泛第三極大數(shù)據(jù)系統(tǒng)

泛第三極大數(shù)據(jù)系統(tǒng)BigDataSystemforPan-ThirdPole標(biāo)題：祁連山區(qū)域1985-2017年30m土地覆蓋分類產(chǎn)品數(shù)據(jù)集（V1.0）Title：LandCoverDatasetinQilianMountainAreafrom1985to2017(V1.0)摘要：本數(shù)據(jù)集包括祁連山區(qū)域1990年至2017年每5年一期的30m土地覆蓋分類產(chǎn)品。該產(chǎn)品首先利用Landsat-8/OLI構(gòu)造2015年時(shí)間序列數(shù)據(jù)，針對(duì)各類地物隨時(shí)間變化呈現(xiàn)的NDVI時(shí)間序列曲線不同，對(duì)不同地物特征進(jìn)行知識(shí)歸納，設(shè)定提取規(guī)則不同地物信息，得到2015年的土地覆蓋分類圖。分類系統(tǒng)參考了IGBP分類系統(tǒng)和FROM_LC分類系統(tǒng)，共分為耕地、林地、草地、灌叢、濕地、水體、不透水面、裸地、冰川和積雪共10大類。由GoogleEarth高清影像和實(shí)地調(diào)研數(shù)據(jù)進(jìn)行精度評(píng)價(jià)，得出2015年土地覆蓋分類產(chǎn)品的總體精度高達(dá)92.19%。以2015年的土地覆蓋分類產(chǎn)品為底圖，按各類別的比例選取大量樣本，基于GoogleEarthEngine平臺(tái)的Landsat系列數(shù)據(jù)和強(qiáng)大地?cái)?shù)據(jù)處理能力，利用深度學(xué)習(xí)的思想，選取隨機(jī)森林分類器，對(duì)波段信息和NDVI、MNDWI、NDBI等指數(shù)進(jìn)行樣本訓(xùn)練，生產(chǎn)出1985-2017年每5年一期的土地覆蓋分類產(chǎn)品。對(duì)2套2015年的分類產(chǎn)品進(jìn)行比較，得出基于GoogleEarthEngine平臺(tái)生產(chǎn)的土地覆蓋分類產(chǎn)品與基于時(shí)間序列方法得到的分類產(chǎn)品具有很好的一致性。總之，祁連山核心區(qū)的土地覆蓋數(shù)據(jù)集具有較高的總體精度，且基于GoogleEarthEngine平臺(tái)樣本訓(xùn)練的方法能夠在時(shí)間和空間上對(duì)現(xiàn)有的分類產(chǎn)品進(jìn)行擴(kuò)展，每5年一期的頻次能夠在長(zhǎng)時(shí)間序列上反映更多的土地覆蓋類型變化信息。Abstract：ThisdatasetcontainslandcoverproductsinQilianMountainAreafrom1985to2017every5years.Thedatasetwasproducedbytwosteps.Firstly,landcoverproductin2015isproducedusingtimeseriesLandsat-8/OLIdata.InviewofthedifferentNDVItimeseriescurvesofvariouslandfeatureswithtimevariation,theknowledgeofdifferentlandfeaturesissummarized,theextractionrulesofdifferentlandfeaturesareset,andthelandcoverclassificationmapin2015isobtained.TheclassificationsystemreferstoIGBPandFROM_LCclassificationsystem.Itisdividedinto10categories:cultivatedland,woodland,grassland,shrub,wetland,waterbody,impermeablesurface,bareland,glacierandsnowcover.AccordingtotheaccuracyevaluationofGoogleEarthhigh-definitionimageandfieldsurveydata,theoverallaccuracyoflandcoverclassificationproductsin2015isashighas92.19%.Secondly,takingthelandcoverclassificationproductsin2015asthebasemap,alargenumberofsamplesareselectedaccordingtotheproportionofdifferenttypes.BasedontheLandsatseriesdataandpowerfuldataprocessingabilityofGoogleEarthEngineplatform,therandomforestclassifierisselectedtotrainthebandinformationandNDVI,MNDWI,NDBIandotherindicesbyusingtheideaofin-depthlearning.Thelandofeachfive-yearperiodfrom1985to2017isproduced.Bycomparingtwoclassifiedproductsin2015,itisconcludedthatthelandcoverclassifiedproductsbasedonGoogleEarthEngineplatformhavegoodconsistencywiththosebasedontimeseriesmethod.Inconclusion,thelandcoverdatasetinthecoreareaofQilianMountainshashighoverallaccuracy,andthemethodbasedonsampletrainingofGoogleEarthEngineplatformcanexpandtheexistingclassificationproductsintimeandspace,andthefrequencyofeveryfiveyearscanreflectmorelandcovertypechangeinformationinlongtimeseries.縮略圖祁連山區(qū)域1985年土地覆蓋分類產(chǎn)品LandCoverProductinQilianMountainAreain1985數(shù)據(jù)文件命名方式和使用方法：文件命名約定：YYYY_QiLianShan_WholeArea_LC30.tif （YYYY：年）數(shù)據(jù)版本號(hào)：V1.0投影：Geographic數(shù)據(jù)格式：GeoTIFF,66777行×40598列填充值：0FileNamingConventionanddatadescription：FileNamingConvention:YYYY_QiLianShan_WholeArea_LC30.tif(YYYY:year)DataVersion：V1.0Projection：+proj=longlat+datum=WGS84+no_defsDataFormat:GeoTIFF,40850rows×17023columesFilledValue：0文件大小：數(shù)據(jù)集~40.4GB，單個(gè)文件~5.04GB時(shí)間范圍:1985–2017年資助項(xiàng)目：泛第三極環(huán)境變化與綠色絲綢之路建設(shè)專項(xiàng)（XDA20100101）數(shù)據(jù)來(lái)源：該數(shù)據(jù)集是利用了GoogleEarthEngine平臺(tái)提供的Landsat系列數(shù)據(jù)：/Datasource：ThedatasetwasproducedusingLandsatseriesdatainGoogleEarthEngineplatform:/空間范圍（東、西、南、北經(jīng)緯度）：93°E-104°E，35°N-40°N時(shí)間分辨率：5年空間分辨率：30m數(shù)據(jù)引用文獻(xiàn)/Citation：Zhong,B.,Yang,A.,Nie,A.,Yao,Y.,Zhang,H.,Wu,S.,&Liu,Q.(2015).Finerresolutionland-covermappingusingmultipleclassifiersandmultisourceremotelysenseddataintheHeiheriverbasin.IEEEJournalofSelectedTopicsinAppliedEarthObservationsandRemoteSensing.8(10),4973-4992.Zhong,B.,Ma,P.,Nie,A.,Yang,A.,Yao,Y.,Lü,W.,&Liu,Q.(2014).LandcovermappingusingtimeseriesHJ-1/CCDdata.ScienceChinaEarthSciences,57(8),1790-1799.參考文獻(xiàn)/Reference：Huang,H.,Chen,Y.,Clinton,N.,Wang,J.,Wang,X.,&Liu,C.,etal.(2017).MappingmajorlandcoverdynamicsinBeijingusingallLandsatimagesingoogleearthengine.RemoteSensingofEnvironment,S0034425717300810.數(shù)據(jù)使用聲明：為尊重知識(shí)產(chǎn)權(quán)、保障數(shù)據(jù)作者的權(quán)益、擴(kuò)展數(shù)據(jù)中心的服務(wù)、評(píng)估數(shù)據(jù)的應(yīng)用潛力，請(qǐng)數(shù)據(jù)使用者在使用數(shù)據(jù)所產(chǎn)生的研究成果中（包括公開(kāi)發(fā)表的論文、論著、數(shù)據(jù)產(chǎn)品和未公開(kāi)發(fā)表的研究報(bào)告、數(shù)據(jù)產(chǎn)品等成果），