题名

以邊緣萃取技術探討濁水溪揚塵潛在發生區位劃定之研究

并列篇名

Delineating Potential Areas of Aeolian Dust Occurrence with Edge Detection Techniques in the Jhuoshei River

DOI

10.29417/JCSWC.201106_42(2).0003

作者

邱郁瑄(Yu-Hsuan Chiu);莊智瑋(Chin-Wei Chuang);林昭遠(Chao-Yuan Lin)

关键词

揚塵潛在發生區位 ; 揚塵 ; 邊緣萃取 ; potential areas of aeolian dust occurrence ; aeolian dust ; edge extraction

期刊名称

中華水土保持學報

卷期/出版年月

42卷2期(2011 / 06 / 01)

页次

107 - 119

内容语文

繁體中文

中文摘要

濁水溪河口因地質鬆軟、土壤乾燥,旱季時水位下降,河床裸地遽增,適逢東北季風來臨,細小泥砂易藉由風力揚起,導致大氣懸浮微粒濃度過高,對鄰近居民之健康造成嚴重威脅,如何有效及快速對揚塵潛在發生區位進行劃定極為重要。另濁水溪每年河道流路變化甚大,且退水期間農墾區位時常變動,致使河川揚塵潛在發生區位需逐年更新,加上光譜解析度不足,傳統影像分類無法將衛星影像中之揚塵潛在發生區位及整地初期瓜田加以區隔,實有賴其他影像處理技術進行劃定,供河川管理機關儘早擬定揚塵抑制對策有其必要。本研究以揚塵最為嚴重之濁水溪自強大橋至西濱大橋為研究樣區,利用各土地利用之影像梯度差異進行邊緣萃取,並藉由面積門檻值給定,進一步萃取出揚塵潛在發生區位。本研究選用Sobel,Laplacian,Prewitt及Canny邊緣偵測技術等進行探討,以建立河川揚塵潛在區位劃定之技術,供日後揚塵治理參考之用。

英文摘要

The estuary of the Jhuoshei River has geological fracturing and dry soils. Therefore, in dry seasons, the increasing amount of bare land and the northeast monsoons cause smaller particle matter to be easily eroded. This causes serious air pollution and seriously threatens the health of nearby residents. Effectively and rapidly delineating potential areas of aeolian dust occurrence is very important. Variations in the flow paths of river channels are frequently governed by annual storms, and land used to grow melons is also changed annually. Therefore, annual delineation of potential areas of aeolian dust is necessary. Because of the lack of spectral resolution, traditional image classification cannot clearly delineate potential areas of aeolian dust occurrence and melon land. Thus, delineating potential areas by other image processing techniques for the reference of related authorities to adopt dust control countermeasures is a vital issue. This study focused on the dust vulnerable segment (Bridge Ziqiang to Xibin) of the Jhuoshei River using image gradient derived from each land use and assigning an area threshold to delineate potential areas of aeolian dust occurrence. There are several methods for edge detection. The correct selection of suitable edge detection method is very important. The Sobel, Laplacian, Prewitt, and Canny methods were chosen for discussion in this study. The techniques for delineating the potential areas of aeolian dust occurrence could be then established for the reference of aeolian dust management in the future.

主题分类 生物農學 > 農業
生物農學 > 森林
生物農學 > 畜牧
生物農學 > 漁業
生物農學 > 生物環境與多樣性
工程學 > 土木與建築工程
工程學 > 市政與環境工程