(1. Peking University Institute of Remote Sensing and Geographic Information System, Beijing,100871; 2. School of Earth Science and Resources, China University of Geosciences, Beijing, 100083)
Taking two TM images of 65438+1February 1999 and April 20001year and other auxiliary data as data sources, the remote sensing images processed by principal component analysis are automatically identified by computer, and the distribution information of various land use types is obtained through human-computer interaction interpretation. Using the spatial analysis and mathematical statistics functions of GIS, this paper analyzes the characteristics of land use change, the dynamic attitude of land use and the transformation between various types in Bazi area of Lijiang. The results showed that the cultivated land area increased by 0.27km2, the urban and rural land area increased by 0. 19km2, the water area decreased by 0. 1 1km2, and the woodland and grassland remained basically unchanged. Various types of transformation are highlighted as the transformation of water areas to urban and rural land, and the transformation of unused land to urban and rural land and cultivated land.
Key words: land use change; Markov matrix; Lijiang, Yunnan; Remote sensing and geographic information system
Regional land use/cover change (LUCC) is one of the important research fields of global environmental change. Land use change data can provide the actual situation of studying the land use situation in this region, ensure the country to grasp the land use change situation in time and accurately, and provide scientific basis for formulating national economic development plans, plans and macro-decisions [1]. Land use change includes three aspects: time change, spatial change and quality change, among which spatial change reflects the spatial type of land use change, the spatial distribution of change types and regional differences, and is the focus of land management and planning [2]. The key to solve this problem lies in how to extract the data of land use change and how to scientifically analyze and process the obtained information. Remote sensing technology has the advantages of fast, accurate and short cycle, and has obvious advantages in monitoring large and medium-scale land use/cover changes. In this paper, remote sensing, GIS and mathematical statistics are used to describe the quantitative and spatial changes of land use in Bazi area of Lijiang during 1999 ~ 200 1, which lays a foundation for land management decision-making, ecological environment protection and rational development of resources in this area and is of great significance to the sustainable development and utilization of regional land.
1 Overview of the study area
Lijiang is located in the middle reaches of the Jinsha River in the northwest of Yunnan Province [3], with geographical coordinates of 99 23 ′ e ~101′ e, 25 59 ′ n ~ 27 56 ′ n, and is adjacent to Diqing, Nujiang, Dali and Sichuan Provinces. The total land area is 93 1.02km2, which is located at the junction of Qinghai-Tibet Plateau and Yunnan-Guizhou Plateau. It belongs to Hengduan Mountain area, with complex terrain, high mountains and deep valleys, and relatively closed long-term natural environment, forming a typical vertical natural belt and three-dimensional climate, with extremely complex animals and plants and rich rare wild animals and plants. The total population of the whole region is 6.5438+0.09 million, including 23 ethnic groups such as Han, Naxi, Yi, Lisu, Tibetan, Bai and Pumi, among which ethnic minorities account for 57% [3].
2 research methods
2. 1 research data
Two Landsat-TM images were taken in February of 1999 and April of 20001year respectively. The ground resolution is 30m, there is no cloud in the study area, and the quality is good. In addition, 1 ∶ 1000 topographic map and 1 ∶ 50000 land use map are used to assist the selection and visual interpretation of training sample areas. There are other statistics (meteorology, hydrology, population, soil, social economy) and so on.
2.2 Main workflow
There are three main methods for change detection using remote sensing image information: post-classification comparison method, multi-temporal image direct intersection method and multi-temporal image classification method [4]. The direct intersection method of multi-temporal images requires two phases to be close, and the change information is usually detected by image difference, ratio or principal component analysis. The operation is simple, but the specific feature types of changes cannot be obtained. The time difference of remote sensing images used in this paper is close to 3 months, so it is not suitable for direct intersection method. Multi-temporal image classification should use static type and dynamic type, and the training samples of dynamic type are generally difficult to determine. This paper mainly adopts the post-classification comparison method.
The main workflow of this study is shown in figure 1: firstly, two TM images are preprocessed (such as geometric correction). By using PCI software, the principal component analysis of the above two images is carried out, and then all kinds of knowledge (such as topographic map, vegetation map, etc.) are fully combined to supervise and classify. ) [3]. In the supervised classification, the Technical Specification for Land Use Investigation in China was adopted, and the land use types were divided into six categories: cultivated land, woodland, grassland, water area, urban and rural industrial and mining settlements and unused land. The classification results are interpreted by man-machine interaction, and the interpretation results are transmitted to the geographic information system software ARC GIS in grid form for data processing. Spatial superposition analysis of the second phase data is carried out to obtain the data of land use change, and then the dynamic change analysis of land use is carried out.
Figure 1 main work flow chart of the project
2.3 Treatment of several key technologies
2.3. 1 geometric registration of remote sensing images
Geometric registration of two time-phase remote sensing images is the basis of dynamic change research. In order to use geoscience information for auxiliary analysis, remote sensing images can be registered in geodetic coordinate system. In this paper, the topographic map of 1 ∶ 1000 is selected as the benchmark, and 12 ground control points are evenly selected on the remote sensing image. Quadratic polynomial fitting is used, and the gray sampling method is bicubic convolution, which is used for geometric registration and gray resampling. Results The overall median error is 0.624, and the registration error is less than 1 pixel, which meets the requirements of dynamic monitoring of land use.
2.3.2 Principal component analysis of remote sensing images
Principal component analysis (also known as principal component transformation) is a method to describe the measured values of multiple variables with several comprehensive indicators without losing information as much as possible [5]. In multi-spectral images, because there are many correlations between the data of each band, most of the information of the image can be represented by a few bands through principal component analysis, so that the information is almost not lost but the data amount can be reduced. In this paper, the principal components of the six infrared bands of the second Landsat-TM image are analyzed, and the first to third principal components are color synthesized to get the processed image.
2.3.3 Generate result data
According to all kinds of auxiliary data, six training samples of land use types are selected from the secondary remote sensing images after principal component analysis, and the maximum voltage method is used for computer automatic identification. Because there are different spectra of the same object, different spectra of foreign objects and mixed pixels in remote sensing images, computer automatic identification is not ideal in distinguishing some categories, such as unused land and urban land. In practical work, it is necessary to assist in visual correction of various measured data to obtain the final classification results of the two phases, and then transfer the classification results into the geographic information system software ARC GIS in the form of grid, carry out spatial superposition analysis on the data of the second phase, obtain land use change data, and make necessary mathematical statistics on the analysis and use of the results.
3 Result analysis
3. 1 Analysis of land use change range
The range of regional land use change is mainly reflected in the total area change of different land use types, which can provide the overall information of regional land use change and land use structure change.
Based on the second-phase remote sensing data, this paper makes a statistical analysis of the two-phase land use data in the Bazi area of Lijiang, and the results are shown in table 1.
Table11999 ~ 2001land use change table
Table 1 shows that in recent three years, the cultivated land area in this area has increased by 0.27km2, the urban and rural land area has increased by 0. 19km2, and the water area has decreased by 0. 1 1km2, while the woodland and grassland have remained basically unchanged.
3.2 Dynamic attitude analysis of land use
The dynamic attitude of land use can be simply described by single dynamic attitude of land use and comprehensive dynamic attitude of land use [6]. Among them, a single land use dynamic attitude can quantitatively describe the speed of a certain land use type change in a certain time range, provide regional differences in land use change and predict the future land use change trend; Comprehensive land use dynamics is used to characterize the speed of regional land use change.
The formula expression of single land use dynamic attitude is:
Innovation of Land Information Technology and Development of Land Science and Technology: Proceedings of the 2006 Annual Conference of china land science Institution.
Where Ua and Ub respectively represent the quantity of a certain land use type at the beginning and end of the study; T stands for long learning period.
The expression of comprehensive land use dynamic attitude is:
Innovation of Land Information Technology and Development of Land Science and Technology: Proceedings of the 2006 Annual Conference of china land science Institution.
In the formula, LUi represents the area converted to other land uses at the end of the study of Class I features at the initial stage of the study; LUi indicates the area converted to other land use types at the end of the study on the characteristics of Class I land; T stands for study period.
According to formulas (1) and (2), the annual change rates of six land use types in Bazi area of Lijiang are calculated. The results show that from 1999 to 200 1, the annual change rate of land use in Bazi area of Lijiang is 0. 17%, among which the change rates of urban and rural land and unused land are the fastest, reaching 0.36% and 0.2 1% respectively, while woodland and grassland remain basically unchanged, while cultivated land and grassland remain unchanged.
3.3 Analysis of regional differences in land use
Due to the differences in natural conditions such as topography and climate, as well as the differences in economic development and population growth rate, the regional differences in land use are significant. The regional difference of land use change can be expressed by the relative change rate of specific land use types. The relative change rate is a good way to reflect the regional differences of land use change, and its expression is:
Innovation of Land Information Technology and Development of Land Science and Technology: Proceedings of the 2006 Annual Conference of china land science Institution.
Where Kb and Ka respectively represent the area of a specific land use type in a certain area at the beginning and end of the study; Cb and Ca represent the area of a specific land use type in the whole region at the beginning and end of the study respectively.
According to administrative divisions, Lijiang is divided into four regions: Lijiang County, ninglang county, Yongsheng County and Huaping County. According to the remote sensing data, the relative change rates of six land use types are calculated respectively, and the results are shown in Table 2.
Table 2 Regional differences of land use from 1999 to 200 1
As can be seen from Table 2, there are obvious differences in land use change: ① In terms of cultivated land, Lijiang County is the largest, reaching 1.53, which is obviously larger than Yongsheng County, and ninglang county and Huaping counties are also larger than Yongsheng County; (2) There is little regional difference in the change of woodland, grassland and water area; ③ The regional difference of urban and rural land use is the most obvious, and Lijiang County is much larger than the other three counties, reaching 5.36; ④ The change of unused land in Huaping County is outstanding, about 4.89, almost three times that of Yongsheng County.
3.4 land use type conversion matrix analysis
The mutual transformation between land use types can be further described by Markov transfer matrix model [7]. Markov chain is a special stochastic process with no aftereffect, which reflects a series of processes in which a metastable system changes from time n to time n+ 1 in a series of specific time intervals, and the state of time n+ 1 is only related to the state of time n ... because the evolution of land use types has the nature of Markov stochastic process: ① In a certain area, different land use types are possible. ② There are many events in the process of mutual transformation between land use types that are difficult to be accurately described by functional relationships, so Markov transfer matrix model can be used to describe the dynamic transformation of land use types.
The key to the application of Markov model in land use type conversion is to determine the transfer probability matrix P between land use types. If the transition probability of the inter-slice region is taken as a matrix element, the transition matrix model is:
Innovation of Land Information Technology and Development of Land Science and Technology: Proceedings of the 2006 Annual Conference of china land science Institution.
Where Pij is the transfer probability of land use type I to land use type J.
The probability matrix of land use change and transfer obtained by computer automatic identification of remote sensing images is shown in Table 3.
Table 31probability matrix unit of land use change transfer from 999 to 200 1 year:%
As can be seen from Table 3:
(1) The increase of cultivated land mainly comes from grassland and unused land, accounting for 1.5 1% and 0.2 1% respectively. Most of the decrease was converted into unused land and grassland, accounting for 4.46% and 2. 10% respectively.
(2) The increase of forest land mainly comes from grassland and cultivated land, which are 4.06% and 1.06% respectively, and 1% of forest land is converted into grassland.
(3) 2. 10% and 1.00% of grassland increase came from cultivated land and woodland, and 4.06% and 1.5 1% of grassland decrease were converted into woodland and cultivated land respectively.
(4) 2.32% of the water area is converted into urban and rural land, 0.36% into unused land, and 0. 18% into forest land, indicating that the reduction of water body is mainly affected by human factors.
(5) The increase of urban and rural land mainly comes from cultivated land, unused land and water area, accounting for 0.74%, 0.6 1% and 0.39% respectively.
(6) 17.06%, 4.46%, 0.93% and 0.58% of urban and rural land, cultivated land, grassland and woodland were converted into unused land respectively, indicating that the land use situation in this area is not very good, and a large number of cultivated land was occupied but not reasonably developed.
4 conclusion
The study of land use/land cover change (LUCC) aims to deeply understand the current situation of land use and its dynamic process, leading causes and evolution mechanism, so as to improve people's ability to predict, manage, make decisions and regulate land use change, which is very important for data collection, analysis and processing. By analyzing the data obtained in this study, we can draw the following conclusions:
During the period of (1)1999 ~ 2001,the land use situation in the Bazi area of Lijiang, Yunnan Province changed to some extent. In recent three years, the cultivated land area increased by 0.27km2, the urban and rural land area increased by 0. 19km2, and the water area decreased by 0. 1 1km2, while the woodland and grassland remained basically unchanged. The transformation of waters to urban and rural land, unused land to urban and rural land and cultivated land highlights various types of transformation.
(2) The annual change rate of land use is 0. 17%. Among them, the growth rate of urban and rural land is the fastest, with an average annual growth rate of 0.36%, and the unused land decreases at a rate of 0.2 1%, while the woodland and grassland remain basically unchanged.
(3) The transformation between land use types is from water to urban and rural land, and from unused land to urban and rural land and cultivated land. In recent three years, the water area of 0.068km2 (2.32%) in this area has been transformed into urban and rural land, and the unused land of 0. 14km2 and 0.07km2 has been transformed into urban and rural land and cultivated land, accounting for 17.06% and 5.46% of the unused land area respectively.
refer to
Zhu, He, He, et al. Spatial and temporal changes of land use around Bohai Sea [J]. Journal of Geography, 200 1, 56 (3): 253 ~ 260.
Wang Siyuan, Liu Jiyuan and Zhang Zengxiang. Land use pattern and its evolution in China in recent 10 years [J]. Journal of Geography, 2002,57 (5): 523 ~ 530.
[3] Lijiang Local Records Office. Lijiang yearbook (1997) [M]. Kunming: Yunnan Nationalities Publishing House, 1997, 293 ~ 294.
Li, a man. Remote sensing detection and analysis of dynamic change of land use/cover in Xiamen [J]. Geography Science, 200 1, 2 1 (6): 537 ~ 544.
Ji Jun Murai et al. Accurate solution of remote sensing [M]. Beijing. Surveying and Mapping Press, 1993,12:194 ~196.
Wang Xiulan, Bao Yuhai. Discussion on the research method of land use dynamic change [J]. Progress of geographical science, 1999,18 (1): 81~ 87.
Li, Wang Yimou, et al. Analysis of land use change in Yulin area based on RS and GIS. Journal of soil and water conservation, 2003, 6: 97 ~ 99.