科技论文写作 程伟.docx

科技论文写作 程伟.docx

《科技论文写作 程伟.docx》由会员分享，可在线阅读，更多相关《科技论文写作 程伟.docx（10页珍藏版）》请在冰豆网上搜索。

科技论文写作程伟

RESEARCHPROPOSAL

Title:

Estimationofsoilmoistureusingpassivemicrowaveandactivemicrowaveremotesensingdatainaheterogeneouslandscape

Name:

WeiCheng

StudentNo.:

SA13008134

Department:

Schooloflifeandscience

Degreesought:

Master

Supervisor:

XiaomingFeng

Date:

2014.6.8

Tableofcontents

Abstract:

1

Keyword:

1

Introduction2

Literaturereview3

3.1Soilmoistureinversionmodel3

3.2Basisofanestablishedmodel5

Methodology6

4.1Thevegetationindex6

4.2Selectionoftheremotesensingdata6

4.3Apolynomialmodel7

4.4Datacollection7

4.5Dataanalysis7

4.6Limitation8

Timeframe8

Outcomes9

Reference9

Abstract:

Soilmoisture（SM）isacriticalparameterinthehydrologicalcycleandclimateresearch.Duetothecomplexityofregionalecosystems,thereisabiglimitationforfieldmeasurementsofsoilmoisture.Inthispaper,soilmoistureisretrievedbyusingpassivemicrowaveandactivemicrowaveremotesensingdataintheheterogeneouslandscapelevel.EstimatedSMvalueswerethencorrelatedwithinsituSMmeasurementsandtheirrelationshipswerestatisticallyanalyzed.Resultsindicatedstatisticallysignificantcorrelationsbetweenthem,whichexhibitsthepossibilitytoestimateSMfromremotesensingdata.Therefore,SMcanbeenretrievedbycombinewithactiveandpassiveremotesensingonalargescaleheterogeneouslandscape.

Keyword:

soilmoisture;remotesensing;heterogeneouslandscape;insitumeasurement;

Introduction

Landsurfaceiscoveredbyavarietyofvegetationcoverandwaterbody,whichresultsinRegionalecologicalsystemcomplexity.Ecosystemisinfluencedbymanydifferentfactors,whichthusincreasestothecomplexityoftheecosystemandleadtoecosystemstabilityindynamicequilibrium.Inregionalecosystem,soilmoisture（SM）isofimportanceforecosystembalance.Infact,soilmoistureisakeyvariableinland–atmosphereinteractions,sounderstandingofsoilmoisturespatial–temporalvariabilityisoneofthemostimportantissuesinmanyscientificdisciplines,suchasenvironmentalscience,agronomy,atmosphericscience.Therefore,theSMresearchisahotspotinmanyscientificfields.

Duetotheimportanceofsoilmoisture,soitiscriticaltomeasuresoilmoisturebyusingvariousdifferentmethods.Traditionally,SMisusuallymeasuredbyinsitumeasurementsandfieldmeasurements.Thesemethodsarethemostaccuratemethodsforestimatingsoilmoisture,butisexpensive,time-consumingandlabor-intensiveandonlyprovidespointmeasurements.Inaddition,theseconventionalmeasurementscannotmeettheneedsoflarge-scaleandlong-termmeasurementsofsoilmoisture.Therefore,technologicaladvancesinsatelliteremotesensinghaveofferedanalternativetotheseconventionalmeasurementsofSMandenabledustomonitoritathighertemporalandspatialresolutionsatlowercostandlesstime.Inthepaper,usingremotesensingmethodretrievesSMonalargescaleheterogeneouslandscape.

Inaheterogeneouslandscape,ecosystemiscomposedofmanydifferentelements,likewaterbody,vegetation,soilandsoon.Theseelementswillexperiencedifferentchangesinenvironment.Thesephysical,chemicalandbiologicalchangeprocessestakingplaceatthelandsurfacestronglyimpacttheamountofwaterstoredwithintheuppersoillayers.Therefore,inordertoreduceotherfactors’influenceinSMinversion,theappropriateinversionmodelshouldalsobeenestablishedwhenusingremotesensingdataestimatesSM.

Inthepaper,themainpurposeistocombinewithactiveandpassiveremotesensingtoretrievesoilmoisturebyapolynomialmodelandverifyitsreliabilitywhencomparingwithinsitumeasurement.Assuredly,moredetailedaimswecanacquireareasfollows:

1Firstly,selectingasuitableinversionmodelisofimportance,anditshouldincludeasmuchasafactor,suchasvegetation,soiltype.

2Thesefactorsshouldbeenanalyzedtofindouttheextentoftheirinfluenceonsoilmoisturemeasurementandmakethemorder.

3Usingremotesensingdata,SMisestimatedbytheestablishedmodelonalargescaleheterogeneouslandscape.

4Comparingwithinsitumeasurement,thebias,standarddeviation,rootmeansquareerror（RMSE）andcorrelationcoefficientarecomputedandstatisticallyanalyzed

5Limitationsofremotesensingmethodtoestimatesoilmoistureandmodelsareanalyzed.

Literaturereview

3.1Soilmoistureinversionmodel

SoilmoistureinversionmodelisofimportancewhenusingremotesensingdatatoretrieveSM,whichcanenhancetheaccuracyofSMinversion.Soilmoistureinversionmodelisanalgorithm,whichcontainsvariousfactorsaffectingsoilmoisturechange.BinFangetal[1]SMathighspatialresolutioniscriticalforstudyingvariousland-airboundaryinteractions’process.However,currentlytheresolutionofpassivemicrowaveretrievedsoilmoistureislow-around25km（SMOSandAMSR-E）.Tosolvethisproblem,asoilmoisturedisaggregationalgorithmbasedonthermalinertiarelationshipbetweendailytemperaturechangeandaveragesoilmoisturemodulatedbyvegetationconditionshasbeenformulated.Thealgorithmcontainssurfacetemperatureandvegetationindex.Althoughtheresultsofthisapproachareveryencouraging,mismatchofthepixelsizeamongthedatasetsusedinthisstudyandtheaccuracyofthedisaggregationalgorithmvariesindifferentseasons.

T.Lacavaetal[2]foundearth'semittedradiationmeasuredfromsatellite（usuallygivenintermsofbrightnesstemperature,BT）stronglydepends,inthemicrowavespectralregion,ontheemissivityand,atalowerextent,onsurfacetemperaturevariations.Inthisspectralregionwaterandsoilhaveverydifferentdielectricpropertieswhichstronglyaffectemissivity.Remotesensingcanreceivethoseemissivity,andcanbeusedinAMSU-basedsoilwetnessindex（SWI）togetSM.Themethodusedinthecatchmentgetagoodresult,buttheAMSU-basedsoilwetnessindicescannotyetbevalidatedinotherdifferentgeographicallocations.Similarly,SWIwasalsoutilizedbychristophPauliketal[3].ButdataheusedwasfromASCAT.

J.A.Sobrinoetal[4]usedAirborneHyperspectralScanner（AHS）andASTERdatatocalculateSM.Thecorrelationbetweenthesurfacetemperature,theNormalizedDifferenceVegetationIndex（NDVI）andtheemissivitywasestablishedbyapolynomialtoretrieveSM.However,themethodhaslimiteduseovermoredenselyvegetatedcrops.

Soilmoistureestimationisalsoagrowingtendencytowardintelligence.Thus,SajjadAhmadetal[5]proposedanovelregressiontechniquecalledSupportVectorMachine（SVM）basedonstatisticallearningtheorythatusesahypothesisspaceoflinearfunctionsbasedonKernelapproach.ThestrengthofSVMliesinminimizingtheempiricalclassificationerrorandmaximizingthegeometricmarginbysolvinginverseproblem.Buttheeffectsofsurfaceroughnessandtopographywerenottakenintoaccountinthemodel.

ApartfromsurfaceSM,NildaSánchezetal[6]consideredprofilesoilmoisture（0-100cmdepth）calculatedbytheFAO56methodology.Buttheresultshowedthatsomewaterdoesremainforacertainperiodoftime,duetheparticularcharacteristicseitherofthesoilprofilecharacteristicsorthevegetationcover.

Bothpassiveandactivemicrowaveremotesensingexistshortcomingsandadvantages.QinLietal[7]utilizedtheirstrength,andthuscombinedpassiveandactivemicrowaveremotesensingwasusedtoretrievesoilmoisture.Theestablishedmodels（THRAandTWRA）includedthebrightnesstemperature（BT）,backscatteringcoefficient（BSC）andSurfaceroughness.TheresultindicatedthatTWRAisbetterthanTHRA,butTWRAismoresuitableforbaresurfaceorlowvegetationsurfaces,whileTHRAdoesnothavethislimitation.

ParinazRahimzadeh-Bajgiranetal[8]estimatingsoilmoisture（SM）wasbasedonevaporativefraction（EF）retrievedfromoptical/thermalinfraredMODISdata.EFmodelusingtheremotelysensedlandsurfacetemperature（Ts）/vegetationindexconceptwasmodifiedbyincorporatingNorthAmericanRegionalReanalysis（NAAR）TadataandusedforSMestimation.Theresultindicatedtheaccuracyofthepredictionswasconsiderablybetterforintermediatesoilmoisturevalues,butthemodeltoaccountforextremeconditionswasstillimproved.

SMindices（likeNSMIandSMGM）canachieveahighaccuracyfornon-vegetationinfluencedsoilsamples,buttheiraccuracyislimitedincaseofthepresenceofvegetation.Since,theincreaseofthevegetationcoverleadstonon-linearvariationsoftheindices.SocalibrationofinfluenceofvegetationiscriticalinSMinversionmodel.D.Spengleretal[9]usedhyperspectralartificial3D-canopymodelstocorrecttheinfluenceofvegetationonsurfacesoilmoistureindices.Whenuptoavegetationcoverof75%,thecorrectionfunctionminimizetheinfluencesofvegetationcoversignificantly.Ifthevegetationisdenserthemethodleadstoinadequatequalitytopredictthesoilmoisturecontent.

3.2Basisofanestablishedmodel

Previousstudiesestablishedavarietyofmodelstoretrievesoilmoisturebyusingremotesensingdata,whichexhibittheirstrengthandweakness.Themodelscontainingasmuchasafactorisbetter.WeshouldestablishamodelwhichincludesthosefactorssignificantlyaffectinginSMestimation.Sothemodelshouldincludethesurfacetemperature,thevegetationIndexandtheemissivity.

Methodology

Inaheterogeneouslandscape,soilmoisturestronglyaffectstheecosystembalance,especiallyaridandsemi-aridregions（liketheLoessPlateau）.Sofortheseregions,SMresearchisessential.Usingremotesensingdataestimatessoilmoistureonalargescalerange.Weselectthemodelincl