简介：

简介：ＣＬＯＳＩＮＧＨＩＬＬＳＦＯＲＲＥＦＯＲＥＳＴＴＩＯＮ──ＡｎＥｆｆｅｃｔｉｖｅＷａｙｔｏＲｅｓｔｏｒｅａｎｄＩｎｃｒｅａｓｅＦｏｒｅｓｔｒｙＰｌａｎｔＣｏｖｅｒＴｏ“ｃｌｏｓｅｈｉｌｌｓ”（ｓａｎｄｐｌｏｔｓ，ｓｈｏａｌｓ）ａｎｄａｌｌｏｗｎａｔｕ...

简介：

简介：Remote-sensingdataforprotectedareasinnorthernTogo,obtainedinthreedifferentyears(2007,2000,and1987),wereusedtoassessandmapchangesinlandcoverandlanduseforthisdroughtpronezone.Thenormalizeddifferencevegetationindex(NDVI)wasappliedtotheimagestomapchangesinvegetation.Anunsupervisedclassification,followedbyclassesrecoding,filtering,identifications,areacomputingandpost-classificationprocesswereappliedtothecompositeofthethreeyearsofNDVIimages.Maximumlikelihoodclassificationwasappliedtothe2007image(ETM+2007)usingasupervisedclassificationprocess.Sevenvegetationclassesweredefinedfromtrainingdatasets.Thesevenclassesincludedthefollowingbiomes:riparianforest,dryforest,floodedvegetation,woodedsavanna,fallows,parkland,andwater.Fortheseclasses,theoverallaccuracyandtheoverallkappastatisticfortheclassifiedmapwere72.5%and0.67,respectively.Dataanalysesindicatedagreatchangeinlandresources;especiallybetween1987and2000probablyduetotheimpactofdemocratizationprocesssocial,economic,andpoliticaldisorderfrom1990.Wide-scalelossofvegetationoccurredduringthisperiod.However,areasofvegetationclearingandregrowthweremorevisiblebetween2000and2007.Themainsourceofconfusioninthecontingencymatrixwasduetoheterogeneitywithincertainclasses.Itcouldalsobeduetospectralhomogeneityamongtheclasses.Thisresearchprovidesabaselineforfutureecologicallandscaperesearchandforthenextmanagementprograminthearea.

简介：WemappedtheforestcoverofKhadimnagarNationalPark(KNP)ofSylhetForestDivisionandestimatedforestchangeoveraperiodof22years(1988-2010)usingLandsatTMimagesandotherGISdata.SupervisedclassificationandNormalizedDifferenceVegetationIndex(NDVI)imageclassificationapproacheswereappliedtotheimagestoproducethreecoverclasses,viz.denseforest,mediumdenseforest,andbareland.Thechangemapwasproducedbydifferencingclassifiedimageriesof1988and2010asbeforeimageandafterimage,respectively,inERDASIMAGINE.Errormatrixandkappastatisticswereusedtoassesstheaccuracyoftheproducedmaps.Overallmapaccuraciesresultingfromsupervisedclassificationof1988and2010imagerieswere84.6%(Kappa0.75)and87.5%(Kappa0.80),respectively.Forestcoverstatisticsresultingfromsupervisedclassificationshowedthatdenseforestandbarelanddeclinedfrom526ha(67%)to417ha(59%)and105ha(13%)to8ha(1%),respectively,whereasmediumdenseforestincreasedfrom155ha(20%)to317ha(40%).ForestcoverchangestatisticsderivedfromNDVIclassificationshowedthatdenseforestdeclinedfrom525ha(67%)to421ha(54%)whilemediumdenseforestincreasedfrom253ha(32%)to356ha(45%).BothsupervisedandNDVIclassificationapproachesshowedsimilartrendsofforestchange,i.e.decreaseofdenseforestandincreaseofmediumdenseforest,whichindicatesdenseforesthasbeenconvertedtomediumdenseforest.Areaofbarelandwasunchanged.Illicitfelling,encroachment,andsettlementnearforestscausedthedenseforestdeclinewhileshortandlongrotationplantationsraisedinvariousyearscausedtheincreaseinareaofmediumdenseforest.ProtectivemeasuresshouldbeundertakentocheckfurtherdegradationofforestatKNP.

简介：Background:Remotesensing-basedinventoriesareessentialinestimatingforestcoverintropicalandsubtropicalcountries,wheregroundinventoriescannotbeperformedperiodicallyatalargescaleowingtohighcostsandforestinaccessibility(e.g.REDDprojects)andaremandatoryforconstructinghistoricalrecordsthatcanbeusedasforestcoverbaselines.Giventheconditionsofsuchinventories,thesurveyareaispartitionedintoagridofimagerysegmentsofpre-fixedsizewheretheproportionofforestcovercanbemeasuredwithinsegmentsusingacombinationofunsupervised(automatedorsemi-automated)classificationofsatelliteimageryandmanual(i.e.visualon-screen)enhancements.Becausevisualon-screenoperationsaretimeexpensiveprocedures,manualclassificationcanbeperformedonlyforasampleofimagerysegmentsselectedatafirststage,whileforestcoverwithineachselectedsegmentisestimatedatasecondstagefromasampleofpixelsselectedwithinthesegment.Becauseforestcoverdataarisingfromunsupervisedsatelliteimageryclassificationmaybefreelyavailable(e.g.Landsatimagery)overtheentiresurveyarea(wall-to-walldata)andarelikelytobegoodproxiesofmanuallyclassifiedcoverdata(sampledata),theycanbeadoptedassuitableauxiliaryinformation.Methods:Thequestionishowtochoosethesampleareaswheremanualclassificationiscarriedout.Wehaveinvestigatedtheefficiencyofone-per-stratumstratifiedsamplingforselectingsegmentsandpixels,wheretocarryoutmanualclassificationandtodeterminetheefficiencyofthedifferenceestimatorforexploitingauxiliaryinformationattheestimationlevel.Theperformanceofthisstrategyiscomparedwithsimplerandomsamplingwithoutreplacement.Results:OurresultswereobtainedtheoreticallyfromthreeartificialpopulationsconstructedfromtheLandsatclassification(forest/nonforest)availableatpixellevelforastudyarealocatedincentralItaly,assumingthreelevelsoferrorratesoftheuns

简介：Arbuscularmycorrhizafungi(AMF)arevitalintheregenerationofvegetationindisturbedecosystemsduetotheirnumerousecologicaladvantagesandthereforearegoodindicatorsofsoilandecosystemhealthatlarge.Thisstudywasaimedatdetermininghowtheseasonal,vegetationcoverdensity,edaphicandanthropogenicfactorsaffectAMFrootcolonization(RC)andsporedensity(SD)inDesa’adryAfromontaneforest.AMFRCandSDintherhizosphereoffivedominantwoodyspecies,Juniperusprocera,Oleaeuropaea,Maytenusarbutifolia,CarissaspinarumandDodonaeaangustifoliagrowinginDesa’aforestwerestudiedduringtherainyandthedryseasonsinthreepermanentstudyvegetationcoverdensityplots(dense,medium,andpoor).Eachplot(160×40m2)hastwomanagementpractices(fencedandunfencedplots)ofarea.A100gsampleofrhizospheresoilfrommoisturefreecompositesoilwasusedtodeterminesporedensity.Sporedensityrangedfrom50to4467spores/100gsoil,andallspecieswerecolonizedbyAMFwithinarangeof4–95%.Glomuswasthedominantgenusintherhizosphereofallspecies.VegetationcoverdensitystronglyaffectedSDandRC.TheSDwassignificantlyhigher(p<0.05)inthepoorvegetationcoverdensitythanintheothertwoandlowestinthedensecover;rootcolonizationshowedthereversetrend.Managementpracticessignificantly(p<0.05)influencedAMFSDandRC,withthefencedplotsbeingmorefavoured.Seasonssignificantly(p<0.05)affectedRCandSD.MoreRCandSDwereobservedinthewetperiodthanthedryperiod.CorrelatingAMFSDandRCwithsoilphysicalandchemicalpropertiesshowednosignificantdifference(p>0.05)exceptfortotalnitrogen.Disturbance,vegetationcoverdensity,seasonandtotalnitrogenaresignificantfactorsthatcontrolthedynamicsandmanagementinterventionstomaintaintheforesthealthofdryAfromontaneforests.