1

Algorithm IntegrationAlgorithm Integration

Presented byPresented by

Walter WolfWalter WolfAWG Integration Team LeadAWG Integration Team Lead

NOAA/NESDIS/STARNOAA/NESDIS/STAR

Algorithm Integration Algorithm Integration Team (AIT) East MembersTeam (AIT) East Members

Walter Wolf – Government LeadWalter Wolf – Government Lead Shanna Sampson – Project LeadShanna Sampson – Project Lead Xingpin Liu – QA Lead & Monitoring ToolsXingpin Liu – QA Lead & Monitoring Tools Wayne MacKenzie – GSP/Harris/AWG LiasonWayne MacKenzie – GSP/Harris/AWG Liason Larisa Koval – Documentation Larisa Koval – Documentation Yunhui Zhao – CM Yunhui Zhao – CM Aiwu Li – Algorithm IntegrationAiwu Li – Algorithm Integration Tianxu Yu – Algorithm IntegrationTianxu Yu – Algorithm Integration Rickey Rollins – Algorithm IntegrationRickey Rollins – Algorithm Integration Veena Jose – Algorithm IntegrationVeena Jose – Algorithm Integration Meizhu Fan – Algorithm testingMeizhu Fan – Algorithm testing Zhaohui Zhang – Algorithm Integration/Framework SupportZhaohui Zhang – Algorithm Integration/Framework Support Kristina Sprietzer – Framework SupportKristina Sprietzer – Framework Support Hua Xie – Sounding IntegrationHua Xie – Sounding Integration Wendy Zhang – Imagery IntegrationWendy Zhang – Imagery Integration Haibing Sun – Physical CollocationsHaibing Sun – Physical Collocations

AIT Midwest MembersAIT Midwest Members

Ray Garcia – AIT Midwest LeadRay Garcia – AIT Midwest Lead William Straka – Algorithm IntegrationWilliam Straka – Algorithm Integration Graeme Martin – Algorithm IntegrationGraeme Martin – Algorithm Integration Eva Schiffer – Glance DevelopmentEva Schiffer – Glance Development Bob Holz – Physical Collocation LeadBob Holz – Physical Collocation Lead Fred Nagle – Navigation ExpertFred Nagle – Navigation Expert Greg Quinn – Physical Collocation SupportGreg Quinn – Physical Collocation Support Ralph Kuehn – Physical Collocation SupportRalph Kuehn – Physical Collocation Support

KeywordsKeywords

AmbrosiaAmbrosia

HousekeepingHousekeeping

EnterpriseEnterprise

ContinuityContinuity

OpportunityOpportunity

4

5

OutlineOutline

HousekeepingHousekeeping» Accomplishments Over the Past YearAccomplishments Over the Past Year» Option 2 Stragglers – Visibility, Rainfall Potential, Option 2 Stragglers – Visibility, Rainfall Potential,

Rainfall ProbabilityRainfall Probability

The Way ForwardThe Way Forward» EnterpriseEnterprise» ContinuityContinuity» Enterprise & ContinuityEnterprise & Continuity

SummarySummary

HousekeepingHousekeeping

6

7

Milestones for Milestones for Baseline Products – Scheduled Baseline Products – Scheduled

Deliveries CompleteDeliveries Complete

Deliveries and Milestones» Draft ATBD – September 2008» 80% ATBD and Algorithm Package –

September 2009» 100% ATBD and Algorithm Package –

September 2010 & December 2010 respectively» REMOVED – Maintenance Delivery of ATBD

and Algorithm Package – September 2012

8

Milestones for Milestones for Option 2 Products – Scheduled Option 2 Products – Scheduled

Deliveries CompleteDeliveries Complete

Deliveries and Milestones» Draft ATBD – September 2008» 80% ATBD and Algorithm Package –

September 2010» 100% ATBD and Algorithm Package –

September 2011–Visibility, Rainfall Potential, and Probability

of Rainfall not included

9

Milestones for Milestones for 2012 Option 2 Products 2012 Option 2 Products

Visibility, Probability of Rainfall and Rainfall Visibility, Probability of Rainfall and Rainfall PotentialPotential

Deliveries and Milestones» Draft ATBD – September 2008» 80% ATBD and Algorithm Package –

September 2010» 100% ATBD and Algorithm Package –

September 2012

Preparing for Preparing for 2011 Delivery2011 Delivery

Conducted 6 Option 2 Algorithm Readiness ReviewsConducted 6 Option 2 Algorithm Readiness Reviews

Reviewed 18 Option 2 ATBDsReviewed 18 Option 2 ATBDs

Updated the Algorithm Interface and Ancillary Data Updated the Algorithm Interface and Ancillary Data Description Document (AIADD)Description Document (AIADD)

Wrote the Wrote the Baseline Algorithm Product Performance Baseline Algorithm Product Performance Monitoring DocumentMonitoring Document

10

AccomplishmentsAccomplishments

Deliver Algorithm Package containing the Option 2 100% Deliver Algorithm Package containing the Option 2 100% algorithms on September 30, 2011algorithms on September 30, 2011» ATBDATBD» Algorithm Interfaces and Ancillary Data Description (AIADD) DocumentAlgorithm Interfaces and Ancillary Data Description (AIADD) Document» Test DataTest Data

– Proxy and Simulated Input Data SetProxy and Simulated Input Data Set– Output Data SetsOutput Data Sets– Associated Coefficient Data SetsAssociated Coefficient Data Sets

» DD250 FormDD250 Form» MD5sum value for each file (check sum)MD5sum value for each file (check sum)

Delivered the Delivered the Baseline Algorithm Product Performance Baseline Algorithm Product Performance Monitoring DocumentMonitoring Document

11

Visibility, Probability of Rainfall Visibility, Probability of Rainfall and Rainfall Potential Products: and Rainfall Potential Products:

Current StatusCurrent Status

Rainfall Probability and Rainfall Potential Rainfall Probability and Rainfall Potential algorithms have been reengineeredalgorithms have been reengineered

Received version 4 internal delivery for the Received version 4 internal delivery for the three productsthree products

Software Reviews will be conducted next Software Reviews will be conducted next weekweek

12

Visibility, Probability of Rainfall Visibility, Probability of Rainfall and Rainfall Potential Products:and Rainfall Potential Products:

Next StepsNext Steps

Internal version 5 deliveriesInternal version 5 deliveries

Integration into the framework and test the Integration into the framework and test the outputsoutputs» Conduct 4 month runs on test data setsConduct 4 month runs on test data sets

Conduct Algorithm Readiness ReviewsConduct Algorithm Readiness Reviews

September 30 deliverySeptember 30 delivery

13

Technical Interchange Technical Interchange Meetings (TIM) with Harris Meetings (TIM) with Harris

and GSPand GSP

Harris/AER is currently implementing the AWG algorithms from Harris/AER is currently implementing the AWG algorithms from the ATBDsthe ATBDs

Harris/AER develop an algorithm design document from which Harris/AER develop an algorithm design document from which the software is developedthe software is developed

Harris/AER ask questions of the AWG for algorithm clarificationHarris/AER ask questions of the AWG for algorithm clarification

AWG answers the questionsAWG answers the questions» Fast turnaround is generally requiredFast turnaround is generally required» AWG has already answered over 1000 questionsAWG has already answered over 1000 questions

This is an iterative processThis is an iterative process14

Baseline TestBaseline TestData Set RedeliveryData Set Redelivery

Through the process of answering the Harris/AER Through the process of answering the Harris/AER questions, occasional bugs are found in the questions, occasional bugs are found in the softwaresoftware

When bugs are fixed, the test data are regenerated When bugs are fixed, the test data are regenerated and redeliveredand redelivered

So far, 17 redeliveries have been made: Lightning, So far, 17 redeliveries have been made: Lightning, Cloud, Aerosol, Soundings, and Imagery baseline Cloud, Aerosol, Soundings, and Imagery baseline algorithm test data setsalgorithm test data sets

15

Baseline Algorithm Baseline Algorithm ValidationValidation

The baseline algorithm product teams are The baseline algorithm product teams are continuing to validate their algorithmscontinuing to validate their algorithms

Routine and deep dive Cal/Val Tools are Routine and deep dive Cal/Val Tools are being developedbeing developed

Documents for the routine Cal/Val Tools will Documents for the routine Cal/Val Tools will be delivered in September 2012be delivered in September 2012

16

Extended Baseline Extended Baseline Algorithm TestingAlgorithm Testing

Continuous data feeds are being set up to process the Continuous data feeds are being set up to process the algorithms on extended data setsalgorithms on extended data sets» Simulated dataSimulated data» SEVIRISEVIRI

Fine tune thresholds, coefficients and look up tablesFine tune thresholds, coefficients and look up tables

These products will be made available to the users for These products will be made available to the users for testing purposestesting purposes

Algorithm upgrades will occur after launchAlgorithm upgrades will occur after launch

17

18

Housekeeping Housekeeping WrapupWrapup

100% Option 2 ATBDs & Algorithm Packages were delivered on 100% Option 2 ATBDs & Algorithm Packages were delivered on September 30, 2011September 30, 2011

Visibility, Probability of Rainfall and Rainfall Potential Algorithm Visibility, Probability of Rainfall and Rainfall Potential Algorithm Readiness Reviews will be completed by September 1, 2011Readiness Reviews will be completed by September 1, 2011

100% Option 2 ATBDs & Algorithm Packages for Rainfall 100% Option 2 ATBDs & Algorithm Packages for Rainfall Potential, Probability of Rainfall, and Visibility will be delivered on Potential, Probability of Rainfall, and Visibility will be delivered on September 30, 2012September 30, 2012

100% Routine Cal/Val Tool ATBDs for the Baseline Algorithms 100% Routine Cal/Val Tool ATBDs for the Baseline Algorithms will be delivered on September 20, 2012will be delivered on September 20, 2012

The Way ForwardThe Way Forward

EnterpriseEnterprise

19

Enterprise ConceptEnterprise Concept

Why the enterprise concept?Why the enterprise concept?» End to end cost savingsEnd to end cost savings» MaintainabilityMaintainability» FlexibilityFlexibility» ScalibilityScalibility

OR…….OR…….

20

Enterprise ConceptEnterprise Concept

Because I am tiredBecause I am tired

Tired of building a new system for every Tired of building a new system for every satellite launchedsatellite launched

Tired of implementing a new system every Tired of implementing a new system every time a new algorithm is transitioned to time a new algorithm is transitioned to operationsoperations

21

OpportunityOpportunity

GOES-R provided an opportunity to implement a new GOES-R provided an opportunity to implement a new system within STAR due to the algorithm requirementssystem within STAR due to the algorithm requirements» One algorithm for each productOne algorithm for each product

– No specific cloud mask for each productNo specific cloud mask for each product

» Product precedenceProduct precedence– Algorithms use prior run algorithms as inputAlgorithms use prior run algorithms as input

» Forward model consistencyForward model consistency

Requirements were put in place to reduce algorithm Requirements were put in place to reduce algorithm development costs and to bring scientific consistency development costs and to bring scientific consistency across productsacross products

22

System RequirementsSystem Requirements

In order to develop and test the algorithms, AWG needed In order to develop and test the algorithms, AWG needed to build a test system that could:to build a test system that could:» Read in and process both geostationary and polar data (SEVIRI, Read in and process both geostationary and polar data (SEVIRI,

GOES and MODIS)GOES and MODIS)

» Run algorithms in precedenceRun algorithms in precedence

» Use common ancillary data setsUse common ancillary data sets

» Use a common forward modelUse a common forward model

» Create one type of output data setCreate one type of output data set

Enterprise type of systemEnterprise type of system» Standardization of programming languages, software, interfaces, Standardization of programming languages, software, interfaces,

libraries and toolslibraries and tools

23

Building the SystemBuilding the System

Ingredients required for designing the systemIngredients required for designing the system» Mulling thoughtsMulling thoughts» Smart people – not me….Smart people – not me….» White boardWhite board» AmbrosiaAmbrosia» MarkersMarkers» AmbrosiaAmbrosia» EraserEraser» AmbrosiaAmbrosia» Understanding spouseUnderstanding spouse

Patience ……Patience ……

24

PatiencePatience

It takes months to put each piece in placeIt takes months to put each piece in place

Software hurdles have to be dealt with along the Software hurdles have to be dealt with along the wayway

Issues have to be dealt with as the pieces are Issues have to be dealt with as the pieces are plugged togetherplugged together

It just takes timeIt just takes time25

Resulting Resulting Processing SystemProcessing System

GOES-R Algorithm Processing FrameworkGOES-R Algorithm Processing Framework» System that runs 57 GOES-R algorithmsSystem that runs 57 GOES-R algorithms» One algorithm for each product (i.e. one cloud mask One algorithm for each product (i.e. one cloud mask

for all products)for all products)– Though new algorithms may be plugged in and Though new algorithms may be plugged in and

tested leading to replacement of algorithmstested leading to replacement of algorithms» Common ancillary data setsCommon ancillary data sets» Common forward modelCommon forward model» Standardized output formatStandardized output format» Inputs both geostationary and polar radiance dataInputs both geostationary and polar radiance data

26

Enterprise yet?Enterprise yet?

At this point, I would not call the Framework At this point, I would not call the Framework an enterprise systeman enterprise system

It is an efficient, scientifically consistent It is an efficient, scientifically consistent processing systemprocessing system

What we are missing is …..What we are missing is …..

27

ContinuityContinuity

28

Algorithm ContinuityAlgorithm Continuity

Achieve continutity by having the same algorithm Achieve continutity by having the same algorithm work on multiple instrumentswork on multiple instruments

For ABI algorithm testing, the cloud mask, phase and For ABI algorithm testing, the cloud mask, phase and height products have been implemented to run on height products have been implemented to run on both SEVIRI and MODIS databoth SEVIRI and MODIS data

The cloud mask, phase and height products created The cloud mask, phase and height products created from MODIS data are used for the aerosol detection from MODIS data are used for the aerosol detection and aerosol optical depth algorithm testingand aerosol optical depth algorithm testing

29

How to AchieveHow to AchieveAlgorithm ContinuityAlgorithm Continuity

Visionary scientistsVisionary scientists» Mike Pavolonis – GEOCAT Mike Pavolonis – GEOCAT

» Andy Heidinger – CLAVR-x/PATMOS-xAndy Heidinger – CLAVR-x/PATMOS-x

» And maybe Jaime Daniels…..And maybe Jaime Daniels…..

They developed offline research systems where They developed offline research systems where they can process multiple types of satellite data they can process multiple types of satellite data through their various algorithmsthrough their various algorithms

They probably just got tired like me…..They probably just got tired like me…..

30

OpportunityOpportunity

Users have requested continuity of NOAA Users have requested continuity of NOAA products between the current satellite products between the current satellite systems and the future systemssystems and the future systems

Continuity of products enables the users to Continuity of products enables the users to prepare for GOES-R productsprepare for GOES-R products

Retrofit state of the art GOES-R algorithms to Retrofit state of the art GOES-R algorithms to work for current satellite instrumentswork for current satellite instruments

31

Current ProjectsCurrent Projects

Upgrade GOES WindsUpgrade GOES Winds» Implement GOES-R cloud mask, cloud phase and cloud height Implement GOES-R cloud mask, cloud phase and cloud height

algorithms for winds due to product procedencealgorithms for winds due to product procedence» GOES-R Framework will be delivered to operations with this projectGOES-R Framework will be delivered to operations with this project

VIIRS VIIRS » Polar WindsPolar Winds» Cloud Mask, Phase, HeightCloud Mask, Phase, Height» Cloud Optical PropertiesCloud Optical Properties» Aerosol Optical DepthAerosol Optical Depth» Aerosol DetectionAerosol Detection» Ice Concentration, Age, and Surface TemperatureIce Concentration, Age, and Surface Temperature

32

Enterprise & ContinuityEnterprise & Continuity

33

Integration / Investment

Bene

fit

Policy and Procedure integration

Organizational and Management Integration

Operations Integration

Systems Integration

Science Integration

Enterprise Management

Gateways (Antennas)

Command and Control Systems

Ingest / Distribution Systems

Product Generation

IT Security

Communications (Networks)

Monitoring, Reporting and Help Desk Services

Configuration Management

Management Controls

Operational Controls

Technical Controls

Examples Include:•OSO/OSDPD ReorganizationFuture activities:•Acquisition Integration•Business Reference Model

Examples Include:•Common Help Desk Services•Common operations staffFuture activities:•Common mission management staff

Examples Include:•Merged data/products•Common toolsFuture activities:•Merged algorithms

Examples Include:•Technical Reference ModelsFuture activities:•Consolidate backup

Examples Include:•ESPC, METOP, NDEFuture activities:•JPSS/GOES-R

Examples Include:•Product Distribution and Access (PDS) Subsystem (Legacy, NDE, GOES-R)Future activities:•Common ingest subsystem

Examples Include:•Multi-mission antennas Future activities:•Exostrategies Study

Examples Include:•Common Communication Element for IJPSFuture activities:•[TBD]

Examples Include:•RATS/CATS•CWSFuture activities:•[TBD]

Examples Include:•PoliciesFuture activities:•System Acquisition•Engineering Principles

Future activities:•Network Monitoring

Future activities:•Consolidated Backup

Future activities:•CCSDS/SOA

NOAA Enterprise Architecture Towards Integrated Algorithm and

Production Generation

STAR Enterprise STAR Enterprise ApproachApproach

STAR has implemented a Framework that STAR has implemented a Framework that enables an enterprise approach for enables an enterprise approach for developing algorithmsdeveloping algorithms

STAR scientists have developed algorithm STAR scientists have developed algorithm software that may process data from multiple software that may process data from multiple instruments to bring product continuity to the instruments to bring product continuity to the usersusers

35

SummarySummary

HousekeepingHousekeeping – – STAR is preparing to delivery the final 3 algorithms and is STAR is preparing to delivery the final 3 algorithms and is redelivering datasetsredelivering datasets

OpportunityOpportunity – – Developing algorithms for multiple satellites has provided Developing algorithms for multiple satellites has provided and opportunity for an enterprise systemand opportunity for an enterprise system

EnterpriseEnterprise – – STAR has implemented a system for an enterprise approach STAR has implemented a system for an enterprise approach for algorithm developmentfor algorithm development

ContinuityContinuity – – STAR has designed algorithms to process data from multiple STAR has designed algorithms to process data from multiple satellites to bring product consistencysatellites to bring product consistency

AmbrosiaAmbrosia – – Makes things happenMakes things happen

36