基于多季相遙感信息的三江平原濕地信息提取
發(fā)布時間：2018-05-04 08:48
本文選題：面向?qū)ο蟮姆椒?/strong> + 多季相Landsat�。� 參考：《中國科學(xué)院研究生院(東北地理與農(nóng)業(yè)生態(tài)研究所)》2013年碩士論文
【摘要】：本文以中國東北地區(qū)三江平原為研究對象，利用研究區(qū)2000年、2006年和2012年多季相Landsat TM/ETM+影像，建立了濕地分類系統(tǒng)，結(jié)合野外調(diào)查數(shù)據(jù)，應(yīng)用多尺度分割算法，根據(jù)影像的豐富信息和物候、時相等特征，采用面向?qū)ο蟮姆诸惙椒�，進行三江平原的濕地分類。且利用遙感解譯數(shù)據(jù)，逐步分析了研究區(qū)2000-2012年濕地的覆被空間特征，并分析了濕地類型的變化特征。結(jié)果表明： 1.2000年、2006年和2012年濕地提取的生產(chǎn)者精度和用戶精度都高于85%，總體精度也都高于85%。面向?qū)ο蟮姆诸惙椒軌蛴行Ю眠b感影像提供的豐富信息，產(chǎn)生較高的分類精度。由于濕地分布破碎復(fù)雜，且中分辨率影像地物邊界模糊，影像分割時，尺度太大難以準確反映濕地分布狀況和邊界，分割尺度過小則不利于對象的提取和空間信息的利用。對于中分辨率遙感影像不同分割尺度的組合使用有利于濕地信息的提取。 2.多季相影像數(shù)據(jù)的使用，改善了季相變化濕地的分類效果和精度，要獲得較高的濕地分類精，降雨量多寡月份的選取和物候規(guī)律的把握是關(guān)鍵。針對特定的年份，最好選取當年降水最大和最小的月份，且結(jié)合能夠明顯區(qū)分植被季相信息月份的遙感影像。利用面向?qū)ο笫褂弥蟹直媛识嗉具b感影像，在大區(qū)域范圍內(nèi)細致且高精度的提取濕地信息是一個成本較低且行之有效的辦法。 3.以季節(jié)變化為時間尺度，同年各季相遙感影像為解譯目標，發(fā)現(xiàn)降水較多季節(jié)和降水較少季節(jié)之間影像監(jiān)測的濕地轉(zhuǎn)化最為劇烈，主要為濕地類型到非濕地類型之間的轉(zhuǎn)化，濕地類型內(nèi)部間的轉(zhuǎn)化和非濕地類型到濕地類型的轉(zhuǎn)化所占面積比例均很小。季節(jié)性的濕地轉(zhuǎn)化主要發(fā)生在草本沼澤和草地之間。降水、氣溫及植被長勢等變化因素對濕地面積季相變化及轉(zhuǎn)化起到很大作用。 4.三江平原濕地主要分布在平原、低洼地區(qū)。主要人工濕地為水田，其余人工濕地面積比例較少；主要天然濕地類型為草本沼澤，其次為河流，其他濕地類型所占面積比例較小。濕地分布主要在保護區(qū)、河岸以及邊遠國界周圍，而且分布較為集中。研究區(qū)2000，2006和2012年各階段濕地覆被變化明顯，，主要發(fā)生在濕地和非濕地之間，濕地內(nèi)部轉(zhuǎn)化次之。濕地總體面積穩(wěn)定，近十年三江平原草本沼澤農(nóng)田化進程趨緩。
[Abstract]:This paper takes Sanjiang Plain in Northeast China as the research object, using the multi-season Landsat TM/ETM images in 2000, 2006 and 2012, establishes the wetland classification system, combines the field survey data, and applies the multi-scale segmentation algorithm. According to abundant information, phenology and time-equal features of images, the wetland classification of Sanjiang Plain is carried out by using object-oriented classification method. The spatial characteristics of wetland cover and the variation characteristics of wetland types in the study area from 2000 to 2012 were analyzed by means of remote sensing interpretation data. The results show that: 1. In 2000, 2006 and 2012, the producer and user accuracy of wetland extraction was higher than 850.The overall precision was also higher than that of 85kyam. The object-oriented classification method can effectively utilize the abundant information provided by remote sensing images and produce higher classification accuracy. Because the distribution of wetland is complicated and the boundary of middle resolution image is blurred, it is difficult to accurately reflect the distribution and boundary of wetland when image segmentation is too large. The small scale of segmentation is not conducive to the extraction of objects and the use of spatial information. The combination of different segmentation scales for middle resolution remote sensing images is beneficial to the extraction of wetland information. 2. The use of multi-season image data improves the classification effect and precision of seasonal wetland. To obtain higher wetland classification precision, the selection of rainfall month and the grasp of phenological regularity are the key. For a specific year, it is better to select the month with the largest and the smallest precipitation in the year, and combine the remote sensing image which can distinguish the vegetation seasonal information month obviously. It is a low cost and effective method to extract wetland information carefully and accurately in large area by using object oriented multi-season remote sensing image with medium resolution. 3. With seasonal variation as time scale and remote sensing images of different seasons as the interpretation target in the same year, it was found that the transformation of wetlands monitored by image between more and less precipitation seasons was the most intense, mainly the transformation from wetland type to non-wetland type. The area ratio of the transformation between the wetland types and the non-wetland types to the wetland types is very small. Seasonal wetland conversion occurs mainly between herbaceous swamps and grasslands. Precipitation, air temperature and vegetation growth play an important role in the seasonal change and transformation of wetland area. 4. Sanjiang plain wetland mainly distributes in plain, low-lying area. The main artificial wetland is paddy field, the other constructed wetland area proportion is less, the main natural wetland type is herbaceous swamp, the next is river, the other wetland type accounts for a small proportion of the area. Wetlands are mainly distributed in protected areas, riverbanks and around remote borders, and the distribution is relatively concentrated. The changes of wetland cover in 2006 and 2012 were obvious, mainly occurred between wetland and non-wetland, followed by internal transformation of wetland. The total area of wetland is stable, and the farmland process of herbaceous swamp in Sanjiang Plain has slowed down in recent ten years.
【學(xué)位授予單位】：中國科學(xué)院研究生院(東北地理與農(nóng)業(yè)生態(tài)研究所)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2013
【分類號】：P237
【參考文獻】
相關(guān)期刊論文 前9條
1 李建平;張柏;張泠;王宗明;宋開山;;濕地遙感監(jiān)測研究現(xiàn)狀與展望[J];地理科學(xué)進展;2007年01期
2 汪愛華,張樹清,何艷芬;RS和GIS支持下的三江平原沼澤濕地動態(tài)變化研究[J];地理科學(xué);2002年05期
3 王根緒,程國棟;荒漠綠洲生態(tài)系統(tǒng)的景觀格局分析——景觀空間方法與應(yīng)用[J];干旱區(qū)研究;1999年03期
4 ;Analysis of net primary productivity of terrestrial vegetation on the Qinghai-Tibet Plateau,based on MODIS remote sensing data[J];Science China(Earth Sciences);2012年08期
5 宮鵬;牛振國;程曉;趙魁義;周德民;虢建宏;梁璐;王曉風;李丹丹;黃華兵;王毅;王坤;李文寧;王顯威;應(yīng)清;楊鎮(zhèn)鐘;葉玉芳;李展;莊大方;遲耀斌;周會珍;閆軍;;中國1990和2000基準年濕地變化遙感[J];中國科學(xué):地球科學(xué);2010年06期
6 牛振國;張海英;王顯威;姚文博;周德民;趙魁義;趙惠;李娜娜;黃華兵;李叢叢;楊軍;柳彩霞;劉爽;王琳;李展;楊鎮(zhèn)鐘;喬飛;鄭姚閩;陳炎磊;盛永偉;高小紅;朱衛(wèi)紅;王文卿;王紅;翁永玲;莊大方;劉紀遠;羅志才;程曉;郭子琪;宮鵬;;1978～2008年中國濕地類型變化[J];科學(xué)通報;2012年16期
7 傅伯杰;呂一河;陳利頂;蘇常紅;姚雪玲;劉宇;;國際景觀生態(tài)學(xué)研究新進展[J];生態(tài)學(xué)報;2008年02期
8 趙其國;高俊峰;;中國濕地資源的生態(tài)功能及其分區(qū)[J];中國生態(tài)農(nóng)業(yè)學(xué)報;2007年01期
9 周德民;宮輝力;胡金明;趙魁義;;三江平原淡水濕地生態(tài)系統(tǒng)景觀格局特征研究——以洪河濕地自然保護區(qū)為例[J];自然資源學(xué)報;2007年01期
本文編號：1842384
本文鏈接：http://sikaile.net/kejilunwen/dizhicehuilunwen/1842384.html
最近更新
教材專著
