didn’t get here in time)● Your ob failed a quality control test● Your ob got lost somewhere or was mistakenly pulled at a duplicate● Your ob was counterweighted by other nearby obs and/or background field● Your ob is not perfect (even if it is an ASOS you maintain)

○ “Whaddya mean it’s not your favorite ob too?!”

Page 8: Surface Ob Issues in RTMA/URMA

RTMA over DC/Potomac River

All plots are from RTMA, 7/16/15, 05Z

Page 9: Surface Ob Issues in RTMA/URMA

You are here

ReaganNat’lAirport(KDCA)

Ft. Belvoir (KDAA)

Page 10: Surface Ob Issues in RTMA/URMA

Now with ALL obs

ReaganNat’lAirport(ASOS)

Page 11: Surface Ob Issues in RTMA/URMA

Background Field

BG does not show ob-ob differences here

Page 12: Surface Ob Issues in RTMA/URMA

Increments (Analysis - Background)

Increments between obs are ‘washed out’

1F increment intervals 8-10 km wide

Page 13: Surface Ob Issues in RTMA/URMA

Ob and analysis valuesSite ID Observed Background Analysis

KDCA 72.1 69.9 69.0

AS102 72.1 69.7 69.0

AXDRA 69.9 69.9 69.9

FTWSH 69.9 69.9 69.2

WSHNP 69.1 69.7 68.7

FTBMR 67.9 69.9 69.2

ALEXD 67.9 69.7 69.0

KDAA 70.6 69.9 69.5

Res

trict

ed/in

visi

ble

mes

onet

s!

Page 14: Surface Ob Issues in RTMA/URMA

Impact of ob missing from maxT, but included in hourly analyses

Page 15: Surface Ob Issues in RTMA/URMA

Future QC Possibilities• CLARUS QC

– Developed by NCAR/USDOT (see document)– To be sent out by MADIS starting mid 2016 (decoder upgrade?)

• Ob processing changes– SDMedit to allow mesonets (but manual effort still needed to change)– Time series QC - look at previous analyses?– Cross-tank or cross-type duplicate checking

• Buddy check from WRFDA (implemented by Jacob Carley)– Currently lets obs in for temperature that match buddies, ignores O-B– Expand usage to throw obs out?– Use for other variables?

• Mesonet metadata database– Funded by National Mesonet, includes sensor info where available– mySQL format (not immediately compatible with WCOSS)– Combine with O-B stats, generate QC lists (already happening some)

Page 16: Surface Ob Issues in RTMA/URMA

Where to submit feedback or ask questions(shameless self-promotion)

[email protected] - RTMA/URMA listserver (best option if seen in RTMA)

National Blend of Models VLab page/forum (email w/o @noaa.gov to login)

[email protected] - email handle for NBM VLab forum

[email protected] - Decoder/ingest group

[email protected] - Obs Processing/BUFR issues

Page 17: Surface Ob Issues in RTMA/URMA

Links with further information...

NCEP Data Processing (Keyser)

RTMA/URMA overview (focus on April 2015 upgrade - made for SOOs/field)

Overview of January 2016 upgrade (made for EMC Director)

RTMA/URMA and ob NBM viewer (email w/o @noaa.gov to login, click “Daily Analysis Review”)

Citizen Weather Observing Program

MADIS (see “Integrated Mesonet”)

MesoWest (large aggregator, more mesonet research from Univ of Utah)