Unidata software and data usage at University of Wisconsin - Madison

Unidata software and data usage at University of Wisconsin -

Madison

Pete PokrandtUW-AOS Computer Systems Admin

Unidata software and data usage at UW-AOS

Evolution of UW-Madison AOS involvement with Unidata

Ongoing research using Unidata software/data Use in courses

Evolution of Unidata involvement at UW Madison

1986 – DIFAX to facsimile machine DDS, PPS to line feed printer

1987 – PC McIDAS 1989 – DIFAX to Dot Matrix printer 1992 – DDS, PPS to Sun Workstation

minimal data archiving to Exabyte tape wxp to plot data DIFAX to laserprinter


1994-1995 – GEMPAK installed, replaced McIDAS as primary data analysis/plotting tool

1995 – switch from satellite feed to IDD DDPLUS, IDS, HDS, MCIDAS, NLDN 1996 – archive DDPLUS, IDS, HDS, MCIDAS 1998 NMC2/SPARE/CONDUIT 2000 NEXRAD, FNEXRAD


2002 – archive CONDUIT grid analyses 2003 NIMAGE, CRAFT, IDV

Some uses of Unidata software/data

Products made available on the internet– Surface, Upper Air plots– NEXRAD Composites– Model plots and animations– Lightning strike plots (Restricted)

Analysis using NCEP Model Grids NCEP Model Grids used to initialize local

mesoscale models

http://www.aos.wisc.edu/weather

Products on the internet

Surface plots


Surface plots


Surface plots


Upper air analyses


Upper air analyses


NEXRAD products and composites– National and Regional Composites

(live link)– Individual site products for regional sites

C:\Documents and Settings\jhansen\Local Settings\Temp\mwcomp1.mon.fli

http://www.aos.wisc.edu/weatherdata/mapmaker_scr2/wxp/mwcomp1.fli

http://mapmaker.aos.wisc.edu/~wxp/nids/nids.html


Model plots and animations– Eta on the AWIPS 212 grid– Eta on the AWIPS 104 grid

– GFS on the 1 degree global grid300 hPa 500 hPa 850 hPa

http://www.aos.wisc.edu/weather/3indexcm.html

C:\Documents and Settings\jhansen\Local Settings\Temp\eta212\c500.fli

C:\Documents and Settings\jhansen\Local Settings\Temp\eta104\c850.fli

C:\Documents and Settings\jhansen\Local Settings\Temp\avnglob\c300.fli







GFS/Ensemble 4-panel plots



1 day forecast



8 day forecast



10 day forecast


Lightning data – plots and loops US region US region loop WI region WI region loop

C:\Documents and Settings\jhansen\Local Settings\Temp\latestus-nldn.gif

C:\Documents and Settings\jhansen\Local Settings\Temp\latest.ltg.fli

C:\Documents and Settings\jhansen\Local Settings\Temp\latestmsn-nldn.gif

C:\Documents and Settings\jhansen\Local Settings\Temp\latest.ltg-msn.fli

Use of NCEP Model Grids

Analysis using NCEP Model Grids

- Steve Decker – GFS Energetics plots

- Justin Mclay – Ensemble Verification

- Allison Hoggarth – PV tracking of easterly waves

GFS Energetics plotsSteven Decker

Horizontal Kinetic Energy per unit mass (KE) at a point can be broken into two parts

- Mean KE is derived from the time mean wind at that point – 28 day time mean

- Eddy KE is derived from current wind minus mean wind: EKE = (1/2)(u’2 + v’2)


Time tendency of EKE is determined by:

d(EKE)/dt = MAEKE + EAEKE + BTG + BCG + AGFC + CURV + RES (d/dt is local derivative)

MAEKE is mean advection of EKEEAEKE is eddy advection of EKEBTG is barotropic generationBCG is baroclinic generation


Time tendency of EKE is determined by:


AGFC is ageostrophic geopotential flux conv.CURV are terms related to earth curvatureRES is a residual, including friction



Advection terms move EKE around but do not create or destroy it



Generation terms create or destroy EKE in various ways



AGFC indicates collection (dispersion) of EKE radiation at (from) a point from (to) elsewhere in the domain



The other terms are usually not important


Using GEMPAK and the 1 degree global GFS data set from the CONDUIT data stream, plots are created twice daily for EKE with AGF vectors, EAEKE, BCG, AGFC and a wave packet envelope function.


300 hPa Geo Hgt EKE and AGF vectors


Time tendency of EKE due to eddy advection


Baroclinic Generation of EKE


Wave Packet Envelope function


Plots and further explanation available at

http://speedy.aos.wisc.edu/~sgdecker/realtime/realtime.html














Ensemble prediction of CAOsJustin Mclay

Daily 00 UTC ensemble initialization is being used in an ongoing assessment of deterministic and ensemble prediction of North American Cold Air Outbreaks (CAOs)

Ensemble forecasts frequently predict “Phantom” or “Sneak” CAOs (Postel 2002, personal communication)


Phantom CAOs – where ensemble suggest a high likelyhood of a CAO, which ultimately does not verify

Sneak CAOs – where ensemble suggests a low, if any likelyhood of a CAO, which ultimately does verify


Current effort is using GFS ensemble forecasts via the CONDUIT data stream to document the performance of the ensemble system with specific regard to CAOs.


Some elements – Relative frequency of Phantom and Sneak CAOs– Relative skill in predicting moderate vs. extreme

CAO– First and second statistical moments of the

ensemble (mean and covariance) are also being investigated for incorporation into statistical post-processing schemes to improve ensemble prediction of CAOs.

PV Tracking of easterly wavesAllison Hoggarth

Using 1 degree global GFS analyses and GEMPAK, evaluate PV (and other quantities) over the tropical Atlantic basin

Is there a way to categorize whether a wave will transform into a tropical depression or not?

Tropical depression #2 (June 2003)

C:\Documents and Settings\jhansen\Local Settings\Temp\june-305-315.fli

Use of NCEP Model Grids

Initialization for local operational mesoscale modeling

- Tripoli – UW-NMS

- Morgan/Kleist – MM5/Adjoint derived forecast sensitivities

Operational UW-NMSTripoli, Pokrandt, Adams, et. al.

Began operational runs in 1992 Data from inside source at NMC, later from

public NMC server Since 2000, via CONDUIT feed – locally

available sooner than via ftp “Storm of the Century”, 1993 Mainly lake breeze, lake effect snow – tied to

the terrain/surface characteristics


Cooperation with NWS-Sullivan, studying predictability of local terrain/topo driven phenomena (lake breeze, lake effect snow)

Fire Weather index prediction Supercell Index – supports severe storm

observation class (Storm chasing) Vis5d animations, GEMPAK output support

synoptic lab courses


Support of various field projects

- Lake ICE (Lake Effect Snow over Lake Michigan

- Recent Pacific field project – instrument testing – needed heavy precipitation over water

MM5/Adjoint derived fcst sensitivityMorgan/Kleist

MM5 Adjoint Modeling System (Zou et al., 1997)

All sensitivities to be described were calculated by integrating the adjoint model “backwards” using dry dynamics, about a moist basic state generated by the forward MM5 run, initialized with Eta initialization

'inx

)( vqpTwvux ,',,,,

'outx

inxR

outx R

ForecastModel

AdjointModel

)('R,R,R,R,RRpTwvux



Real-Time Forecast Sensitivities Goal: To understand the characteristics and sensitivity to initial

conditions of short range numerical weather prediction (NWP) forecasts and forecast errors over the continental United States

Available: – Sensitivity plots (updated twice daily) for two response functions:

36 hour energy-weighted forecast error 36 hour forecast of average temperature over Wisconsin

– Adjoint-derived ensemble of forecasts of average temperature over Wisconsin (soon to be available)

0h 12h

24h 36h


Sensitivity Based “Ensembles” Could run several forward models with different

initial conditions (Eta, NGM, GFS, NOGAPS,etc), get an ensemble of average temps over WI box

Instead, multiply the sensitivity gradient by each initial condition to get estimates of the ensemble members

Use in after-the-fact analysis

Use of archived datasets for after-the-fact modeling and analysis- Hitchman/Buker – UW-NMS/middle atmosphere modeling

- Martin – GEMPAK libraries to create new datasets

Middle Atmosphere modelingMarcus Buker, Matt Hitchman

Real-time forecasting for flight planning for various field projects (POLARIS, SOLVE, TRACE-P)

After-the-fact simulations to interpret observations


POLARIS (Photochemical Ozone Loss in the Arctic Region In Summer)– Regional scale simulations were run for the

campaign area (50-70N, 120W-70E)– Ozone & passive tracers initialized to monitor

constituent transport across the tropopause– Found ozone is lost from the stratosphere to the

troposphere by stretching/folding of tropopause by breaking Rossby waves.


SOLVE (SAGE III Ozone Loss and Validation Experiment)– Ozone loss in wintertime boreal polar region is

highly dependent on existence of polar stratospheric clouds – chemical makeup is conduscive for photochemical destruction of ozone.

– Form in coldest parts of stratosphere (~-80C), in areas where bouyancy waves induce relatively strong vertical motion


SOLVE (SAGE III Ozone Loss and Validation Experiment)– Mountain waves are a major contributor to this type

of phenomenon– Hitchman et al. (2003) used UWNMS to show that

non-orographic bouyancy waves can also produce extensive areas of PSC formation, especially in early winter


TRACE-P (TRansport And Chemical Evolution over the Pacific)

– UW-NMS simulations ongoing for flight dates in March, 2001.– Trying to differentiate between ozone from ground sources

and transport from the stratosphere, to determine contribution of tropospheric pollution from east Asian sector.

– Testing new methodology to get ozone flux between stratosphere/troposphere in regions of strong tropospheric activity

GEMPAK to create new data setsJon Martin

Use of GEMPAK libraries and locally written programs

Read existing data sets, perform calculations, save out to new data set.

Can be done recursively, or to trim size of a data set, compute complex functions, etc.

Unidata in UW Courses

GEMPAK/GARP – in class and in research ldm – to get data Maps online Tripoli – storm chasing Synoptic Lab – case studies

The future

IDV, THREDDS CRAFT

Questions?

Thank you!

Unidata software and data usage at University of Wisconsin - Madison

Documents

data difax

internetlightning data

local derivative maeke

data usage

internetsurface plotsproducts

panel plotsproducts

mean advection

local derivative agfc