COSMIC retreat October 2005 UCAR COSMIC Project Office CDAAC Readiness Status

COSMIC retreat October 2005 UCAR COSMIC Project Office

CDAAC Readiness Status


Thanks CDAAC team!

• Doug Hunt - General architecture + lots more• Maggie Sleziak-Sallee - Operator interface• Sergey Sokolovskiy - Inversion algorithm• Bill Schreiner - Orbits, Clocks, fiducial network• Stig Syndergaard - Ionosphere• Martin Lohmann - Q/C + Errors• Karl Hudnut - Computers/facilities + System support• Jim Johnson - Fiducial data, BitGrabber network, Rx tests• TaeKwon Wee - 1DVar code• Lidia Cucurul - Data assimilation and encouragement to do better


Missing from CDAAC 1.0• TIP instrument processing:

• Level 1b radiance and pointing information file• Level 2 inversion (combined with RO)• Difficult to debug until we have real data

• COSMIC level 0 to 1 code• Open Loop capability• Dual cluster functionality• Operator interface to payload SOH data• TBB processing capability

Issues Reported at 2004 Retreat:







TIP State of Health

● TIP State of health extraction done: tipSOH.pl

●Generates tipSOH data type which contains:

●High voltage status plot●Filter heater status plot●Command and response summary●Based on IDL code from Scott Budzien of NRL


TIP Radiance code● TIP pointing location from attitude (part of tipRadiance.pl)

is complete.● Awaiting code from NRL for converting TIP raw counts to

radiances.







COSMIC zero to one code

● cosmicZero2one.pl complete●Tested with several sample data files from NSPO, including 9 dumps during the recent early orbit simulation.●Operates on either USN download files or NSPO raw satellite dump files●Generates:

●High rate opnGps GPS data●Low rate BINEX GPS data●TIP raw data●Satellite attitude data, including rough position and velocity and inertial quaternions●GOX state of health extract







Open loop zero to one code

● JPL open loop code was highly experimental●Written in perl and C●Used a 300+ Megabyte intermediate ASCII file●Slow—10 to 15 minutes to run

● Updated this code into one clean efficient C code and a perl wrapper

●More compact (80 Megabyte) binary output file●Much faster—1 minute or less to run●Consolidated JPL Bjfmtl.c code. Now one version works for COSMIC, CHAMP and SACC open loop

● New high rate data format: opnGps. Replaces atmGps BINEX file which was hard to adapt to new open loop requirements


Statistics of comparison of the RO inverted N to ECMWF analysis. global

- better penetration;- larger standard deviation

PLL OL


Statistics of comparison of the RO inverted N to ECMWF analysis. tropics

- better penetration;- larger standard deviation- smaller negative N-bias

PLL OL


OL RO signals capture PBL

OL PLL


OL RO signals capture PBL

A,B,C - sea

D - sea-land boundary

E,F - land







CDAAC Hardware Status

● All Hardware has been purchased● Primary processing cluster is in place and being tested● Primary I/O server is running and receiving NR CAN● Web based data available to collaborators

● Back up processing and I/O servers are in place● Failover procedures for I/O and processing strings due

mid December


CDAAC Operations Status

●Monitoring of I/O server and processing cluster performance in a rudimentary state.●Complete monitoring scripts, including dial out to pager/cell phone for on call operator due mid December


CDAAC Computer Room Status

● AC power and backup UPS in current room is adequate only for short (15 minute) outages

● Cooling is barely adequate and not redundant

● As a result of extensive analysis and survey of needs a new facility is being prepared.

● Backup generator will guarantee AC power for up to 2 day outage

● Adequate cooling with redundant systems will be available● UCAR maintenance is preparing the room and will have it

available by end of 2005● CDAAC clusters will move to new facility Jan 2006







Operator Interface - Leo Dumps - Added COSMICT (COSMIC test) mission and changed display

to show results for up to 6 LEO Ids instead of only one.

- Leo dump queries now show both: dumps created with links to

dump info, and dumps that were not created due to an error.


Operator Interface - Leo Dumps - A click on the dump ID will dynamically generate a status page with:• * Refractivity statistics for all occultations

• * Number of files produced of the following types:

• o atmPhs

• o atmPrf

• o avnPrf

• o wetPrf

• * Failure percentages and reasons


Operator Interface - Fiducial Information - The main screen allows display for the current day, with the option to view up

to ten days back. The display shows:

• * Average Latency Map for a given day for CosHrf and IgsHrf files (New)

• * Fiducial RealTime Data Inventory (New)

• * Hour count of CanHrf, CosHrf, and IgsHrf files for the present day (New)

• * MP1_AVG and MP2_AVG (GPS data quality indicators), EPOCHS_SUM (the total number of epochs for this day) and NOBS_SUM (the total number of observations for this day) for each site.


CDAAC - Internal - Besides The Operator Interface, the following sections have been updated for

the next release:

- Operator Manual: http://jay:8080/cdaac/internal/opsManual/index.html

- Source Code data flow diagrams: http://jay:8080/cdaac/internal/sourceDoc/index.html







TBB processing

● No progress


Other new CDAAC activities

● Improved fiducial network and data handling● Development of Data Bit Systems● Single difference processing greatly reduces processing

time and data volume● Receiver Testing● Data QC● Ionospheric processing






FIDUCIAL SITE STATUS●Real time processing

● ~30 stations in a real time stream from NR Canada (canHrf)●Three netRS receivers in South Pacific (cosHrf)●The balance of stations we use for real time (total = 50) are fetched from the IGS (CDDIS) based on latency and coverage●Script rateIgsHrf.pl determines which igsHrf sites to fetch. This list will be updated monthly.

will add “ntrip” sites






• Bit Grabber System• 10 systems built• Currently deployed• User guide on web• Data already proven

beneficial• Considering patent

for use of data bits in RO processing


Host agreed

Host needed

Global Bit Grabber network






CDAAC POD Status

• Near real-time POD demonstrated with CHAMP and current ground network data

• Single-Difference processing with 30-sec GPS clocks to be used for excess phase generation in Post-Processing and Near Real-Time

• Include CDAAC functionality to handle LEO S/C maneuvers• Include COSMIC satellite GPS antenna phase centers and possibly variations• Modify Bernese software to process data from 2 POD antennas • Install occultation prediction software

New Results:

Future Work:


CDAAC GPS Ground Networks

Near Real-Time

Post-Processing


CDAAC Zenith Troposphere Delay Results

Post-Processing

Near Real-Time

Note differentcolor scales


Post-Processed and NRT CHAMP Orbit Position Overlap Results (vs. JPL)

(Daily 24 hour arcs) (Arcs for every CHAMP dump)

Near Real-TimePost-Processing


Computation of excess atmospheric delay

• Double Difference • Advantage: Station clock errors

removed, satellite clock errors mostly removed (differential light time creates different transmit times), general and special relativistic effects removed

• Problem: Fid. site MP, atmos. noise, thermal noise

• Single Difference• LEO clock errors removed • use solved-for GPS clocks• Main advantage: Minimizes

double difference errorsQuickTime™ and a

TIFF (LZW) decompressorare needed to see this picture.


Spectral Analysis of GPS Clocks

• Analyzed spectrum of difference of 2 GPS clocks, with 1-sec data

• 30-sec fiducial sampling adequate for RO data processing

30-secPeriod


DD vs. SD Post-Processing Status(2005.121-151)

Single-Difference with30-sec GPS clocks

Double-Difference with1-sec ground data


DD vs. SD Near Real-Time Status(2005.140-150)

Single-Difference with30-sec GPS clocks

Double-Difference with1-sec ground data


Other new CDAAC activites




Receiver Testing (a)

•Planning to test all COSMIC flight receivers outdoors on roof of NSPO

•Working with BRE engineers and JPL on test plan and procedure

•Tests planned for 1st week of Nov. ‘05

•Will set up web site for on-line analysis of test data in real-time (to be used by test engineers)

•Tests contain some risk

•Interpretation / consequences of bad data

•Tight schedule to fix/change receivers


Receiver Testing (b) •Need to keep the tests simple !!

•Check SNRs and compare to SNRs from UCAR roof tests (and to commercial reference receiver)

•Check cycle slips - to see if receiver remains locked

•Need to test 4 antennas / satellite


NSPO Roof Pre-Survey

Satellite Visibility

L1 signal to noise ratio (dbhz)






CDAAC QC and error estimation progress since last year

• Roam QC has been fine-tuned and new QC checks have been introduced to remove errors related to:• Truncation of L1 when this signal is lost• Truncation of L2 when this signal is lost• Truncation of radio holographic bending angle profile • ‘Bad’ signals (tracking errors and 1 second spike (CHAMP)

• Dynamic error estimation for bending angles and refractivities below 20 km – errors estimated without using external data

• Both QC and error estimation techniques have not yet been tested with open loop data – this will be done in the coming future


Old processing vs. new processing - deviations from ECMWF refractivity

CHAMP August 2002


Global Statistics of dynamically estimated measurement errors CHAMP August 2002






Ionosphere processing

• 1) TIP:- running NRL’s radiance code and from the orbits and attitude info calculate the pointing location on the Earths surface.

• 2) Absolute TEC:- Cycle slip correction, QC, and DCB calibration.

• 3) Profiles:- Abel transform + hopefully some kind of an attempt to correct for the horizontal gradients via a model or TEC maps.

• 4) Scintillation maps for the 50 Hz data (generated by Bill S).Unclear what we will do with TBB. Plan to have an antenna on the roof.


Other new CDAAC features

● Daemon management infrastructure: daemons.pl● Technique for syncing up systems with CVS: cvsSync.pl● Improved file transfer technique 'QCP' helps TACC latency● New var1d code - tuned so that results are consistent with

radio occultation refractivity● COSMIC/CHAMP 6 LEO test case and operator interface:

got rid of many bugs in handling multiple LEOs. Now can display status of six satellites.

● Scripts in place to compare results between two clusters (leo-a and leo-b or TACC and CDAAC) on an ongoing basis.

● Improved Bernese V5.0 menu utility speeds processing up up to 5 fold


Summary

● CDAAC operational version is almost done ● Next delivery to NSPO/CWB delayed until Nov. 8 05● Remaining Tasks

●Tip code from NRL needs to be incorporated●TBB - need to install a receiver on roof get data ●Decoding / display of real-time payload health and diagnostics data needs to be improved●Display of results - show&tell needs to be finalized●….

Documents

COSMIC retreat October 2005 UCAR COSMIC Project Office CDAAC Readiness Status