摘要
面对日益加剧的人口-资源-环境问题,全球变化研究成为近年来国际上最为活跃的研究领域之一。随着研究的深入,人们认识到由人类活动所导致的土地利用/土地覆被变化(LUCC)是引起生态环境和气候变化的主要驱动力,同时又对人类社会的可持续发展产生重大的影响。正因为LUCC在全球变化和社会可持续发展中的重要地位,自20世纪90年代以来,许多全球变化研究计划都将LUCC研究作为主要研究内容。山东省是一个经济大省,近年来经济发展与土地利用间的矛盾日见凸显,通过对山东省土地利用格局及其驱动因素的研究,可及时、准确地了解土地利用变化情况,监测土地利用变化趋势、布局及规模,掌握建设用地和耕地的使用情况,为城市建设提供基础数据,为国土资源调查、规划、管理、保护与合理利用服务。
以中分辨率遥感数据源MODIS归一化植被指数(NDVI)年度序列数据为基础,叠加高程、夜间灯光数据,采用非监督分类与监督分类相结合的方法,对山东省土地利用进行分类。用典型相关、多元回归等方法对土地利用格局及空间决定要素、土地利用动态及驱动要素进行了研究。
2001年土地利用分类总精度为84.8%,Kappa系数为0.83。城市建筑用地、林业用地、耕地分别占全省总面积的2.13%,23.37%,52.53%。分类结果与《山东植被》和统计数据基本吻合。
在NDVI时间序列的基础上叠加夜间灯光强度,对于城市建筑用地的识别非常有效,其生产者精度、用户精度分别为87.01%、95.71%。
一年两熟旱作和双季稻NDVI的季节动态曲线为双峰型,其余的土地利用类型NDVI的季节动态曲线为单峰型。
单因素方差分析表明,夜间灯光强度在土地利用类型间的分布存在显著差异。夜间灯光强度以建筑用地最高,城乡交错带次之。
各土地利用类型的分布高程均存在显著差异,其中以混交林分布平均海拔最高,芦苇沼泽、盐生草甸、水体分布平均海拔最低。
以县域为单元提取8个土地利用指标和14个潜在空间决定要素进行典型相关分析。研究结果很好的揭示了该区土地利用格局的分异规律和成因。结果表明研究区土地利用格局是自然、人文要素综合作用的结果。在县域单元尺度上,耕地、草地、城市建筑用地、落叶阔叶林地、常绿针叶林/混交林地的分布受相应因素的影响均表现出一定的规律,反映土地综合利用状态的土地利用程度、斑块密度和辛普森多样性指数也呈现出一定的规律,同时不同土地利用类型之间的分布也相互影响。耕地、草地的分布、斑块密度、景观多样性与人口分布、气候、海陆位置有关;耕地与草地分布呈负相关;耕地分布较多的地区,景观整体的破碎度和多样性下降。城市建筑用地分布受区域经济发展水平和交通便利性的影响。落叶阔叶林分布到黄河和地级市中心的距离呈正相关。常绿针叶林/混交林的分布和土地开发利用程度受地形与气候的制约。
多尺度水平多元逐步回归分析显示,影响土地利用格局的空间因素表现出尺度效应,以土地利用强度为例,到海岸线距离、到市中心距离、年均温、到黄河距离在三个尺度上持续起作用,其他的变量仅在特定尺度上起作用。所有的变量都没有表现出有规则的尺度依赖变化规律。线性回归模型的拟合效果总体较差。
研究区土地利用变化的主要方向是城市扩展、植被增加和耕地丧失。转入城市建设用地的土地利用类型中,以城乡交错带、一年两熟旱作、一年一熟旱作、落叶阔叶林作为占较大比例;转入林地的土地利用类型中,以一年一熟粮作、郁闭灌丛、一年两熟粮作、城乡交错带占较大比例;耕地转出的土地利用类型中,以落叶阔叶林、城市建筑用地、城乡交错带、混交林、常绿针叶林占较大比例。临沂、潍坊、青岛、烟台是耕地丧失较多的四个市,临沂、烟台、泰安、潍坊、济宁是植被增加较多的五个市,青岛、潍坊、烟台、东营是城市扩展较快的四个市。
研究区土地利用动态同样受自然人为因素综合作用驱动。县域尺度典型相关分析表明,城市扩展和耕地丧失的发生呈正相关,即两者相伴发生。在经济密度较高,距京杭运河和黄河较远,距海岸线较近,年均温、河网密度和人均地区生产总值变化率较低的地方,城市扩展和耕地丧失的发生率较大。土地利用程度的变化率和植被的增加率呈负相关,即植被增加率较高的地区,土地利用程度变化较小。植被增加率受气候因子如年降雨量、最冷月均温、年均温和地形因子如海拔的制约。人口密度、经济密度较高的地区植被增加率较高,表明经济发展对植被恢复的促进作用。
单一要素分段频数统计表明,距海岸线和交通干线越近,海拔越低,耕地丧失、城市扩展、植被增加发生的频率越高;黄河、京杭运河对城市扩展的影响不明显,城市扩展发生频率随到中心市区距离的增加略微减小。黄河、京杭运河、中心市区对耕地丧失、植被增加的影响均表现为以某一距离为阈值,小于此阈值时发生频率随距离的增加而增加,超过此阈值之后发生频率随距离的增加而减小。
多尺度水平多元回归分析表明土地利用变化的驱动因素及作用强度在不同尺度上均有变化,表现出明显的尺度效应。海拔和有机质在三个尺度上对植被增加持续作用。到海岸线距离、年降雨量、人口密度在三个尺度上对耕地丧失持续作用。到海岸线距离和到市中心距离在三个尺度上对城市扩展持续作用。海拔对植被增加、人口密度对耕地丧失、到海岸线距离对城市扩展的影响随尺度增加而增强。线性模型的拟合效果很差。
Global change has been the most active field of study these years with the problem of population, resources and environment increasingly growing. It is realized that land-use/cover change (LUCC) is the major force driving environment and climate change, and it also has significant impacts on sustainable development of human society. LUCC has been the main contents of many global change research plan since 1990s, due to its importance in the global change and sustainable development.
As a major economic province, contradiction between economic development and land-use has been aggravated in Shandong province. Through the study of land-use change and its driving forces, land-use change can be understood in a timely and accurate manner, and trend, distribution and scale of land-use change can be monitored, so that we can provide basic data for urban construction, and we can serve for investigation, planning, conservation and rational utilization of land resources.
Unsupervised and supervised methods for classification were used to classify land-use types, based on annual series of MODIS-NDVI combined with dem and night light data. Canonical correlation and multiple regression were used for exploring land-use pattern, change dynamics and their driving forces.
Overall accuracy of land-use classification for 2001 is 84.8%, and the Kappa Coefficient is 0.83. Urban-built land, wood land and farmland respectively take up 2.13%, 23.37%, 52.53% of the whole, which is approximate with statistical data.
Combination of night light data into NDVI time series greatly improved accuracy of identification of urban-built land, producer accuracy and user accuracy of which are 87.01%, 95.71% respectively.
Seasonal profiles of two rotation dry croplands and two rotation rice are bimodal, while profiles of other land-use types are unimodal.
Nighttime light density shows significant difference between different land-use types. Nighttime light density of urban-built land is the highest of all, and that of urban-rural crossbelt is the second highest.
Elevation shows significant difference between between different landuse types. Elevation of mixed forest is the highest and Bulrash marsh, salt meadow and water have the lowest elevation.
Canonical Correlation Analysis is conducted with eight land use index and fourteen potential spatial determinants involved in county level. Land-use pattern and its causes are clearly revealed and interpreted using this method. Results show that land-use pattern is determined by biophysical factors combined with socio-economic factors. In county level, distribution of farmland, grassland, urban-built land, deciduous forest and coniferous/mixed forest, as well as Land use Intensity, Patch Density and Simpsons Diversity Index all show some patterns affected by corresponding factors. Relationship also exists among certain land use index. Distribution of farmland and grassland, patch density and diversity is affected by population density, climate and distance to coastal line. Negative relationship exists between distribution of farmland and grassland. Fragmentation and diversity of the whole landscape decrease when proportion of farmland increases. Urban-built land is mainly influenced by economic level and convenience of traffic. Distribution of deciduous forest is relevant to distance to Yellow River and region-level city centers. Both distribution of coniferous/mixed forest and land-use intensity are restrained by terrain and climate.
Multiple regressions on multiple scales show scale effect of spatial factors. Take land-use intensity for example, distance to coastal line, distance to city centers, annual average temperature and distance to Yellow River consistently influence land-use intensity, while other factors only have influences on certain scales. No factor shows any regular trend with increase or decrease of scales. Regression model fitting results are not so satisfying.
Urban expansion, vegetation increase and farm loss are the three major aspects of land-use change in Shandong province. Urban-rural crossbelt, one rotation dry croplands, two rotation dry croplands and decidous forest take up large part of area which converts into urban-built land. One rotation dry croplands, closed shrubland, two rotation dry croplands and urban-rural crossbelt take up large part of area which converts into wood land. A large part of area which comes from farmlands is transformed into decidous forest, urban-built land, urban-rural crossbelt, mixed forest and coniferous forest.
City of Linyi, Weifang, Qingdao and Yantai are the four cities which lose more farmlands. City of Linyi, Yantai, Taian, Weifang and Jining are the five cities in which vegetation increases more. City of Qingdao, Weifang, Yantai, Dongying are the four cities in which urban-built land expands more quickly.
Land-use dynamic is driven by both biophysical and human factors. Canonical Correlation Analysis in county level shows that urban expansion and farm loss are positively related, which reveals the two happens closely. Urban expansion and farm loss happen more frequently where economic density is higher, distances to canal and yellow river are further, distance to coastal line is nearer, and annual average temperature, hydrological density and rate of change of Per Capita GDP are lower. Rates of change of land-use intensity and increase of vegetation are negatively correlated. Rate of increase of vegetation is restrained by such climatic factors like annual rainfall, average temperature of the coldest month, annual average temperature, and terrain. Rate of increase of vegetation is higher where population density and economic density are higher, which indicates economical development promotes vegetation restoration.
Frequency statistic for single factor shows that farm loss, urban expansion and vegetation increase happen more frequently where distances to coastal line and main roads are nearer, and elevation is lower. Yellow River, Jing Hang Canal have no impacts on urban expansion. Rate of urban expansion decreases slightly when distance to city centers increases. Thresholds of distances to Yellow River, Jing Hang Canal and city centers exist when they influence farm loss and vegetation increase. When the value is under the threshold, farm loss and vegetation increase increase with the distance increasing, while they decrease with the distance increasing when the value is over the threshold.
Multiple regressions on multiple scales show scale effect of spatial factors. Elevation and soil organic matter consistently influence vegetation increase. Distance to coastal line, annual rainfall and population density consistently influence farm loss. Distance to coastal line and Distance to city centers consistently influence urban expansion. Elevation, population, and distance to coastal line show regular change with increase or decrease of scales when they influence vegetation increase, farm loss and urban expansion respectively. Regression model fitting results are not so satisfying.
引文
[1]刘彦随,陈百明.中国可持续发展问题与土地利用/覆被变化研究[J].地理研究,2002,2l(3):324-330.
[2]韦素琼,陈健飞.基于闽台对比的福建耕地变化趋势演绎[J].自然资源学报,2005,20(2):206-211.
[3]路云阁,蔡运龙等.走向土地变化科学[J].中国土地科学,2005,20(1):55-61.
[4]Lambin E.F.,Baulies X.,Bockstael N.,et al.Land-use and Land-cover change (LUCC):implementation strategy[R].IGBP Report No.48 and HDP Report No.10.Stochkholm:IGBP.1999.
[5]Veldkamp A.,Lambin E.F..Predicting land use change[J].Agriculture,Ecosystems and Environment,2001,85:1-6.
[6]Ojima D.,Lavorel S.,Graumich L.,et al.Terrestrial human-environment systems:the future of land research in IGBP Ⅱ[J].In:Global Change Newsletter Issue No.50,2002.
[7]Moran E.F..News on the land project[J].In:Global Change Newsletter Issue No.54,2003.
[8]Justice C.,Townshend J..Special issue on the moderate resolution imaging spectro-radiometer(MODIS):a new generation of land surface monitoring[J].Remote Sensing of Environment.2002,83:1
[9]Lu H.,Raupach M.R.,McVicar T.R.et al.Decomposition of vegetation cover into woody and herbaceous components using AVHRR NDVI time series[J].Remote Sensing of Environment,2003,86:1-18.
[10]Zanoni V.M.,Goward S.N..A new direction in Earth observations from space:IKONOS[J].Remote Sensing of Environment,2003,88:1-2.
[11]Ola H.,Geoffrey J.H.,Andre B.,et al.Detecting dominant landscape objects through multiple scales:An integration of object-specific methods and watershed segmentation[J].Landscape Ecology,2004,19:59-76.
[12]Liu J.Y.,Zhuang D.F.,Luo D.,et al.Land-cover classification of China:integrated analysis of AVHRR imagery and geophysical data[J].International Journal of Remote Sensing,2003,24(12):2485-2500.
[13]Wen C.G.,Tateishi R.30-second degree grid land-cover classification of Asia[J].International Journal of Remote Sensing,2001,22(18):3845-3854.
[14]Anderson L.E.The causes of deforestation in Brazilian Amazon[J].Journal of Environment and Development,1996,5:309-328.
[15]Reid R.S.,Krushka R.L.,Muthui N.,et al.Land-use and land-cover dynamics in response to changes in climatic,biological and socio-political forces:the case of southwestern Ethiopia[J].Landscape Ecology,2000,15:339-355.
[16]Mertens B.,Sunderlin W.,Ndoye O.,et al.Impact of macro-economic change on deforestation in South Cameroon:integration of household survey and remotely-sensed data[J].World Development,2000,28:983-999.
[17]蒙吉军,李正国.河西走廊张掖绿洲LUCC的驱动力分析[J].地理科学,2003,23(4):464-470.
[18]邹亚荣,张增祥,周全斌等.中国农牧交错区土地利用变化空间格局与驱动力分析[J].自然资源学报,2003,18(2):222-227.
[19]Veldkamp A.,Fresco L.O.CLUE-CR:an integrated multi-scale model to simulate land use change scenarios in Costa Rica[J].Ecological Modelling,1996,91:231-248.
[20]Veldkamp A.,Fresco L.O.Reconstructing land use drivers and their spatial scale dependence for Costa Ri ca(1973 and 1984)[J].Agricultural Systems,1997,55:19-43.
[21]Verberg P.H.,Welmoed S.,Veldkamp A.,et al.Modeling the spatial dynamics of regional land use:the CLUE-S model[J].Environmental Management,2002,30(3):391-405.
[22]Lambin E.F.,Turner B.L.,Geist H.J.,et al.The causes of land-use and land-cover change:moving beyond the myths[J].Global Environmental Change,2001,11:261-269.
[23]徐银良,胡宁.山东省城市化过程中的土地利用问题研究[J].山东师范大学学报(自然科学版),2004,19(1):68-71.
[24]晓梅,杨勤业,张洪业.山东省耕地变化趋势及驱动因子分析研究[J].地理研究,2001,18(3):298-306.
[25]王仁卿.山东森林植被恢复的理论方法和实践[J].山东林业科技,2001,3:11-15.
[26]Wen C.G.,Tateishi R.30-second degree grid land-cover classification of Asia[J].International Journal of Remote Sensing,2001,22(18):3845-3854.
[27]Friedl M.A.,Woodcock C.,Gopal S.,et al.A note on procedures used for accuracy assessment in land cover maps derived from AVHRR data[J].International Journal of Remote Sensing,2000,21(5):1073-1077.
[28]Friedl M.A.,D.K.McIver,J.C.F.Hodges et al.Global land cover mapping from MODIS:algorithms and early results[J].Remote Sensing of Environment,2002,83,287-302.
[29]Giri C.,Shrestha S..1996.Land-cover mapping and monitoring from NOAA AVHRR data in Bangladesh[J].International Journal of Remote Sensing,1996,17(14):2749-2759.
[30]Brown J.F.,Loveland T.R.,Merchant J.W..Using multi-source data in global land-cover characterization:concept,requirements,and methods[J].Photogram metric Engineering & Remote Sensing,1993,59(6):977-987
[31]Meyer W..Human Impact on the Earth.Cambridge:Cambridge University Press,1996.
[32]Schneider A.Classification of urban areas at continental scales using remotely sensed data(C).IEEE,2001.
[33]桌莉,李强等.基于夜间灯光数据的中国城市用地扩展类型[J].地理学报.2006,612,169-178.
[34]Ma M.G.,Veroustraete F..Reconstructing pathfinder AVHRR land NDVI time-series data for the Northwest of China[J].Advances in Space Research,2006,835-840.
[35]赵萍,傅云飞.基于分类回归树分析的遥感影像土地利用/覆被分类研究[J]. 遥感学报,2005,9(6):709-716
[36]Bauer E.,Kohavi R..An empirical comparison of voting classification algorithms:bagging,boosting,and variants[J].Machine Learning,1998,36,105-142.
[37]周生路,黄劲松.东南沿海低山丘陵区土地利用结构的地域分异研究—以温州市为例[J].土壤学报,2003,40(1):37-45.
[38]姚晓军,张明军,孙美平.甘肃省土地利用程度地域分异规律研究[J].干旱区研究,2007,03:312-315.
[39]庄大方,刘纪远.中国土地利用程度的区域分异及模型研究[J].自然资源学报,1997:12(2):105-111.
[40]王思远,张曾祥,周全斌,等.中国土地利用格局及其影响因子分析[J].生态学报,2002,23(4):649-656.
[41]龙花楼,李秀彬.长江沿线样带土地利用格局及其影响因子分析[J].地理学报,2001,56(4):417-425.
[42]邱炳文,王钦敏,陈崇成等.福建省土地利用多尺度空间自相关分析[J];自然资源学报,2007,22(02):311-321.
[43]谢花林,刘黎明,李波,张新时.土地利用变化的多尺度空间自相关分析—以内蒙古翁牛特旗为例[J].地理学报,2006,61(4):389-400.
[44]张明,朱会义,何书金.典型相关分析在土地利用结构研究中的应用[J].地理研究,2001,20(6):761-767.
[45]蒋志刚,李春旺,曾岩.生物实验设计与数据分析.北京:高等教育出版社,2003.481.
[46]刘纪远主编.西藏自治区土地利用.北京:科学出版社,1992.60.
[47]李月臣,陈晋,宫鹏等.基于NDVI时间序列数据的土地覆盖变化检测指标设计[J].应用基础与工程科学学报,2005,13(3):261-275.
[48]Lenney M.P.,Woodcock C.E.,Collins J.B.,et al.The status of agricultural lands in Egypt:The use of multi-temporal NDVI features derived from Lands at TM[J].Remote Sensing on Environment,1996,56:8-20.
[49]Townshend J.,Justice C.,Li W.,Gurney C.,et al.Global land-cover classification by Remote Sensing:present capabilities and future possibilities[J].Remote Sensing Environment,1991,35:243-255.