由于适宜水禽栖息的生境面积不断缩小,致使湿地生态系统受到严重破坏,生物种群数量急剧下降.通过野外采点调查建立可靠的栖息地适宜点群.基于遥感和GIS空间技术获取20个评价因子,并结合二项Logistic回归模型,构造出基于像元水平的栖息地适宜性模型,经过筛选后保留对适宜性影响力较大、拟合优度高的10个评价因子,分别从社会人文、植被长势和地形地貌三大方面分析.再应用K-means聚类分析法将适宜性点群分级,得到建三江区栖息地适宜性评价分级图,有效分析适宜性5个等级的空间尺度,其中最适宜栖息,中适宜栖息,基本适宜栖息,不适宜栖息和不可用地分别占总用地的24.73%,17.69%,29.8%,17.62%和10.16%,前三类占总的72.22%.说明建三江区的适宜水禽栖息的用地较多,主要分布在流域、低河漫滩等人为干扰少的湿地沼泽平原,此研究对水禽生境的保护和可持续发展都具有指导意义.

waterfowls habitat and population quantity has been seriously damaged due to continuously deteriorating of Wetland ecosystem. There are satisfactory variety of ecosystems, and there are resources and abundant plants or anmi als species in the Jiansanjiang area.Although the biodiversity of the region is rich and mi portant, it is suffering from manmade destroy and disturbance. Analysisofhigh strengthwetland reclamation process and environmentchangeswas conduc-ted to study the directand indirecteffects onwet land waterfowlshabitat. Peoplewere raised to pay attention to the protec-tion of Wetland and Biodiversity extensively. Habitatsuitabilitymethod iswidely applied in natural land resourcemanage-ment, especially it isverymeaningful research forbiodiversityprotection and use fulspacesquestion ofanmi als. Ithas also very mi portant significance to quickly realize and determine the besthabitatposition in spatial scale domain.Jiansanjiang area lies in the northwest of Heilongjiang Province in China, as formostof the year, where the clmi ate is characterized as temperate semiarid continentalmonsoon. There are three large rivers in the region, including Heilongjiang River, SonghuaRiver andWusuliRiver, which are mi portantnationalwetland in Sanjiang Plain. Themaps of topographic and vegetation distribution in Jiansanjiang area following as referencematerials, it is an application ofGPS on field control pointposition established habitat suitability point cluster. The acquisition of 21 single evaluation factors is based on information extraction by remote sensing andGeography Information System spatial analysismethod, and analysis ofBinaryLogistic regressionmode,l whichwas integrated into habitat suitabilitymodule on pixel levels. Alongwith other factors, thesewas a great influence, high goodness-of-fitand big standardized residuals on 10 single suitability evaluation factors finally, synthetically evaluated by settlementdensity, population density, road Euclid Distance, settlementEuclid Distance, NDVI, vegetation types, wetness, texture, reliefandDEM etc. These factors emphasize three aspects: humanity society, vegetation growth topography and geomorphology respectively. There is integrated classification criterion of Jiansanjiang habitat with K-means cluster analysis, effectively analyzed five levels of suitability on spatial scale. As a result,the percentage of the highly suitable habitat, suitable habitat,moderately suitable habitat, not suitable habitat and permanently not suitable habitat respectively are 24.73%, 17.69%, 29.8%, 17.62% and 10.16%, thatof the former three typeswas about72.22% of the tota,l thosewhowilluse evidence to support their view that the abundant land of waterfowls suitable habitat in Jiansanjiang. The effective strategiesbased on the evaluation results can be put forward as the reference for protection ofhabitat suitability land and sustainable reproduction ofpopulation.The expermi ental results show thatourmethod achieveshigh fusion accuracy (general classification accuracy: 87.3%,kappa’s coefficient: 0.852), it gives a proof of grade separation mi age and actual ecological environment is very smi ilar.The paper offered an innovation thought: application of statistical Binary-Logisticmodule based on pixel level inRS miage.Thismethod can be helpful for the large-scale natural ecological landscape planning, determination of human disturbance and anmi al protection scale domain, biodiversity sustainable development for future geographic spatial analysis.