用于湿地气候效应模拟的三江平原下垫面数据获取

Acquirement of Underlying Surface Data of the Sanjiang Plain for Simulations of Wetland Climate Effect

三江平原是我国沼泽湿地的重要分布区。在20世纪的后50 a中,三江平原在经过几次大规模的开荒后,其下垫面发生了明显的阶段性变化,湿地大面积消失,农田面积不断增加。为了认识20世纪下半叶不同时期三江平原湿地的气候效应,拟采用第三代区域气候模式(RegCM3)来开展研究。而目前被广泛用于RegCM3的GL-CC数据,所反映的研究区下垫面几乎不存在沼泽湿地,这与事实严重不符,所以必须获取三江平原有沼泽湿地存在的真实的下垫面数据,才能得到可靠的湿地气候效应的模拟结果。分别选择了4期(1954年、1985年、1996年和2000年)有阶段代表性的航片、TM影像数据为主要数据源,利用ERDAS8.7和Arcview3.2等软件,首先得到4期研究区的土地利用数据,然后通过土地利用数据类型向模式数据类型的转换,从4期研究区土地利用数据中提取出了三江平原下垫面的模式数据。经过误差分析,认为获取的4期三江平原下垫面模式数据准确的反映出研究区几个时期的湿地下垫面实况,可以作为三江平原湿地气候效应模拟的基础数据。

The study of climate effect of wetland is one of focuses of wetland field. RegCM3 （the third generation RegCM） is chosen for the research of climate effect of wetland in the Sanjiang Plain. At present, terrain data inputted RegCM3 are mainly from GLCC （Global Land Cover Characteristics） database. Data in the database are derived from lkm AVHRR NDVI composites spanning a 12 - month period from April 1992 to March 1993. It is obviously that the GLCC data cannot meet the requirement to simulate the climate of a certain region under different surface situations during various periods, especially for the region of a greater change such as the Sanjiang Plain. The Sanjiang Plain （45°01′N - 48°27′N, 130°13 ′E - 135°05′26＂E） is an important region distributing marshes in China. In 1949, the wetlands＇ area accounted for 80.2% of the area of the plain part of the Sanjiang Plain. Because wetlands have been losing since 1949 and their landscape fragments in a high degree at present, the situations of the underlying surface have been changing during various periods in the study region. Data of the aircraft photo on July 12, 1954, TM （Thematic Mapper） images on September 5, 1985, September 21, 1996, September 20, 2000, GLCC （ Global Land Cover Characteristics） database and field investigation material of the region were chosen as basic data sources in the study based on facts of wetlands and climate changes of the study region in the second haft of 20 century. The results calculated according to data of 4 phases chosen showed that the areas of wetlands accounted for 51.4%, 18.2%, 14.3% and 13.9% and ones of paddy field were 1%, 9. 8%, 7.1% and 17.1% of the plain part of the region respectively. ,So 4 years chosen were representative of undisturbed phase and, phases after large - scaled reclamation and of paddy field increase. On the plane of soft wares of ERDAS 8.7 and Arcview3.2, the land cover interpretation was based on extensive use of computer - assisted image processing and supervised methods and the classification process was not automated entirely and more closely resembles a traditional manual image interpretation philosophy. Considering extensive applicability of data and real reflection of the underlying surface, the preliminary data of land cover of the region were obtained using TM3, TM4 and TM5 bands＇ composites based on classification system of land use in China. The composites of TM3, TM4 and TM5 just wetlands left were overlapped with the ones of TMS, TM6 and TM7 to get the boundaries for acquirement of accurate data of wetlands. Then, 4 phases＂ grid data and maps of land use of the study region were obtained. The simulation domain is 42°N -50°N, 127°30′E - 136°30′E and center point of the domain is （46°N,132°E）. The number of level grid points is 109 ×97 and level resolution is 5′ ×5′. The classification system of land use in China include 6 types of first class and 25 types of second class, however, the one of land cover of terrain data of RegCM3 only has 20 classes. First, 25 types of second class of land use need to be turned into 20 classes of land cover of RegCM3 according to LCCS （ Land Cover Classification System） standards. In the Sanjiang Plain, 19 types of land use were turned into 12 classes of land cover, which were irrigated crop, crop, evergreen needle leaf tree, mixed woodland, deciduous broad leaf tree, deciduous shrub, forest/field mosaic, bog or marsh, short grass, tall grass, inland water, water and land mixture. There are 1194 grid points based on spatial resolution of 5′ × 5′ . The areas of various classes inside each grid point were calculated. If area of the class were the maximum in whole point, a certain class would be defined as the class of the point. Finally, 4 phases＇ grid model data of land cover for simulations of climate effect of wetlands of the Sanjiang Plain were obtained.