将空间自我相关与泛线性模式结合_第1页
将空间自我相关与泛线性模式结合_第2页
将空间自我相关与泛线性模式结合_第3页
将空间自我相关与泛线性模式结合_第4页
将空间自我相关与泛线性模式结合_第5页
已阅读5页,还剩27页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

1、Incorporating spatial autocorrelation into the general linear model with an application to the yellowfin tuna (Thunnus albacares) longline CPUE data 將空間自我相關與泛線性模式結合,並應 用在黃鰭鮪魚延繩釣的CPUE數據資料中。指導老師 王勝平 學生 賴冠宇FrameResultsMethodsIntroduction DiscussionCPUEStandardCase studyStandard-GLMStandard-GLM/HBMSpati

2、al-GLMSpatial-GLM/HBM Evaluation of the spatial-GLMIntroductionCPUEn Often been utilized to obtain a relative index of the abundance of a fish stock.n Affected by changes of year, season, area of fishing and various environmental factors.StandardMany statistical methods have been used to standardize

3、 them to account for such variations.n generalized linear models (standard- GLM) the most commonly usedn general additive models (GAM)n neural networks (NN)n regression trees (RT)The spatial autocorrelation brings a potentially major problem in standardizing the CPUEs by use of the standard-GLM, GAM

4、, NN or RT.ProblemStandardSolutionHBM Standard-GLM = Standard-GLM/HBM Habitat-based models Spatial autocorrelationStandard-GLM/HBM = Spatial-GLM/HBMMore accurate Standardized CPUEs Relative indices of the abundance of the fish Case studyWe use these models to analyze data on the CPUE of yellowfin tu

5、na (Thunnus albacares) of the Japanese longline fisheries in the Indian Ocean.Finallyn Evaluate the results of the spatial-GLMsMethodsIntroductionn standard-GLM and spatial-GLM were used to analyze the Japanese yellowfin tuna longline CPUE data in the Indian Ocean.n Each method was employed with and

6、 without HBM. The data were detailed in Nishida et al. (2003) and summarized in Table 1.MethodsMethodsStandard-GLMThe standard-GLM without HBM is of the form:Standard-GLMStandard-GLMnFollowing Bigelow and Nishida:HBF(the number of hooks between two floats) was divided into six classes: 56、79、1011、12

7、15、1620、2125. n The Japanese tuna longline fisheries with 34 HBF Swordfish. Excluded the data from this type of gear.n Also added the effects : SST (sea surface temperature ) TD (thermocline depth ) to the model. May have affected the distribution and abundance of yellowfin tuna.Standard-GLM/HBMStan

8、dard- GLM/HBM :Spatial-GLMEq. (1) can be written as:The assumption of independent CPUE or in the standard-GLM, Eq. (3), is obviously violated for a fish population, for it ignores the spatial autocorrelation in the many features of the population.After all, fish move together with a positive spatial

9、 dependency. The more closely in space the observations are made, the more similar they are.Spatial-GLMWith cov(i, j) = ij 0, i is not equal to j:Spatial-GLMSpatial-GLMSpatial-GLM/HBMLRT(likelihood ratio test) used for the spatial-GLM can be applied to the effective CPUE, as defined in Eq. (2), to y

10、ield the spatial-GLM/HBM. For both spatial-GLM and spatial-GLM/HBM, the spatial autocorrelation structure was modeled as covariograms, as defined in Eq.(5)In the spatial-GLM, three distancerelated parameters (sill, range and nuggets) were estimated, along with those in the standard-GLM. A case study

11、Four GLMs (standard-GLM, standard-GLM/HBM, spatial-GLM and spatial-GLM/HBM) were used to analyze yellowfin tuna CPUE data (19582001) of the Japanese longline fisheries in the Indian Ocean, together with some environmental factors.In the spatial-GLM and spatial-GLM/HBM, four distance models (Gaussian

12、, exponential, linear, and spherical) were also examined.ResultsMethodsIntroductionResultsOf the four distance models, the Gaussian model had the best goodness-of-fit. The ResultsResultsResultsResultsResultsResultsResultsMethodsIntroduction Discussion Evaluation of the spatial-GLMAlthough the tempor

13、al trends in the CPUEs from the spatial-GLMs did not differ greatly from those from the standard-GLMs, the spatial-GLMs are preferred for analyzing the CPUE data on yellowfin tuna, especially if there is strong spatial autocorrelation among the data.This is because the spatial-GLMs took account of t

14、he spatial autocorrelation effectively and yielded more realistic estimates of the variances. This is not surprising, especially considering the semivariograms and covariograms from the standard-GLMs (Fig. 3).A covariogram is a function of the distance between data points that measures how strong th

15、eir spatial autocorrelation is. A positive spatial autocorrelation manifests itself in a decrease with distance to zero at some distance, where observations are no longer autocorrelated.The likelihood ratio tests (LRT) for spatial independence of H0: r = 0 showed that the test statistics were highly significant in all cases (Table 3).The

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论