200806 NASA ESTC

From Seamonster
Jump to: navigation, search

NASA Earth Science Technology Conference June 24-26, 2008

A1P1 Ashit Talukder (JPL) - Cyclone ID and feedback between QuickSCAT and TRMM

work w/ Sci co-I to do cyclogensis (Africa..)

Q: Kumar (prediction of cyclone path? for sci..)

A1P2 Helen Conover (UAH) - Sensor Web protocols for satellite data into forecast models

OGC/Mike Botts, etc..

goal: explore SWE

using:

  • SOS: requesting, filtering retreiving sensor system info and observations (sensor obervation services)
  • O&M schema (XML)
  • SAS: web service interface for advertising publishing and subscribing to alerts from sensors
  • sensorML

SensorML process chain w/ both SOS/SAS integrated.. (Event Identification Subsystem)

(satellite footprint prediction SOS uses SensorML to determine sat location...)

Leveraging SAS package from 52North (XMPP implementation, using IM) (modified to includ geographic bounding box) (52North is mostly working with in situ) (52North is German group)

goes to WRF for US East Coast (so NCAT/ISET ?)

Lessons Learned: careful with upgrades (SOS 1.0 now...)

Oceans Interop Experiment ("cookbooks for buoys..)

Q: Karen Moe (sensor tasking?)

Michael Seablom - SW simulations

demo SW (for met), quantify savings and efficiency of using SW

improve weather forecast (5 degree error -> $10M/day cost to Calif power companies)

extend mission life by turning power off (LIDAR, laser life, for example)

model drives observing system (ugly system diagram)

run using assimilation of synthetic LIDAR, then take some out and look for impact on forecast accuracy

using adjoing studies to id sensitive regions is the right way. they did the "poor man's approach"--take two 72 hour forecasts, difference them, highlight big changes

cool movie with STK!

Michael Burl - Adaptive Sky

One year AIST (object-centric datasets w/ Adaptive Sky)

why SW...

some co-registration of "imagery" of different sensors (on A-train, AQUA/TERRA)

lots of Adaptive sky applications/demos

segmentation and tracking of images

Bezymiana eruption--track w/ 3 (or 4?) sats and 5 instruments.. to track ash cloud. using adaptive sky

seems to be more feature tracking than lots of SW (so this is a tool for integration into SW?)

seems to be software "component"?

(Q: Rob Sherwood - release of software for other SW PIs).

Kevin Knuth - info theory for climate variables

challenge is identifying relevant variables

thermodynamic/info theory generalities (systems, states)

histograms as PDF estimator. But problem w/ nbins..

use posterior probability to find optimal histograms

ok, lots of setup, now looking at cloud cover and cold tongue index (?)

use in robotics to automat sensor placement

Peter Fox - semantic mediation

defn semantics, RDF/OWL (no XML in talk before lunch)

don't need "complete semantics" to get started

barrier to interdisciplinary research: (is dataset A or B the right one to start with?)

-> looking at volcano, SOX, atm.. need quick overview of datasets, science fields, etc.

SESDI (project name)

need to shift away from instrument based queries to measurement based queries

page on volcanot-atmosphere: lots of data sources, processes, etc... Main task: detectiong and attribution

"mashup" is such an ugly word, and all context is lost..

GIMA ontology..

SEDRE (scientists semantically register data sets -- a desktop application) -- couldn't find this online.

LUNCH

Kumar's presentation -- track Education impacts

John Kinnebrew - MACRO

Apply MACRO to SEAMONSTER.

CORBA Component Model (CCM) overview -- nice slide build.

Deployment Infrastructure Overview - www.cs.wustl.edu/~schmidt/PDF/DanCE.pdf (not there!)

preserve state at shutdown (everything (middleware) is local.. lots of overhead) ccm_store (CCM?)

Antidio Viguria - SnoMotes

Colorado demo and description of slope finding, etc

two videos from Mendenhall

market based algorithms

Godwin Shen - distributed wavelet processing

entropy coding (NSTC 2007) (variable rate encoding)

way into routing optimization

tmote sky implementation

mote implementation (in lab)

DAY2

Costas Tsatsoulis - SW as multiagent, negotiating systems

SW: "sense on demand"

agents as coalition (no central authority)

architecture of reasoning component -- what not to do.. :)

  1. broadcast task to all
  2. each agent keep list of all capabilities

What to do: use a "matchmaker" -- central registry

matchmaker not enough--need complete (and agile) sensor model (sensorML)

cast in "proposal" model. Issue an RFP, agent receives RFP, decides to reply, compose a bid (commercial agents/sensors have a motivation to lie?)

Experiments w/ this.. Use SensorML to describe EO-1 sensors, then try this in matchmaker

simulated estuary, w/ pressure transducer, wind, rain. do a hurricane event simulation

want wind sensors, location is part of bid

looks like it is all written in java - would work on sentilla? would this work on tinyOS?

Ken Witt - SWAMO

distributed agent autonomy on satellite for operations

ST5, FlatSat, MidSTAR-1 (US Naval Academy SSP; IP in space) (test platform)

Genong Yu - asynchrony in sensor web

using OGC standards...

callback methods and some XMPP

using OGC SAS (XMPP) and WNS (SAS and SPS) for asynchrony

Yunling Lou - Forest Fire / UAVSAR

sensor web use case

flying on G3 right now, targetting Globalhawk

using Steve Chien's CASPER for autonomous planning

comms (high bandwidth for radar)

John Dolan - Ocean Sensing (HAB)

working at Wallops so far

OASIS - NOAA funded--three "good size" (15'?) boat/robots

uses rhodamine for synthetic HAB (with aerostat for validation/tracking)

uses software from SPAWAR (multi-robot control system)

instrumented drift buoy (GPS, data xmit) to get current information

just doing presence/absence for now

adaptive exploring and mapping

lots of work on pragmatic aspects of deployment (drift compensation) and then modeling of adaptive sampling

Ashit Talukder - coastal SW and autonomous control

current state of technology -- poor response to small scale, transient events (plume tracking..)

using NYOPS

observing (in measurements) fresh water pulse in NY harbor and out into ocean

modeling comms (sensor to logger)

optimize: sample rate, UUV location, assign sensors to datalogger

non-optimal, but fast solutions

compare adaptive vs uniform sampling with 1000 sensors

the storm surge we detected... (they are doing the adaptive sampling, or modelling it..) UUV is using adaptive controls

another one with 4 UUVs

goal optimization is a bit of a black art (how to weight..) -- just did serial ranking, rather than weighting

Payman Arabshahi - ocean, acoustics, ...

ROMS (Regional Ocean Modeling Systems)

TRLs for subsystems--acoustics is TRL-3, sensors/platforms are up at 8 or 9

calculating link capacity for underwater acoustics is really hard (talk to the Navy...)

lots of work on software underwater modems (test in SEAK?)

HV connector failed, cost $1M to fix (1km deep, 60 km offshore)

LUNCH

Andrea Donnellan - Quakesim

missed, out to GSFC with Anthony

Sharon Kedar - OASIS

Another OASIS ? (Volcano stuff..) (earlier OASIS was boats)

caught tail end.

using imotes

Ashley Davies - model-driven volcano

modeling different types of eruptions to better do "real-time" reporting and on-board (EO-1) data anslysis

looking at DR Congo Niragongo and Nyamuragira

ground volcano nodes in a pelican with BASIC stamp, USB, two undergrad engineering students funded by space grant


BREAK

Dan Mandl - SW 2.0

missed, talking with Ashit Talukder

Wei Ye - Sat SW Architecture

use WildBlue for satellite internet (few hundred kilobits/sec)

use motes (Mica2)

Paul Houster - prototype SW in land surface

Mahta Moghaddam - Soil Moisture

model of a few sensors to then scale up to 1000s

physical model, measurement model, control strategy, performance criterion, experiments & actuation

iUSBDAQ (actuation control unit is HYTEK iUSBDAQ-U120816) with wireless to desktop in lab

DAY3

Meemong Lee - SOX

not just physical sensor web, but concept of sensor web itself can be infused

"trade space"? (trade offs in design/coordination?)

metric for observation coordination optimization

model: SOX--has 4 components. OSSE is a "relatively well known" observation system simulation experiment framework

integrate individual instrument (on spacecraft) transfer functions to model atmospheric state as observed over many platforms and sensors (challenge)

Rob: TeamX takes science and designs platform; SOX take science and designs sensor web

JPL internal only for now (going through review)

Don Sullivan - processing services

laptop problems...

sensors > networks > webs > decision making

mentioned Kepler and BPEL (on the slide at least..)

wildfire/smoke

Bluesky (open source modeling framework)

will be running Saturday, all real time, online

Ikhana Web Service Flow (lots of OGC acronyms!) incorporated OGC XML from OASIS

OGC is not the only standards body developing in this arena

build SW in real time to work in real time

UAS/UAV work

FAA really good w/ this (Dryden support-in Edwards) - "FAA trusts us"

USFS -> particulate modeling

http://getbluesky.org/

Heavner / SEAMONSTER

Heather Conover (OIE/Ocean stuff links!)

Ashit - motes..

Boise Prof - mote power, letter of support req

SWAMO on microservers (Ken Witt, ...)

BREAK

David Lary

modelling/sensor web validation/action infrastructure (built for atmospheric use case, next step is ocean)

mostly sat/remote sensing built

interested in collab