Some Results from Analysis of ER-2 Data
Hi,
here there is a sample of some interesting objects we found in ER-2 data doing semi automatic detection
Candidate_Name      ,Filename   ,Date     ,UTC    , Longitude,Latitude,  X,  Y, Pk, N59438221_121552_35980,59438221.tif,2007-02-02,18:17:34,-121.55193,35.98013,2114,3125, 97, 659438221_121543_35974,59438221.tif,2007-02-02,18:17:34,-121.54253,35.97372,2435,2568, 97, 959438222_121580_35985,59438222.tif,2007-02-02,18:18:04,-121.58027,35.98466,1606,1497,193, 8
 …..
Â
The whole sample inÂ
 http://skydev.pha.jhu.edu/nieto/tenacious
It would be great if we could collect coordinates in a common way and do cross-matches between images from different data sources or data analysis.
Jeff Valenti and Maria Nieto-Santisteban
February 5th, 2007 at 3:34 am
Can you be more specific about the coordinates integration problem? If we need to knock up some code to canonicalize and load into a DBMS, goodness knows there are many able hands here to serve…
February 5th, 2007 at 5:14 am
Do you guys have the original TIF images ?
Igor.
February 5th, 2007 at 6:13 am
Maybe having access to the satellite data originally gathered from Ikonos and comparing the GPS coordinates you guys have uncovered will reduce the number of artifacts ? I am sure you have already thought of this.
With regards to wind data, when we launch balloons we check this database for predictions: http://weather.uwyo.edu/surface/meteogram/
but I am also sure there is Coast Guard database for that as well.
Igor.
February 5th, 2007 at 7:51 am
Nice work. What is the next step to actually see what these objects are? The ER-2 images were collected a few days ago, and the objects may have moved by now.
February 5th, 2007 at 9:31 am
The problem is not loading into the database but to get the candidates in the same schema.
Candidate_Name , Filename,Date,UTC, Longitude, Latitude, X, Y, Pk,
We need to transform from X, Y into Lon, Lat. Tricky some times but be possible.
We have code to do the finding and match in SQL. We need the positions. With that we can start trying to find the moving objects.
TIF came from the ftp://bg-login2.sdsc.edu/ER2
Maria & Jeff
February 5th, 2007 at 11:24 am
hi,
Not sure if it will help but the skyserver sql code has stored procedures and user defined functions for the conversion from X,Y to lat/lon.
If you need the actual code please let me know.
Best Regards
Trevor
February 5th, 2007 at 1:07 pm
Does anyone know — what color is the below-waterline paint on Tenacious?
If the keel is lost, a sailboat may float inverted for weeks. The color from above would then be a combination of the topsides color (red) plus the bottom paint color. I am thinking in particular of images like this:
http://skydev.pha.jhu.edu/nieto/tenacious/images/ER-2/59438224_121732_35972.jpg
February 5th, 2007 at 1:30 pm
I believe it was David Tennenhouse who yesterday said there were over 1500 possible hits logged already. What I’d like to do is start filtering these out so we can get the most likely targets to the Coast Guard ASAP. How I’d like to do this is to have an inbound feed of potential sightings and build a mapping and use it for filtering. What are vital here are timestamps along with lat-long coordinates and some ID of the dataset from which they came, type of data and specific item (”granule”) from which they came. (All of the above are within my area of expertise, if I can help.) Note that I’m particularly keen on integrating data from disparate datasets for this coordinated effort.)
I’d like to integrate this data with wind/current data, also with date information attached, to try and construct trajectories for the potential sightings. (This is not an area of expertise for me, so pointers either to algorithms, systems, or people woudl be great.) On the somewhat reasonable presumption that Jim’s not able to steer, we could confirm or eliminate based upon a set of criteria, including:
- lack of sigtings in previous/subsequent data
- clear change in direction that doesn’t fit wind/current changes
- lack of motion
…And probably others that may come to mind.
Trouble is, I’m having a hard time getting started. I’ve tried to read the entire blog and all subthreads and I still haven’t figured out coordination of available data sets, collection of hits, etc. I could use any and all help. …Solving problems like converting from x/y in an image to lat long should be straight forward and already have been done for all datasets involved - who’s collecting these?
I volunteer for any/all of these collection activities that aren’t already handled so I can get to work sorting through the inbound “possible” list. I also am ready to help with collecting or at least providing pointers to whatever processing algorithms are involved, such as conversions to lat/long. I may be able to help automate much of this as automating data processing (across organizations) is what my company does (among other things)…
Should I leave this as a comment here, or make a more obvious posting for help along these lines?
Richard
February 5th, 2007 at 2:51 pm
Richard,
For the marine data, there seems to be a software that does that from observation + modeling: http://www.oceanweather.com/data/index.html
they say they do not have archived data off-hand and reserve that access to commercial/paying entities.
http://www.oceanweather.com/data/SW-US/index.html
the text version is here with the buoys numbers:
http://www.oceanweather.com/data/SW-US/marine_text.html
the buoys are listed in the NOAA site, for instance:
http://www.ndbc.noaa.gov/station_page.php?station=46063
with the quickscat wind data asociated with them:
http://www.ndbc.noaa.gov/quikscat.php?station=46063
but most importantly, all the data 250 km around that buoy is here:
http://www.ndbc.noaa.gov/radial_search.php?lat1=34.27N&lon1=120.70W&uom=E&dist=250
but all these are for the past day, historical data for each of the buoy can be gotten from there:
http://www.ndbc.noaa.gov/maps/monterey_bay_hist.shtml
but in this context, historical means 2006. There seems to be no January or February 2007 data.
contacting them might be a good option:
http://www.ndbc.noaa.gov/contact_us.shtml
By the way, you should not make it a comment, you should make it a full post because right now, even if there are good data taken on three different occasions, the next effort will focused on removing hypotheses by cross referencing the findings from the three databases (ikonos, er-2, ikonos…) and other sightings. More importantly, I believe that some of the experienced sailors watching this effort will help a lot in evaluating whether any of the scenario being considered are suboptimal or making sense.
Just having the current sightings from the JHU folks, and the results of the Mechanical Turk overlayed on the same map (with different colors to show the difference in time) as well as sighting information would help a lot.
Igor.