RGPS Science Working Group Meeting (RGPSWG)
February 18-19, 1998
Jet Propulsion Laboratory

Please send request to Ron Kwok (ron.kwok@jpl.nasa.gov) if you are interested in receiving paper copies of material presented at the meeting.

Objectives of Meeting
- Provide overview of RGPS status
- Review investigations and data requirements
- Discuss validation issues
- Decide on RGPS processing/scheduling priorities
- Discuss data distribution/packaging/documentation
Summary of Recommendations of RGPSWG
The RGPS capabilities should be focused on completing the analysis of the follow ing image datasets prior to delivery of system to ASF (end of April 1999):
- Winter/Spring 1996-97, 3-day coverage within the ASF mask, 6-day coverage in t he Tromso mask. A 10 km grid (as opposed to 5 km) will be used to speed up processi ng.
- Summer 1997, 2 streams. (Sampling: 3-day, 10 km)
- Winter/Spring 1997-98, 2 streams to cover the SHEBA site (Sampling: 3-day, 10 km). We recognize that the set of products produced during this timeframe does not ad dress the needs of all investigators present at the meeting but is based on the verificati on and validation needs of the processor, the throughput of the RGPS and availability o f data in the current ASF archive.
Tromso data
- Tromso/DLT problem: The capability to ingest DLT data from Tromso does not exi st. A plan to address this issue based on the requirements of the RGPS should be put in place by ASF/ASFD.
- It is important to continue to acquire data in the Tromso mask although the nu mber of datatakes required to cover the mask should be optimized to reduce cost.
- The transition to the receipt of Eastern Arctic data from Svalbard rather than Tromso should be studied by ASF. What are issues involving overpass conflicts, data transmission from Svalbard, etc.?
SWB Processor throughput
- The SWB processor throughput should be increased to keep up with the demands of the RGPS.
Problems/Issues to be addressed by ASF and ASFD
The following outstanding issues/problems with ScanSAR wide B (SWB) image data should be fixed as quickly as possible.
- Radiometric calibration - the tool for converting the digital numbers in SWB data to normalized backscatter values has not been completed.
- Geolocation problems in ScanSAR data - the frame-to-frame misregistration error due to problems in the PRF ambiguity determination code in the SWB processor need to be resolved before RGPS processing can proceed.
- The data loss during the onset of melt in the past two seasons should be investigated. Is this a recurring problem due to the spacecraft/planning limitations?
- Hidden dataframes in the catalog - data frames that contain good radar data but are hidden from users should be recovered and made available to the RGPS.
- The reason for the bottleneck should be assessed to determine how to increase throughput. is network or cpu speed the issue? asf is running tests to determine this.
Further delay in the delivery of calibrated data to the RGPS will severely impact the current schedule and cost of RGPS development.
Data Distribution
All products will be available on the JPL RGPS website via FTP while the RGPS is resident at JPL. The capability for users to subscribe to notification messages on availability of RGPS products should be implemented.
Outreach/Report of Science Results
We will investigate the AGU EOS article once some good preliminary results are in. The various RGPS web sites should be linked together (ASF, JPL, UW) and have other sites link to us, such as the Global Change Master Directory. Conferences: Fall AGU is too early given the current processing schedule; IGARSS '99 was considered as a timely forum for publicizing RGPS results.
Next Meeting
The next RGPS meeting will be in about 1 year. The RGPS validation project plans to have a report out by the end of 1998, in time for review before themeeting.
More detailed minutes (from the notes of Harry Stern, APL)
These notes give a more-or-less chronological account of the meeting.

Drew gave a historical introduction to RGPS, from the early days of automated ic e tracking (1980s, IMAG, RAGTAG) to the ERS-1 GPS (1992-94) to the conception of RGPS (September 1993) and the subsequent development and meetings.

Ron gave the current status of Radarsat and RGPS (copies of his viewgraphs were handed out). Radarsat data comes to ASF from 3 sources: direct downlink within the ASF mask; direct downlink to Tromso with tapes sent to ASF; and on-board recorder (OBR) data downlinked to Gatineau and sent to ASF. Cycle 15 (November 4-27, 1996) is the first complete 24-day cycle for which data are available. The Arctic Snapshot has been attempting to acquire 3-day repeat coverage within the ASF mask and 6-day repeat coverage within the Tromso mask.

Major outages: May 16 to June 7, 1997 (satellite eclipse) and September 13 to Oc tober 30, 1997 (cycles 28 and 29, Antarctic Imaging Campaign (AIC)). *** Action Item: Ben will look into the satellite eclipse (which will happen every year) and see whether it is possible to avoid the loss of data during that critical time (melt onset).

Problems:

Radarsat data are still uncalibrated. RGPS needs calibrated data for the multiyear ice algorithm.
Geolocation errors. Some images (estimates range from 25% to 50%) have a 4-km along-track geolocation error due to PRF ambiguity. This creates artificial "openings" and "closings" perpendicular to the satellite track when the error is present in one image but not the other of a repeat pair.
Tromso media unreadable. Starting in February 1998, data from Tromso are arri ving at ASF on DLTs. ASF has no present or planned capability to read these tapes.

Some sample RGPS products are available on Ron's RGPS web site (http://www- radar.jpl.nasa.gov/rgps/radarsat.html). Ron and Glenn are preparing a Data Users Handbook for release in April 1998.

Delivery of RGPS to ASF is scheduled for April 1999. This date cannot be pushed back by any considerable amount because sustaining engineering ends at the end of 199 9, and a break-in period will be needed at ASF. The transition could take from 1 to 3 mo nths. The Working Group needs to decide which data RGPS should process before the transiti on.

Verne talked about the impact of RGPS on ASF. RGPS is a top priority for ASF. RGPS consumes about 1/3 of all ASF Radarsat data acquisitions. ASF must pay for the Tromso data, which takes away funds from other ASF activities. The annual cost is approximately $400k, or 5% of the ASF budget. The cost of Tromso data is "per datatake", so the cost can be lowered by planning long datatakes to cover the re ception area.

ASF data processing capacity is broken down as follows:
RGPS 33% AIC 27% NIC 20% All other users 20%.
AIC processing is scheduled to be completed in October 1998, but the upcoming NR A will probably soak up that processing capacity. Currently it takes about 1 cycle (24 days) to process 1 cycle of Radarsat data. Options for speeding up the processing are:

Upgrade the processor;
Get a new computer onto which some of the processing can be shifted.

Prasad directed us to prioritize the scientific needs of RGPS and ASF. The major relevant issues or tradeoffs are: the need for increased processing capacity, the importa nce of Tromso data, and the development of the capability to read DLTs.

Various investigators gave presentations about their work and the RGPS products they will need. Presentations were made by: Seelye Martin, Ron Lindsay, Rick Forster, Ron Kwok, Drew Rothrock, Harry Stern, Ben Holt, Jim Wilson and Marcus Speckhahn, Kim Partington, Mark Drinkwater, Igor Appel, and Lyn McNutt (on the second day).

Validation. Ron presented comparisons between buoy drift and RGPS-computed ice motion. Harry talked about the NASA-funded project called "Validation of RGPS Products" (Ron Lindsay, PI). Copies of the proposal are available from Harry. Members of the RGPS Science Working Group are encouraged to offer their views on validation issues and experience with RGPS products. Florence talked about a Ra darsat multiyear ice algorithm that uses dynamic thresholding to classify pixels. The Radarsat data need not be calibrated.

Nettie talked about the data acquisition and ordering process. The steps go som ething like this: Mission planning for the Snapshot is done by Ron at JPL. The NIC requireme nts are merged with this at ASF and submitted to CSA, which plans the acquisitions. Dat a acquired at ASF go into the queue to be scanned. Data acquired at other stations must be ordered by ASF. Scanned data are "visible" to users in the IMS. The RGPS has a direct line to the IMS, bypassing the usual interface. Ron orders the images (by cycle ) and ASF delivers them on tape.

Items that Nettie is looking into:

Create an ordering scheme that will allow more efficient processing, as she did with RAMP.
Figure out a way to track the data from mission planning to delivery.

Prasad talked about MTPE (now Earth Sciences) and Polar Programs. He said we need better outreach to the public. He also said NASA is using "metrics" to mea sure performance, such as the number of publications and citations. He wants a 1-2 p age summary of accomplishments from each project.

Ron talked about throughput issues. RGPS data are organized into "streams". A s tream contains the trajectories of all grid points defined on an initial datatake. St reams can be processed independently, and RGPS products are based on streams.

Operator intervention is necessary but time-consuming. Inspector Jim checks and corrects tracked points. Additional points are not added.

RGPS processing could be speeded up if the grid density were changed from 5 km t o 10 km. Other constraints: on-line storage capacity of 600 images, and the ability to process (at most) 2 regions or time spans simultaneously. It takes about 5 weeks for RGPS t o process 1 cycle of data, so ASF should be able to keep RGPS supplied with imagery at th eir current estimated throughput of 1 cycle per cycle.

We had a group discussion about processing scenarios. Drew made a chart showing the needs of the different investigators. The data processing must be prioritized b ecause of the throughput limitations. Here is a summary:

Winter/Spring 1996-97. Full Arctic imagery is available (ASF and Tromso). This is a strong point in favor of doing this period first, as the intention of RGPS has a lways been to cover the whole Arctic.

Summer 1997. Full Artic imagery is available. Summer is important to some investigators. The end of the satellite eclipse (June 7) to the start of the AI C (September 13) makes a natural period for processing.

Winter/Spring 1997-98. Full Arctic imagery is NOT available (Tromso/DLT problem ). However, this is the SHEBA year, so there is an increased interest in this perio d among some RGPS and non-RGPS investigators.

The proposed processing scenario is:

Winter/Spring 1996-97, 3-day coverage within the ASF mask, 6-day coverage in the Tromso mask. Use 10 km grid (as opposed to 5 km) to speed up processing.
Summer 1997, 2 streams (for faster results).
Winter/Spring 1997-98, 2 streams to cover the SHEBA site. In addition, the needs of certain other investigators could be met through RGPS Eulerian ice tracking. The goal would be to have all this done by the time of delivery of RGPS to ASF, April 1999.

Rick Guritz showed his SAR mosaics from cycle 15 and talked about banding, saturation over bright targets, along-track timing errors, and calibration.

Data distribution. Ron will continue to make RPGS products available via the web. He will compile a list of subscribers who will be notified by e-mail when new products are available. After the RGPS is delivered to ASF, then what? Perhaps the products would be made available through V1 IMS. There are no plans for web distribution at this point.

Outreach / publications. There was talk of an AGU EOS article once some good preliminary results are in, perhaps a year from now? We should link the RGPS web sites together (ASF, JPL, UW) and have other sites link to us, such as the Global Change Master Directory. Conferences: Fall AGU is too early; perhaps IGARSS '99 would be a good forum for publicizing RGPS results.

The next RGPS meeting will be in about 1 year. The RGPS validation project plans to have a report out by the end of 1998, in time for review before the meeting.

Attendees:

Igor Appel	Fairweather	optima@arctic.net	907-258-3446
Tom Bicknell	JPL	t.bicknell@jpl.nasa.gov	818-354-2523
David Cuddy	JPL	dcuddy@jpl.nasa.gov	818-354-2099
Glenn Cunningham	JPL	glenn@rgps1.jpl.nasa.gov	818-354-8328
Mark Drinkwater	JPL	mrd@pacific.jpl.nasa.gov	818-354-8189
Florence Fetterer	CU/NSIDC	fetterer@kryos.colorado.edu	303-492-442
Rick Forster	BPRC/OSU	forster@iceberg.mps.ohio-state.edu	614-292-1063
Charles Fowler	CU	cfowler@frodo.colorado.edu	303-492-0975
Prasad Gogineni	NASA/HQ	sgoginen@hq.nasa.gov	202-358-0746
Rick Guritz	ASF	rguritz@images.alaska.edu	907-474-7886
Ben Holt	JPL	ben@pacific.jpl.nasa.gov	818-354-5473
Verne Kaupp	ASF	vkaupp@images.alaska.edu	907-474-7263
Ron Kwok	JPL	ron@rgps1.jpl.nasa.gov	818-354-5614
Nettie LaBelle-Hamer	ASF	nettie@borealis.gi.alaska.edu	907-474-6167
Chi Lin	JPL	clin@solstice.jpl.nasa.gov	818-354-7209
Ron Lindsay	UW	lindsay@apl.washington.edu	206-543-5409
Xiang Lu	JPL	xiang@pacific.jpl.nasa.gov	818-354-6655
Chuck Luther	ONR	lutherc@onr.navy.mil	703-696-4123
Seelye Martin	UW	seelye@ocean.washington.edu	206-543-6438
Lyn McNutt	UAF/GI	lyn@dino.gi.alaska.edu	907-474-6077
Dave Nichols	JPL	dnichols@jpl.nasa.gov	818-354-8912
Kim Partington	NIC	kpartington@natice.noaa.gov	301-457-5314
Mitch Roth	UAF/ASF	roth@cs.uaf.edu	907-474-7098
Drew Rothrock	UW	rothrock@apl.washington.edu	206-685-2262
Jim Solinsky	Neptune	solinsk@millennianet.com	619-523-2023
Marcus Speckhahn	NPS	mmspeckh@oc.mps.navy.mil	408-372-7221
Harry Stern	UW	harry@apl.washington.edu	206-543-7253
Bjorn Thomsen	DCRS	bbt@emi.dtu.dk	818-354-8022
Jim Weirick	JPL	James.Weirick@jpl.nasa.gov	818-354-0296
Jim Wilson	NPS/Neptune	jhwilson@fea.net	714-366-6554
Yanling Yu	UW	yanling@apl.washington.edu	206-543-1354

RGPS Science Working Group Meeting (RGPSWG) February 18-19, 1998 Jet Propulsion Laboratory

RGPS Science Working Group Meeting (RGPSWG)
February 18-19, 1998
Jet Propulsion Laboratory