CASA: A NEW PARADIGM FOR END USER DRIVEN DATA COLLECTION
Brenda PhilipsDirector, Industry, Gov’t, and End User
PartnershipsERC for Collaborative Adaptive
Sensing of the Atmosphere
AMS Corporate Forum 2007March 22, 2007
Outline
Motivation and Background User Driven Adaptive Scanning End User Policy in Test bed Next Steps
Engineering Research Centers
Research and develop technologies to generate a well-defined class of engineered systems with social and economic impacts.
Faculty, students and industry/practitioners work in a multi-disciplinary environment reflecting real-world technology.
CASA’s Focus: New weather observation system paradigm based on low-power, low-cost networks of radars.
Year 4 of a 10-year research project
10/27 F1 tornado illustrates limitations of current observation technology
Source: NWS Tallahassee Forecast Office http://www.srh.noaa.gov/tlh/Oct27event/
Watch Issued. No Warning.
“…did not have the correct rotational characteristic we expect from a tornadic storm. It’s tough to see small tornados…It’s a problem at large distances.”
Mismatch: technology vs. temporal and spatial scales for decision making
10,000 ft
tornado
wind
earth surface
snow
3.05
km
0 40 80 120 160 200 240RANGE (km)
Horz. Scale: 1” = 50 kmVert. Scale: 1” -=- 2 km
5.4
km
1 km
2 km
4 km
gap
NEXRAD 250 km spacing Horizon problem
causes coverage gap
~ 2 km resolution
5 min. updates Function
Autonomously “Sit and spin”
surveillance with “data push”
Mismatch: technology vs. temporal and spatial scales for decision makingNEXRAD 250 km spacing Horizon
Problem: Middle to upper troposphere coverage
~ 2 km resolution
5 min. updates Function
Autonomously “Sit and spin”
surveillance with “data push”
10,000 ft
tornado
wind
earth surface
snow
3.05
km
0 40 80 120 160 200 240RANGE (km)
Horz. Scale: 1” = 50 kmVert. Scale: 1” -=- 2 km
5.4
km
1 km 2
km
4 km
gap
CASA addresses “mismatch” with DCAS (distributed collaborative adaptive sensing)
10,000 ft
tornado
wind
earth surface
snow
3.05
km
3.05
km
0 40 80 120 160 200 240RANGE (km)
Short range (~ 30 km) radars
Lower troposphere coverage
100’s meter resolution
Avg. 30 second updates
Adaptive Scanning based on user needs, “data pull”
“Sense the Atmosphere where and when user needs are greatest”
End User Integration links technology with warning and response
streamingstorage
storage
queryinterface
data
Resource planning,optimization
data policy
resource allocation
SNR
Meteorological DetectionAlgorithms
1 2 3 4 5 6 7 8 9A G3G3G3G3G3G3G3G3G3B G3G3G3G3G3G3G3G3G3C G3G3G3G3G3G3G3G3G3DG3G3G3G3G3G3G3G3G3E G3G3G3G3G3G3G3G3G3F G3G3G3G3G3G3G3G3G3GG3G3G3G3G3G3G3G3G3HR1R1R2R2R1G3C2G3G3I R1F1F2, R1F2,H2R1G3C2G3G3J R1H1,F1H1,F1T2,R1R1G3C2G3G3KR1H1T2,H1T2,R1R1G3G3G3G3
Feature Repository
MC&C: Meteorological command and control
DCAS User Driven Adaptive Scanning
User Interaction with new technology for decision
makingImpact of DCAS Technology on Communications, Public Response, and Vulnerability
Social and Economic Value of CASA Data
We interact with practitioners in the public and private sectors
streamingstorage
storage
queryinterface
data
Resource planning,optimization
datapolicy
resource allocation
SNR
Meteorological DetectionAlgorithms
1 2 3 4 5 6 7 8 9A G3G3G3G3G3G3G3G3 G3B G3G3G3G3G3G3G3G3 G3C G3G3G3G3G3G3G3G3 G3D G3G3G3G3G3G3G3G3 G3E G3G3G3G3G3G3G3G3G3F G3G3G3G3G3G3G3G3G3G G3G3G3G3G3G3G3G3G3H R1R1R2R2R1G3C2G3G3I R1F1F2, R1F2,H2R1G3C2G3G3J R1H1,F1H1,F1T2,R1R1G3C2G3G3K R1H1T2,H1T2,R1R1G3G3G3G3
Feature Repository
MC&C: Meteorological command and control
Data
IntermediariesValue Added Data Users
Private Sector: Value-Added Products
Products and Services, Decision Support Systems
Federal Government
Institutions/Industry
Academic Institutions/Researchers
Research
Emergency Response
Public
State Government Public
We have a multidisciplinary team
Brenda Philips, MBA, UMass Ben Aguirre, Sociology, UDelEllen Bass, Human Factors, UVaWalter Diaz, Political Science, UPRM Kevin Kloesel, Meteorology, OUDave Pepyne, ECE, UMassHavidan Rodriguez, Sociology, UDel
Roman Krzysztofowicz, Decision Sciences, UVa,
Oklahoma Test Bed: Severe Storms
4-node mechanically scanned radar network
36 km range 25 – 100 m resolution Adaptive: Multi-
elevation sector scans, pinpointing
Pilot User Group: WFO Norman, EMs, Researchers, (Media)
7000 sq. km (90 km long)
2 tornados
4 tornado warnings
50 severe storm warnings
resource allocation
MC&C: Meteorological command and control
1 2 3 4 5 6 7 8 9A G3 G3 G3 G3 G3 G3 G3 G3 G3B G3 G3 G3 G3 G3 G3 G3 G3 G3C G3 G3 G3 G3 G3 G3 G3 G3 G3D G3 G3 G3 G3 G3 G3 G3 G3 G3E G3 G3 G3 G3 G3 G3 G3 G3 G3F G3 G3 G3 G3 G3 G3 G3 G3 G3G G3 G3 G3 G3 G3 G3 G3 G3 G3H R1 R1 R2 R2 R1 G3 C2 G3 G3I R1 F1 F2, R1 F2,H2 R1 G3 C2 G3 G3J R1 H1,F1 H1,F1 T2,R1 R1 G3 C2 G3 G3K R1 H1 T2,H1 T2,R1 R1 G3 G3 G3 G3
Feature Repository
query
optimization
meteorologicaltask
generation
userrules,
Weights
End users: NWS,emergencyresponse
streamingstorage
detection algorithms
resource allocation
MC&C: Meteorological command and control
1 2 3 4 5 6 7 8 9A G3 G3 G3 G3 G3 G3 G3 G3 G3B G3 G3 G3 G3 G3 G3 G3 G3 G3C G3 G3 G3 G3 G3 G3 G3 G3 G3D G3 G3 G3 G3 G3 G3 G3 G3 G3E G3 G3 G3 G3 G3 G3 G3 G3 G3F G3 G3 G3 G3 G3 G3 G3 G3 G3G G3 G3 G3 G3 G3 G3 G3 G3 G3H R1 R1 R2 R2 R1 G3 C2 G3 G3I R1 F1 F2, R1 F2,H2 R1 G3 C2 G3 G3J R1 H1,F1 H1,F1 T2,R1 R1 G3 C2 G3 G3K R1 H1 T2,H1 T2,R1 R1 G3 G3 G3 G3
Feature Repository
query
optimization
meteorologicaltask
generation
userrules,
Weights
End users: NWS,emergencyresponse
streamingstorage
detection algorithms
resource allocation
MC&C: Meteorological command and control
1 2 3 4 5 6 7 8 9A G3 G3 G3 G3 G3 G3 G3 G3 G3B G3 G3 G3 G3 G3 G3 G3 G3 G3C G3 G3 G3 G3 G3 G3 G3 G3 G3D G3 G3 G3 G3 G3 G3 G3 G3 G3E G3 G3 G3 G3 G3 G3 G3 G3 G3F G3 G3 G3 G3 G3 G3 G3 G3 G3G G3 G3 G3 G3 G3 G3 G3 G3 G3H R1 R1 R2 R2 R1 G3 C2 G3 G3I R1 F1 F2, R1 F2,H2 R1 G3 C2 G3 G3J R1 H1,F1 H1,F1 T2,R1 R1 G3 C2 G3 G3K R1 H1 T2,H1 T2,R1 R1 G3 G3 G3 G3
Feature Repository1 2 3 4 5 6 7 8 9
A G3 G3 G3 G3 G3 G3 G3 G3 G3B G3 G3 G3 G3 G3 G3 G3 G3 G3C G3 G3 G3 G3 G3 G3 G3 G3 G3D G3 G3 G3 G3 G3 G3 G3 G3 G3
1 2 3 4 5 6 7 8 9A G3 G3 G3 G3 G3 G3 G3 G3 G3B G3 G3 G3 G3 G3 G3 G3 G3 G3C G3 G3 G3 G3 G3 G3 G3 G3 G3D G3 G3 G3 G3 G3 G3 G3 G3 G3E G3 G3 G3 G3 G3 G3 G3 G3 G3F G3 G3 G3 G3 G3 G3 G3 G3 G3G G3 G3 G3 G3 G3 G3 G3 G3 G3H R1 R1 R2 R2 R1 G3 C2 G3 G3I R1 F1 F2, R1
E G3 G3 G3 G3 G3 G3 G3 G3 G3F G3 G3 G3 G3 G3 G3 G3 G3 G3G G3 G3 G3 G3 G3 G3 G3 G3 G3H R1 R1 R2 R2 R1 G3 C2 G3 G3I R1 F1 F2, R1 F2,H2 R1 G3 C2 G3 G3J R1 H1,F1 H1,F1 T2,R1 R1 G3 C2 G3 G3K R1 H1 T2,H1 T2,R1 R1 G3 G3 G3 G3
Feature Repository
query
optimizationoptimization
meteorologicaltask
generation
meteorologicaltask
generation
userrules,
Weights
userrules,
Weights
End users: NWS,emergencyresponse
streamingstorage
streamingstorage
detection algorithmsdetection algorithmsdetection algorithms
1. Radar network scans atmosphere (sector, pinpointing, 360 degree) and sends data to MC&C
2. Detection algorithms identify weather features in radar data.
3. Weather features are “posted” in Feature Repository, a 3-d grid of radar coverage area.
4. Tasks are generated based on clustering similar weather features.
5. Optimal radar scan configuration developed based on:
• How Important is a task to users?
• What is the quality of the scan?
ResearchersMedia
user policy
Data Pull: system optimally collects data based on user needs, evolving weather and radar capabilities.
User Policy: How Important is the data to users? User Needs User Needs translated into rules for system
What detected weather feature? How many radars? What horizontal and vertical coverage? How often should it be scanned?
“ Needs” evolving from subjective to more objective measures
Stated preferences Space/Time variablity of weather Socio-economic impacts
USER RULE FOR EMERGENCY MANAGERS: If rotation is detected, then scan the lowest two elevations radars every 30 seconds.
Goal: Geographically specific information for public notification, responder deployment
User Policy: How Important is the data to users? User Weights Determines the relative importance of
different user groups in case of resource conflict.
Mechanism established, Policy for setting weights not established
Still understanding the level of resource contention in the system
Weighting could be set for socio-economic benefit, profits, for specific weather events, etc.
ttasksCionsconfigurat
CtQktUC,,
,,maxarg*
End User Policy – How important is task t to the users?
ggroups
gg ktUwktU,
,),(
User Weights User Rules
End User Policy – How important is task t to the users?
ggroups
gg ktUwktU,
,),(
User Weights User Rules
Another view of optimization
Challenges for developing user rules (preferences) Eliciting preferences
from users for a new sensing paradigm Iterative process Subject Matter Experts Use qualitative approaches
initially
Getting system designers (computer scientists, engineers) to understand user decision process and translate that into code.
Initial approach to user rules focuses on weather features
Feedback from NWS: “That’s not how we make decisions…”
Goal: Issue warnings, communicate expertise save lives, property
Radar data increases or reduces forecaster confidence
“mental movie” focus on areas of uncertainty; not always
determined by radar data
Training
Expectations
Staffing Issues
Coordination
Conditions
observed
outside
Location,
expected
impact
Ongoing
Satellite
Model
1,2,3,etc
guidance
Radar
1,2,3
Data
Conceptual
models
Equipment status
Gut feeling
Warn
Training
Expectations
Staffing Issues
Don’t Warn
Communications
Conditions
observed
outside
Location,
expected
impact
Ongoing
Mesoanalysis
Satellite
Model
1,2,3,etc
guidance
Radar
1,2,3,
Data
Equipment status
Ground truth
“NWS Warning Process”, Liz Quoetone, NWS Warning Decision Training Branch, presentation at 2005 CASA workshop
Changes to rules
Incorporation of interval-based scanning
Expansion of the definition of a storm
cell
Introduction of contiguous scans Dynamic Data Requests
Current rules focus on “time since last scanned”
Rules Ruletrigger
SectorSelection
Elevations #Radars
Contiguous Samplinginterval
NWS
N1 time 360 Lowest two 1 Yes 1 / min
N2 storm task size full volume 1 Yes 1 / 2.5 min
Researcher
R1 rotation task size full volume 2+ Yes 1 / 30 sec
R2 reflectivity task size Full volume 2 Yes 1 / min
R2 velocity task size lowest two 2+ Yes 1/ min
R3 time 360 Full volume 1 No 1/ 5 min
EMs
E1 time 360 lowest 1 Yes 1 / min
E2 reflectivity over AOI
task size lowest 1 Yes 1 / 2.5 min
E3 velocity over AOI
task size lowest 2 Yes 1/ 2.5 min
OS
O1 time 360 lowest two 1 No 1 / 5 min
Table 1. User Rules, End User Policy, Version 1
User WeightsFigure 3.2aFigure 3.2a
User Weights
August 15 frontal boundary with isolated storm cells
3 radars operational Single radar attenuation correction Clutter, velocity, and networked attenuation alg.
not yet installed DCAS running based on reflectivity rules and
detections.
22:08:33 22:08:53 22:09:59 22:10:56
22:11:58 22:12:59 22:12:47
20060815-222919
20060815-222906
20060815-222924 20060815-222928
20060815-222911 20060815-222915
3.00 4.00 6.00
8.00 11.00 14.00
20060815-231941 20060815-231945 20060815-231950
20060815-231955 20060815-232000 20060815-232005
Next Steps Study user behavior in test bed with actual data Study impact of weights on system function Develop revisions to end user policy with “ Needs”
evolving from subjective to more objective measures
Stated preferences, observed preferences Space/Time variability of weather Socio-economic impacts
Expand user base to include public, private companies
Launch Decision Sciences Project integration socio-economic factors into adaptive scanning Creating an end-to-end decision model of network
Thank you
System Parameters
Operating frequency 9.3 GHz
Wavelength 0.03 m Antenna Diameter 1.20 m Antenna Beamwidth 1.8
deg Antenna Gain 38 dB Max radar scanning speed
35 deg/sec Max radar acceleration 50
deg/sec2 Maximum range 36 km
Range resolution 26 m Effective Transmitter
Power 12.5 kW Average Transmitter Power
25 W Dual Pulse Repetition Frequency 1.6kHz, 2.4 kHz Noise Figure 5.5 dB System Losses -20 dB Mean Sensitivity 2.8 dBZ
Extra Slides
Test beds instantiate end-to-end system concepts
Rain, Urban Flooding (Houston)
Rain mapping, distributed hydro. modeling, flood predicting & response in an urban zone.
Wind, storm prediction (Oklahoma)
Wind mapping (100’s m resolution, 10’s second update) for detecting, pinpointing, forecasting wind events; 30 km node spacing.
Rain, mountainous terrain (Puerto Rico – student led)
Off-the-Grid Radar Network for quantitative precipitation estimation (QPE) over complex terrain, student-led project
User Rules and Weights
rule trigger
sector selection
elevations # radars contiguous sample frequency
data quality
NWS Rule 1 time 360 lowest 1 Yes 1 / min High Rule 2 storm task size full
volume 1 Yes 1 / 2.5
min High
Researcher Rule 1 rotation task size full
volume 2+ **** Yes 1 / 30
sec? High
Rule 2 reflectivity task size lowest two
1 Yes 1 / min High
Rule 2 velocity task size lowest two
2+ **** Yes 1/ min High
Rule 3 time 360 to get all 7 every 15 min
1 No 1/ 5 min High
EMs Rule 1 time 360 lowest 1 Yes 1 / min High Rule 2 reflectivity
over AOI task size lowest 1 Yes 1 / min High
Rule 3 velocity over AOI
task size lowest 2+ **** Yes 1/ 2.5 min
OS Rule 1 time 360 lowest
two 1 No 1 / 5 min Low
End User Policy Rules, Version 1End User Policy Control
Tasks (t) (blue) Scans (C) (green)
• Mechanism in place for adjusting weights (Wg)
• Default: all weights = 1.0
• Rules based on scan frequency and/or feature
• Utility based on time since the rule was last scanned (Ug)
ttasks
CionsconfiguratCtQktUJ
,,
),(,max
End User Policy – How important is task t to the users?
ggroups
gg ktUwktU,
,),(
User Weights User Rules
ttasks
CionsconfiguratCtQktUJ
,,
),(,max
End User Policy – How important is task t to the users?
ggroups
gg ktUwktU,
,),(
User Weights User Rules
ttasksCionsconfigurat
CtQktUC,,
,,maxarg*
How important is task t to the users?
Research Organization
Sensing
Distributing
Analysis & Prediction
Education
TechnicalIntegration
End-userIntegration
We interact with practitioners in the public and private sectors
streamingstorage
storage
queryinterface
data
Resource planning,optimization
datapolicy
resource allocation
SNR
Meteorological DetectionAlgorithms
1 2 3 4 5 6 7 8 9A G3G3G3G3G3G3G3G3 G3B G3G3G3G3G3G3G3G3 G3C G3G3G3G3G3G3G3G3 G3D G3G3G3G3G3G3G3G3 G3E G3G3G3G3G3G3G3G3G3F G3G3G3G3G3G3G3G3G3G G3G3G3G3G3G3G3G3G3H R1R1R2R2R1G3C2G3G3I R1F1F2, R1F2,H2R1G3C2G3G3J R1H1,F1H1,F1T2,R1R1G3C2G3G3K R1H1T2,H1T2,R1R1G3G3G3G3
Feature Repository
MC&C: Meteorological command and control
Data
IntermediariesValue Added Data Users
Private Sector: Value-Added Products
Products and Services, Decision Support Systems
Federal Government
Institutions/Industry
Academic Institutions/Researchers
Research
Emergency Response
Public
State Government Public
August 15 Storm East-to-west warm
frontal boundary with isolated storm cells; second area of stratiform rain with embedded convection
NWS issued one thunderstorm warning within the network for Grady County at 2130 UTC
Several severe wind reports were recorded just south of IP1 at approximately 0000 UTC.
CASA Test Bed 3 radars operational Single radar attenuation
correction Clutter, velocity, and
networked attenuation alg. not yet installed
DCAS running based on reflectivity rules and detections.
CASA data Frederick Radar
Phases of Response: High Level Decisions
3 days 2 days 24 hrs 1 hr. 4 hrs. Event~18 min.
PublicNotification
Spotter/Resp
Deployment
Spotter, ResponderAlert
Emerg.Response
Emergency Managers (Towns/Streets)
ProtectiveAction/
No Action
Understanding, believing,confirming, personalizing,
action necessary,action feasible
ImpactPublic (Streets)
Severe Weather Watch
MesoscaleDiscussion
1 Day Convective
Outlook
2 Day Convective
Outlook
3 Day Convective
Outlook
NWS–SPC (Regions)
DCAS: NowcastingSht. NWP, Storm Genesis
Boundary Sensing
DCAS: Feature Detection,Severe WX Sensing,
Nowcasting
DCAS: Ensemble ForecastsClear-Air Sensing
CASA Researchers (Rgn, Cty, Town, Sts.)
NWS–WFO (Counties/ Towns) WARNING
Short Term Forecast
Special WX Statement
Sht. Term ForecastWX Statement
Pre storm Environment Watch Warning Event
Spotter, ResponderResource Assessment
Hazardous Weather Outlook
Key Influencers
Existing practices
Existing sources of information and their perceived uncertainty: existing weather data and models, media info, ground truth
Organizational Issues: procedures, culture, training, evaluation metrics.
3 days 2 days 24 hrs 1 hr.4 hrs. Event~18 min.
PublicNotification
Spotter/Resp
Deployment
Spotter, ResponderAlert
Emerg.Response
Emergency Managers (Towns/Streets)
ProtectiveAction/
No Action
Understanding, believing,confirming, personalizing,
action necessary,action feasible
ImpactPublic (Streets)
Severe Weather
Watch
MesoscaleDiscussion
1 Day Convective
Outlook
2 Day Convective
Outlook
3 Day Convective
Outlook
NWS–SPC (Regions)
DCAS: NowcastingSht. NWP, Storm Genesis
Boundary Sensing
DCAS: Feature Detection,Severe WX Sensing,
Nowcasting
DCAS: Ensemble ForecastsClear-Air Sensing
CASA Researchers (Rgn, Cty, Town, Sts.)
NWS–WFO (Counties/ Towns) WARNING
Short Term Forecast
Special WX Statement
Sht. Term ForecastWX Statement
Pre storm Environment Watch Warning Event
Spotter, ResponderResource Assessment
Hazardous Weather Outlook
3 days 2 days 24 hrs 1 hr.4 hrs. Event~18 min.
PublicNotification
Spotter/Resp
Deployment
Spotter, ResponderAlert
Emerg.Response
Emergency Managers (Towns/Streets)
ProtectiveAction/
No Action
Understanding, believing,confirming, personalizing,
action necessary,action feasible
ImpactPublic (Streets)
Severe Weather
Watch
MesoscaleDiscussion
1 Day Convective
Outlook
2 Day Convective
Outlook
3 Day Convective
Outlook
NWS–SPC (Regions)
Severe Weather
Watch
MesoscaleDiscussion
1 Day Convective
Outlook
2 Day Convective
Outlook
3 Day Convective
Outlook
NWS–SPC (Regions)
DCAS: NowcastingSht. NWP, Storm Genesis
Boundary Sensing
DCAS: Feature Detection,Severe WX Sensing,
Nowcasting
DCAS: Ensemble ForecastsClear-Air Sensing
CASA Researchers (Rgn, Cty, Town, Sts.)
NWS–WFO (Counties/ Towns) WARNING
Short Term Forecast
Special WX Statement
Sht. Term ForecastWX Statement
Pre storm Environment Watch Warning Event
Spotter, ResponderResource Assessment
Hazardous Weather Outlook
Societal Issues: access to information and training; differences based on education, gender, race, income.
Risk Communication: tone and content of message, Multi-directional communications, social networks, etc.
NWS, EM, ResearcherObservation/Data Needs
NWS Warning
Socio-economicImpacts
ActualWeather
Socio-economic
Vulnerability
OtherWeather
InformationSources
PublicResponseDecisions
DCASResource
Optimization
DCAS Sensing
Capabilities
GISInformation
RiskCommunication/
PerceptionLead
Time/WarningArea
DCASScanningStrategy
DCASUser
Interface
DCASDetectingPredicting
EM Decisions
3 yr project for end-to-end decision model
• To develop an end-to-end integrated decision model for DCAS systems (Integrated System Model) from targeted observation, detection, forecast, warning, risk perception and response to socioeconomic impact.
• Use socioeconomic measures to drive CASA’s resource allocation and optimization.
• Implement decision model in an expanded DCAS system emulator that simulates warning, response, and impact.