Publications

Detailed Information

Diagnosis of tropical cyclone track patterns in WRF using the fuzzy c-means clustering method : 퍼지 군집분류 방법을 이용한 WRF 모의태풍 진로 군집 분석

DC Field Value Language
dc.contributor.advisor허창회-
dc.contributor.author홍선영-
dc.date.accessioned2017-07-19T08:51:04Z-
dc.date.available2017-07-19T08:51:04Z-
dc.date.issued2013-02-
dc.identifier.other000000010307-
dc.identifier.urihttps://hdl.handle.net/10371/131369-
dc.description학위논문 (석사)-- 서울대학교 대학원 : 지구환경과학부, 2013. 2. 허창회.-
dc.description.abstractThis study investigated the variability of simulated tropical cyclone (TC) tracks over the western North Pacific (WNP) from the regional climate model (RCM) to compare to the observed TC tracks using the fuzzy clustering analysis. To understand variability of TC tracks specifically, TC tracks are classified to six representative patterns with fuzzy c-means clustering method.
The optimum cluster number is determined by cross checking four objective validity measures. First, 519 observed TC tracks formed from 1982 to 2010 during July to October are classified to six clusters and 549 simulated TC tracks are classified based on the clusters for observed TC tracks. These clusters are described as TCs landfalling countries of East Asia (e.g., East China, Taiwan, Korea and Japan) (C1), TCs affecting Japan with long trajectories (C2), early recurving tracks passing the east of Japan (C3), TCs moving the easternmost region over the WNP (C4), TCs over the South China Sea (SCS) with short straight trajectories (C5) and TCs moving across the Philippines with straight trajectories (C6).
In comparison between observed track clusters and simulated clusters, C1 and C2 have remarkable differences in the percentage (number). The C1 of simulated TC tracks decreased significantly from 19.5% (101) to 13.1% (72) and C2 increased most from 15% (79) to 24% (132). These differences in percentage of C1 and C2 lead to the variability in spatial distribution of TC genesis frequency and track density. For the simulation, TC genesis frequency decreased in the Philippine Sea (PS), where the genesis region of TCs for C1 and increased in the southeastern part of the WNP where the TCs for C2 are mainly formed. The track density of simulated TC is lower than observation in the PS and East China Sea but higher in the eastern part of WNP. The other clusters (C3–6) are successfully simulated in the percentage and spatial distribution.
These significant discrepancies in C1 and C2 can be explained by large-scale environments. The membership coefficient weighted composites for the TC genesis day are conducted to investigate the relationship between variability of dominant track patterns (e.g., C1 and C2) and large-scale circulation. The C1 is the cluster related to the La Niña and simulated TCs of C1 are formed less because low level cyclonic wind is underestimated in the PS and vertical wind shear is overestimated than those of observation. But for C2, El Niño related cluster, the genesis of simulated TCs are enhanced by stronger positive relative vorticity and monsoon trough extending to the genesis region (10°–20°N, 145°–160°E). Moreover, simulated TCs of C2 recurve earlier because of eastern retreating for the North Pacific subtropical High (NPSH) than observation. Therefore, simulated TC tracks of C2 are steered more eastward than tracks of observation.
The results show that the WRF model has good performance for simulating TC tracks but distinctive variability in C1 and C2. This study is expected to improve the ability of regional climate model further.
-
dc.description.tableofcontentsTable of Contents

Abstract
Table of contents
List of Figures
List of Tables
1. Introduction
2. Data and method
2.1. Data
2.2.WRF model and experimental design
2.2.1. Model configuration
2.2.2. Climatology of simulated TCs
2.3. Pattern Classification of TC track patterns
2.3.1. Classifying TC tracks with fuzzy clustering method
3. Track patterns of simulated TC
3.1. Spatial distribution of track patterns
3.2. Mean properties
3.3. The genesis frequency difference of clusters
3.4. Interannual and seasonal variability
3.5. TC landfalls
4. Relationship with Large-scale environments
4.1. Relationship between C1 and large-scale circulation
4.1.1. SST and low-level circulations
4.1.2.Vertical wind shear
4.2. Relationship between C2 and large-scale circulations
4.2.1. SST and low-level circulations
4.2.2. The monsoon trough
4.2.3.Steering circulations
5. Summary and Discussion
Appendix
The algorithm of Fuzzy c-means Clustering Method
Initialization of TC tracks for fuzzy clustering
Iterative method for minimizing fuzzy c-means functional
Validity scalar measure for the optimum cluster number
Reference
국문초록
-
dc.formatapplication/pdf-
dc.format.extent1489286 bytes-
dc.format.mediumapplication/pdf-
dc.language.isoen-
dc.publisher서울대학교 대학원-
dc.subject.ddc550-
dc.titleDiagnosis of tropical cyclone track patterns in WRF using the fuzzy c-means clustering method-
dc.title.alternative퍼지 군집분류 방법을 이용한 WRF 모의태풍 진로 군집 분석-
dc.typeThesis-
dc.contributor.AlternativeAuthorHong, sunyoung-
dc.description.degreeMaster-
dc.citation.pages87-
dc.contributor.affiliation자연과학대학 지구환경과학부-
dc.date.awarded2013-02-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share