| The research regarded Jiangsu Dongtai city Dongtai forest farm as research zone,under help of UAV high image and floor investigation data,firstly through technology which based on objects to divide tree canopy and extract number of trees;secondly establish elements’ regression model like crown width,age and DBH,and use diameter distribution model to evaluate forest storage volume;at last compare different methods for evaluating accuracy of storage volume.The research got some main conclusions are as follows:1.Utilize multi-scale division,face to object category technology’s forest crown width extraction to UAV high image,it got favorable extraction effects.The best division scale of poplar crown within research zone is 8~20,tight and shape index is 0~1,bright characteristic,contiguity index,shape index are available to extraction of forest crown width,and the best image range is sample site scale.Making accuracy comparison of actual measurement of forest crown width and image extraction crown width,the result without obvious difference.2.Establish poplar actual measurement DBH and regression model of average canopy’s image extraction,its model R2 is 0.75.Make crown width model calculation’s storage volume as assisted element,actual storage volume as main element to make poplar storage volume double aggression evaluation,its double aggression evaluation’s formula R2 is 0.87,the calculated poplar storage volume of research zone is 97113.69m3,its evaluation section is 94910.86~103441.14m3,evaluated accuracy is 95.70%.3.Through age element,utilize forest division age and forest division average DBH to establish aggression model to evaluate forest division storage volume,poplar age-DBH model R2 is 0.703 in research zone,got forest number tress through UAV high resolution image,calculate forest division storage volume of every small forest group,and got poplar total storage volume in research zone after accumulation is 103507.2m3,after using image to extract crown width element was added,two linear regression model R2 was improved to 0.766,utilize age and crown width two linear model to calculate poplar storage volume in research zone,the evaluated result was 98610.9m3.4.Classify small forest group in research zone according to their density and age,fitting diameter distribution to different types of forest division,the distribution model is Weibull distribution,normal distribution,Logistic distribution,through χ2 test to verify fitting result,and found that acceptance rate of Weibull distribution model is 100%;acceptance rate of normal distribution model is 25%;but acceptance rate of Logistic distribution model is 8.3%,Weibull distribution has greatly optimized fitting to pure poplar forest in research zone.Use different types of Weibull distribution parameter,substitute every model into counterpart small poplar group,combine poplar number of trees which extracted by crown width,seek the all diameter distribution of small poplar forest,finally calculated poplar storage volume in Dongtai forest farm is 93980.17m3 through unitary volume table.5.Use storage volume which got from sample site investigation of every small group,seek small group’s storage through area method,accumulate all of the small group’s storage volume as total storage volume in research zone,its value is 96141.85m3.Regard storage volume of small group in research zone as benchmark,compare poplar storage volume accuracy by different methods,the results indicated that when use double aggression sampling to evaluate can get the smallest difference,that is 1.03%,evaluation method of age-DBH linear model appeared largest difference,that is 7.66%,the order of evaluation difference from small to large is double aggression sampling,diameter distribution model,two-linear aggression model,age-DBH linear model.The research result provided various referential methods for investigation of UAV storage volume,which decreased many workloads compare with traditional forest investigation. |
PDF Full Text Request