Advanced 3 StaCsCcal Meteorology

5
METR 5433 Spring 2018 Time: T/Th 2:30 - 3:45 PM Room: NWC 5930 Data analysis is a rou-ne part of many types of research in the atmospheric sciences. As such, having the right set of tools and the prowess on how to use those tools is an important part to understanding the behavior of the climate system. This course offers an overview of some advanced sta4s4cal methods used to interpret data in the atmospheric and oceanic sciences. It is designed to be an applied course: i.e., the goal is to gain a working knowledge of the sta4s4cal tools most commonly used in the atmospheric sciences. Major topics to be covered include: (A) regression/ correla4on and epoch analyses; (B) 4me series analysis (e.g., power spectra, filtering, wavelet analysis); (C) matrix methods for signal decomposi4on (e.g., EOFs, CCA); and (D) objec4ve mapping and covariance modeling. The course is intended for graduate students and senior undergraduates (with permission). Although previous knowledge of probability and sta4s4cs is preferable, a short review of sta4s4cal measures will be provided. You will also need a working knowledge of a soOware package to analyze data (e.g., Python, MATLAB, IDL, NCL, etc.). This will be important because of the highly applied nature of the course. Course DescripCon Instructor Dr. Jason C. Furtado Office: Na4onal Weather Center (NWC), Rm. 5240 Phone: 405.325.1391 Email: [email protected] Office Hours: By appointment. Advanced StaCsCcal Meteorology

Transcript of Advanced 3 StaCsCcal Meteorology

Page 1: Advanced 3 StaCsCcal Meteorology

METR5433

Spring2018Time:T/Th2:30-3:45PMRoom:NWC5930

Dataanalysisisarou-nepartofmanytypesofresearchintheatmosphericsciences.Assuch,havingtherightsetoftoolsandtheprowessonhowtousethosetoolsisanimportantparttounderstandingthebehavioroftheclimatesystem.

Thiscourseoffersanoverviewofsomeadvancedsta4s4calmethodsusedtointerpretdataintheatmosphericandoceanicsciences.Itisdesignedtobe

anappliedcourse:i.e.,thegoalistogainaworkingknowledgeofthesta4s4caltoolsmostcommonlyusedintheatmosphericsciences.Majortopicstobecoveredinclude:(A)regression/correla4onandepochanalyses;(B)4meseriesanalysis(e.g.,powerspectra,filtering,waveletanalysis);(C)matrixmethodsforsignaldecomposi4on(e.g.,EOFs,CCA);and(D)objec4vemappingandcovariancemodeling.

Thecourseisintendedforgraduatestudentsandseniorundergraduates(withpermission).Althoughpreviousknowledgeofprobabilityandsta4s4csispreferable,ashortreviewofsta4s4calmeasureswillbeprovided.YouwillalsoneedaworkingknowledgeofasoOwarepackagetoanalyzedata(e.g.,Python,MATLAB,IDL,NCL,etc.).Thiswillbeimportantbecauseofthehighlyappliednatureofthecourse.

CourseDescripCon

Instructor

Dr.JasonC.FurtadoOffice:Na4onalWeatherCenter(NWC),Rm.5240Phone:405.325.1391Email:[email protected]

OfficeHours:Byappointment.

AdvancedStaCsCcalMeteorology

Page 2: Advanced 3 StaCsCcal Meteorology

RequiredTextThere isnorequiredtext for theclass.Mostof theclasswillbetaughtwithmyownpersonalnotes.However,thereareseveralsourcesandtextsthatwillbeuseful(Textbookswillbeavailableonreserveoryoucanpurchasethem):

• Dennis Hartmann class notes on objec4ve analysis: hap://www.atmos.washington.edu/~dennis/552_Notes_Op.html

• DiscreteInverseandStateEs0ma0onProblems-CarlWunsch,CambridgePress• Sta0s0calMethodsintheAtmosphericSciencesSecondEdi0on-DanielWilks,AcademicPress• Sta0s0calAnalysisinClimateResearch-HansvonStorchandFrancisW.Zwiers

CourseWebPageThewebpagewillbeaccessibleviahaps://canvas.ou.edu(logonusingyourOU4+4).Thereyouwillfind course materials (e.g., class notes, assignments, and examples), grades, and other news andannouncementsaboutthecourse.

GradingHomeworkAssignments: 70%FinalProject: 30%

Homework Assignments. Therewill be about 6-7 homework assignments throughout the semester.HomeworkassignmentsmustbetypedandelectronicallysubmiaedthroughCanvas.Allplotsincludedwith your assignment should have proper units, labels, colorbars, and informa4ve cap4ons. Thesehomeworkassignmentsareintendedforyoutoapplytheknowledgeyoulearninthecoursedirectlytodata (either synthe4c or real). Some4mes, I will allow you to subs4tute your own research data (ifapplicable) in lieuoftheprovideddatatocompleteaproblemintheassignment.Thissubs4tu4onisdonesothatyouhaveachancetoactuallyseehowtoapplythesetechniquestoyourownresearchwork.Youmayworkwithothersontheassignments,butyoumustturninyourownwork.

FinalProject.Thefinalprojectwillbeapaperandoralpresenta4oninwhichyoumustuseoneormoresta0s0cal techniques learned in the course to answer a real research ques4on. The project is to bechosenbasedonasetofques4onsthatyouwouldliketoanswerratherthanthetypeofdataanalysistechniqueyouwould like toapply. Youwillbe required to submitanabstractof yourwork forpriorapproval.Moredetailswillbeprovidedinclass.

1. Applysta4s4caltheorydirectlytoreal-worldmeteorologicalandclimatedatatodiscernspa4otemporalcharacteris4cs.

2. Gainfurthersta4s4cal(anddynamical)understandingoftheenvironmentalsystemofinterestandanswerrealworldresearchques4onsintheatmosphericsciences.

3. Cri4callyevaluatejournalar4clesandresearchpresenta4onswhichemploythesetechniques.

4. Developapersonaldata“toolkit”ofsta4s4calmethodsthatcanbeappliedtoyourownresearchproblemsandtasks.

GOALS

Page 3: Advanced 3 StaCsCcal Meteorology

CompuCngAmaingoalofthecourseistohaveyouworkwithdatausingcomputersoOwarepackagesanddevelopyour own “sta4s4cal toolbox” for later use. All studentswho do not have a School ofMeteorology(SoM) computer account may obtain one from Shawn Riley (NWC 5640). Python and MATLAB arereadilyavailable foruseon theMetLabworksta4ons.Python isopen-sourceandcanbe installedonyour own machine. NOTE: You are free to use whatever soOware package with which you feelcomfortable.IwillprimarilyusePythoninthiscourseforin-classexamples,solu4ons,etc.Ifthereareques4onsorissueswithaccesstosoOware,pleaseseemeduringthefirstweekofclass.

CourseStyleTheoverallstructureoftheclasswillconsistoflectures,bothtradi4onalandinterac4ve,coveringthemajor topics. I will also present examples in class of using the actual techniques to analyzemeteorological and climate data. Many examples will be in Jupyter Notebooks (i.e., Python-basedinterac4veexamples).Ques4onsandinterac4onsduringclassarewelcomeandhighlyencouraged. Ifyoudon’taskques4onswhenthingsareunclear,thenneitherofusbenefitfromclassroomlecture.

• Arrivetoclasson4meandpreparedtolearn.

• Submitassignments4mely.Nolatesubmissionsareallowedwithoutpriorapproval.

• TakeanacCveroleinthelearningprocessandaskquesConswhenneeded.

• Seekassistancefrommeifyoudonotunderstandthematerialorneedhelpwithanassignment.

• Becourteoustootherstudents.Placeallphonesonvibrate/silence,donottext/usesocialmediaduringclass,andkeeptalkingtoaminimum.

EXPECTATIONSOFTHESTUDENT

Page 4: Advanced 3 StaCsCcal Meteorology

ReasonableAccommodaConPolicy TheUniversityofOklahomaiscommiaedtoprovidingreasonableaccommoda4onforallstudentswithdisabili4es.Studentswithdisabili4eswhorequireaccommoda4oninthiscoursearerequestedtospeakwithmeassoonaspossible.Studentswithdisabili4esmustberegisteredwiththeDisabilityResourceCenterprior to receivingaccommoda4ons in thiscourse.TheDisabilityResourceCenter is located inUniversityCommunityCenter(730CollegeAve).Phone:405.325.3852.E-mail:[email protected]

AcademicMisconductChea4ngisstrictlyprohibitedattheUniversityofOklahoma.Simplyput, itdevaluesyourdegreeandendsupmarringyourcharacterandreputa4on.Forspecificdefini4onsonwhatcons4tuteschea4ng,review the Student’sGuide toAcademic Integrity at hap://integrity.ou.edu/students.html. If you arecaughtchea4ng,Iamobligatedtoreportit.Sanc4onsforacademicmisconductincludeexpulsionfromtheUniversityandanFinthiscourse.BOTTOMLINE:Don’tcheat-it’snotworthit.

To be successful in this class, all workmust be yours and yours alone. Youmay work together onhomeworkassignments,butyoumustsubmityourownoriginalworkforgrading.

ReligiousHolidaysOU policy is to excuse absences of students that result from religious observances and to providewithoutpenalty for thereschedulingofexamina4onsandaddi4onal requiredclassworkthatmayfallonreligiousholidays.Anystudentwhohasareligiousholidayfallonadayanassignmentisdue,pleaseseemenolaterthanoneweekbeforethedeadlinesoastomakeotherarrangements.

TitleIXResourcesandReporCngRequirementFor any concerns regarding gender-based discrimina4on, sexual harassment, sexual assault, da4ng/domes4cviolence,orstalking,theUniversityoffersavarietyofresources.Tolearnmoreortoreportanincident,pleasecontacttheSexualMisconductOfficeat405.325.2215orsmo@ou.edu.Incidentscanalsobereportedconfiden4allytoOUAdvocates(405.615.0013)24hoursaday,7daysaweek. Pleasebeadvisedthataprofessor/GA/TAisrequiredtoreportinstancesofsexualharassment,sexualassault,ordiscrimina4ontotheSexualMisconductOffice. Inquiriesregardingnon-discrimina4onpoliciesmaybe directed to: Bobby J. Mason, University Equal Opportunity Officer and Title IX Coordinator [email protected],pleasevisithap://www.ou.edu/eoo.html.

AdjustmentsforPregnancy/ChildbirthRelatedIssuesShouldyouneedmodifica4onsoradjustments toyour course requirementsbecauseofdocumentedpregnancy-relatedorchildbirth-related issues,pleasecontactmeortheDisabilityResourceCenterat405.325.3852assoonaspossible.Also,pleaseseehap://www.ou.edu/eoo/faqs/pregnancy-faqs.htmlforanswerstocommonlyaskedques4ons.

Page 5: Advanced 3 StaCsCcal Meteorology

FinalProjectPresentaConswillbedoneduringthefinalweekofclassesandduringtheFinalExamPeriod.

FINALEXAMPERIOD:FRIDAY,MAY11,20181:30-3:30PM

***NOCLASS:Thursday,Feb15(AGUOceanSciencesMee4ng)

CourseOutline

I. FundamentalStaCsCcs+LeastSquaresMethods(a) Reviewoffundamentalsta4s4calmeasures/Sta4s4caltests(b) Composite/Epochanalysis(c) SignificanceTes4ng(d) Correla4ontheory/Regressionandcorrela4onanalysis/Mul4-variateregression(e) Applica4onsofregression/correla4ontheory(e.g.,covariancemodeling)

II. MatrixMethods(a) Linearalgebrareview(b) Empiricalorthogonalfunc4ons(EOFs)/principalcomponentanalysis(PCA)(c) Extendedandmul4variateEOFs(d) Maximumcovarianceanalysis(MCA)&canonicalcorrela4onanalysis(CCA)

III. TimeSeriesAnalysis(a) Autocorrela4on(b) Harmonicanalysis,powerspectralanalysis,andsignificancetes4ngforspectralpeaks(c) Cross-spectralanalysis(d) Filteringandfilterdesigns-Bestprac4cestouse

IV. AddiConalTopics(asCmeallows)(a) Objec4veMapping/Kriging(b) Waveletanalysis