CHINESE GEOGRAPHICAL SCIENCE Volume I 1, Number 4, pp. 366 - 376, 2001 Science Press, Beijing, China A STUDY ON LANDSCAPE MOSAIC STRUCTURE IN URBAN-RURAL AREA IN NORTHWEST OF CHINA WITH RS AND GIS -A Case Study of Xigu District in Lanzhou City XU Jian-hua 1, LU Yan 1, AI Nan-shan 2, YUE Wen-ze 3 (1. Department of Geography, East China Normal University, Shanghai 200062, P. R. China; 2. Environmental Science and Engineering School, Sichuan University, Chengdu 610064, P.R. China; 3. Department of Geography, Lanzhou University, Lanzhou 730000, P. R. China) ABSTRACT: At the study area of Xigu District in Lanzhou City, using RS & GIS as tools we apply Diversity, Domi- nance, Fragmentation, Isolation and so on to study the quantitative, fractal and spatial characters of landscapes" structures in the four sub-regions divided by the morphological features. Using the Fractal Theory to establish the fractal structure models, we analyze the complexity and stability of various landscapes" distribution with fractal dimension value. The spatial distribution characteristics of landscape mosaic structure are also expounded. At the end of the paper we discuss the relevant problems on the main factors which control and effect on the spatial pattern of landscapes as well as on landscape optimization and management. KEY WORDS: landscape mosaic; spatial structure; fractal; Xigu District in Lanzhou City CLC number: P901; Q149 1 REGIONAL BACKGROUND Documentcode: A AND RESEARCH Article ID: 1002-0063(2001)04-0366-11 west to east, divides the whole district into two parts. The Huangshui River, the Zhuanglang River and the METHODS Xianshui River respectively flow into the Huanghe River 1.1 Regional Background in the west and north of the region. The whole geomorphology of the region is that north is much lower In Lanzhou City, Xigu District, which covers an than south, but both are higher than the middle valley area of 103019'48 " - 103 ° 4 1 ' 0 2 " E and 35°38'16 " - (Fig. 1). According 36°13'57"N, is located at the upper Huanghe(Yellow) the whole district is divided into four sub-regions. 1) River and middle part of Gansu Province. It is an in- The river valley area with an altitude of 1528 - 1600m dustrial-agricultural, is located at the 1st and 2nd fluvial terraces by the urban-rural county level adminis- trative region with an area of 3 6 2 . 9 8 k m 2. The Huanghe River, which goes through it from Huanghe to geomorphological characters, River and the Huangshui River. area of 7386ha, It has an accounting for 19. 8% of the whole Received date: 2001-09-25 Foundation item: Under the auspices of the National Natural Science Foundation of China (No. 40171069). Biography: XU Jian-hua( 1965 - ), male, a native of Gansu province, professor. His research interest includes Geo-computation and GIS. A Study on Landscape Mosaic Structure in Urban-Rural Area in Northwest of China With RS and GIS 367 district, and it is also the center of regional economy, east mountainous area to 291mm in the northwest society and culture. 2) The terrace & gentle slope area, mountainous area. located at the 4th and 5th terraces of two banks of the Huanghe River, with an area of 4401ha, which ac- 1.2 Data and Research Method counts for 11.8% of the whole district, is mainly composed of eight terraces with an altitude of 1607 - 1800m All the data needed include 1) 1:8000 aerial re- and a slope of less than 10 °, namely Dajia terrace, mote sensing image data(1998), 2) 1:50 000 TM im- Zhangjia terrace, Qingshi terrace, Zhama terrace, Li- age data(2000), 3) 1:10 000 relief map(1998), 4) ugou other data including climate, hydrology, vegetation, terrace, Fanjia terrace, Zhangjia terrace, Guangjia terrace, etc. 3) The southern mountainous soil and so on. area is located above the southern terraces, with an al- The research method consists of six major steps: 1) titude of 1800 - 2621.4m and an area of 9588ha, ac- use 1 : 5 0 000TM image to interpret and classify the counting for 25.6% landscapes, 2 ) s e l e c t 1:8000 of the whole district. 4) The aerial remote sensing northern mountainous area is located above the northern image and 1:10 000 relief map to make a survey of land terraces of the Huanghe River, with an altitude of cover/land use and landscape patches, 3) digitize im- 1600- 2064. 6m and an area of 16 001ha, accounting ages and relief map data( remote sensing image, relief for 42.8% of the whole district. map, survey data and so on), 4) integrate data into a GIS system by image calibration, registration and data processing for building spatial and attribute databases, 5) apply the quantitative methods to the quantitative characteristics of the landscape mosaic structure and fractal theory to fractal characteristics of the distribution pattern of various landscapes, 6) with spatial analytical methods of GIS(DENESH, 1995), study the main factors which control and influence the spatial pattern of landscapes. Various methods and techniques used in this study are briefly described in Fig. 2. 2 LANDSCAPE MOSAIC AND QUANTITATIVE CHARACTERISTICS In a region, various landscape patches integrate systematically into a landscape mosaic (FORMAN et al., 1986), which shows spatial pattern of landscapes and is Fig. 1 The digital elevation model of Xigu District the result of the natural factors and human activities. in LanzhouCity Xigu District is typically continental and semi-arid Some scholars think that the essence of landscape ecology is to study the landscape mosaic structure in that climate and has obvious vertical climate change. Aver- landscapes age daily temperature difference and annual average 1995; XIAO, 1991). temperature are respectively 13.6~C and 8.5°C, are diverse mosaic entities(FORMAN, ex- treme high temperature 39.7~C and extreme lower tem- 2. 1 Landscape Types and Mosaic Map perature - 23.4°C. And the change rate of rainfall and distinct spatial discrepancy are much larger. Annual By TM interpretation we classify the landscapes rainfall gradually decreases from 500ram in the south- into ten kinds of landscapes, which are respectively 368 XU Jian-hua, LU Yan, AI Nan-shah et al. Relief map (1:10000) ] I Acrialremoteseosingimage,,:8ooo, ] I I [ ClassiJ} landscape types [ l'iVl image I 1:50000 ) I ] i I Interpretation [ I [ I ] I Land use/co~ er surve~ [ I Digitization [ [ I Image calibration, registration and ~ector/raster exchange I i I D I Overlay Database management I I Spatial structure of landscape mosaic } I I I Quantilati~ e analytical method [ Fractal model I Quantitative characteristic of landscape mosaic [ I Fractal characteristic of landscape mosaic Fig. 2 Research methods and techniques bare land, water area, urban area, rural settlement and In order to show and analyze the spatial structure of landscape mosaic, we made the landscape mosaic map industrial area far from settlement. The special meaning of Xigu District in Lanzhou City from the spatial and of each landscape is listed in Table 1. attribute databases of GIS, which is shown in Fig. 3. farmland, orchard, woodland, scattered woodland, Table 1 Landscapetypes and remarks Landscape types Remarks Farmland Orchard Grassland Woodland Covered by crops and vegetation Including apple orchard, peach orchard, pear orchard, apricot orchard, etc. Wasteland, pasture Natural and artificial woodlands with canopy of trees larger than 30% Scattered woodland Woodland with canopy between 10% - 3 0 % of area and newly-made woodland with no canopy but viable rate no less than 40% of reasonable tree numbers Bare land Land without vegetation coverage, including bare loess, rock and gravel Water area Urban area Rural settlement Industrial area far from settlement Including river, creek, alluvial flat and pond 2. 2 Including satellite towns and the built-up area in the region planned by City Planning Rural residential area Including various factories and plants, which are far from settlement and not in the urban area The Quantitative Characteristics of Landscape Mosaic to comprehensively quantify the landscape mosaic structure. 1 ) Diversity, which is a general description of 2. 2. 1 General quantitative characteristics of landscape mosaic we can use diversity, dominance and fragmentation abundance and balance of landscapes(TURNER et al., 1991; ZHANG et al., 2000), is defined as follows: A Study on Landscape Mosaic Structure in Urban-Rural Area in Northwest of China With RS and GIS 369 N w-E $ Legend Farmland ~ B a r e l a n d Orchard ] [ Water area Grassland ~ U r b a n a r e a _---~_._-'i~i~_Woodland ~ Rural settlement ~Scattered ~ Industrial area far woodland r~'~'~','~from settlement ~ Fig. 3 Landscape mosaic map of Xigu District $ i = In formula (1), 3) Fragmentation is used to show the number of (1) H = - EP,InP, patches in one unit area in the mosaic, and it is defined 1 S is the number of landscape types; Pi as follows: S is the area percent of landscape type i in the whole mosaic; H is diversity index, which is much larger, the more abundant landscapes (3) F = ~.~nja k= 1 In formula (3), S is as same as that in formula ( 1 ) ; nk 2) Dominance is an indictor that shows what extent is the patches' number of landscape type k; A is the one or few landscapes dominate in mosaic structure, whole area of all types of landscapes; and it is defined as follows: tion, which is much larger, the more patches are in a S n = InS + ~ P,lnP, i = F is fragmenta- unit area. (2) We use formulas ( 1 ), 1 (2) and (3) to calculate where D represents dominance, which is much larger, diversity, dominance and fragmentation as the following the more highly mosaic structure is dominated by one or in Table 2. few landscapes; S and Pi are as same as those in for- Table 2 shows that diversity ranks in the four sub-regions as river valley area(2. 0375)> terrace & mula (1). Table 2 Landscape diversity index, dominance and fragmentation in sub-regions Area River valley area Terrace & gentle slope area Southern mountainous area Norhtern mountainous area The whole district Diversity (H) 2.0357 1. 8544 1. 3080 1. 2395 1. 827 Dominance(D) 0. 2669 0. 4482 0. 9987 1. 1584 0. 658 27. 4772 51. 4398 9. 4733 8. 3118 12. 147 Fragmentation(F) (patehes/km 2) XU Jian-hua, LU Yah, AI Nan-shah et al. 370 gentle slope area(1. 8544) > southern mountainous area and the patches with an area of lkm 2 are as many as (1. 3080)> northern mountainous area(1. 2395), above 51. 4398, and the fragmentation of the river val- but there is an adverse rank in terms of dominance, namely ley area is much larger, 27.4772 patches with an area northern mountainous area(1. 1584)> southern moun- of lkm 2. But in the south and northern mountainous tainous area(0.9987)> terrace & gentle slope area area the fragmentations are small, respectively 9. 4733 (0.4482)> river valley area (0.2669). Obviously in and 8. 3118 patches with an area of lkm 2. the river valley area various landscapes are distributed 2. 2. 2 Landscape isolation equably without distinct dominant landscape types. But Isolation shows that the distribution of different in the southern mountainous area and northern moun- patches that belong to the same type of landscape in the tainous area they are distributed extremely unevenly and there are obvious dominant landscapes. In the northern landscape mosaic. It is defined as the follows(XIAO, 1991) : mountainous area, the dominant landscape is grassland, 1 ni~-/AA_~k which comprises of 85.96% of the whole area. In the (4) grassland and farmland, which respectively account for where A is the isolation of landscape type k, nk and A are as same as those in formula(3), Ak is the area of 60.93% and 28.51% of the whole area. And diversity landscape type k. southern mountainous area, the dominant landseapes are and dominance in the terrace & gentle slope area are For the same type landscape, isolation shows that between those in the river valley area and those in the what extent the landscape is cut and isolated by other south and northern mountainous areas. landscapes or corridors. That is, the larger the isolation is, the more dispersedly the patches distribute. Fragmentation represents what extent the land- As for the four sub-regions in this region, we apply scapes are cut by various corridors(creeks, roads and etc. ). The larger fragmentation, the more formula(4) to figure out the isolation of various land- strongly scapes listed in Table 3. landscapes are cut by corridors. Obviously the frag- Table 3 shows that farmland is most densely dis- mentation of terraee& gentle slope area is the largest Table 3 Isolation of various landscapes in sub-regions Area Famland Orchard Grassland Woodland Scattered woodland Bare land Water area Urban area Rural Industrial area far settlement from settlement River 6. 1682 12. 3953 7. 942 51. 2730 36. 5440 11. 0947 3. 8840 O. 2244 13. 1183 5.9180 Terrace & gentle slope area 6. 2548 8. 4266 7. 1522 44. 1453 23. 9741 29. 2099 - 10. 4207 12. 4217 17. 1712 Southern mountainous area 9. 5215 28. 8777 3. 8371 9. 9431 14. 4415 38. 7253 - 21. 0350 - Northern mountainous 13. 3231 area 21. 7048 1.4431 - 1. 3171 - 22. 8955 - valley area tributed in the river valley area and terrace & gentle - terrace & gentle slope area, more densely in the slope area, more densely in the southern mountains, southern mountainous area, but it is distributed most but in the northern mountainous area it is distributed densely in the northern mountainous area. Woodland more dispersedly. Orchard mostly is compactly dis- (including woodland and scattered woodland) hardly tributed in the terrace & gentle slope area, more com- exists in the northern mountainous area and is sparsely pactly in the river valley area, however it is distributed distributed in the river valley area and level & gentle more dispersedly in the south and northern mountainous slope area, but most densely distributed in the southern areas. Grassland is more sparsely distributed in the mountainous area. Bare land is most densely distributed A Study on L a n d s c a p e Mosaic Structure in U r b a n - R u r a l A r e a in Northwest of C h i n a W i t h RS and GIS 371 in the northern mountainous area, more densely in the river valley area, but more sparsely in the terrace & below formula(DONG, 1991): gentle slope area, most sparsely in the southern moun- Taking n = 2, we get the fractal structure model in tainous area. Urban area is mainly located at the river valley area and level gentle slope area, more densely at 2-dimension Euclid space: [P(r) ]1/o, = kr ~, -o,)/o,[a(r ) ~1/2 the river valley area, but more sparsely at the terrace & where P ( r ) IS(r) ]~-T, = kr <n-I -o~.,>/oo ,[ V(r) ]-~ (6) (7) gentle slope area. Rural settlement is distributed in all is circumference; A ( r ) is area; k is constant; D1 is fractal dimension in 2-dimension Euclid the four sub-regions, most densely in the terrace & space. gentle slope area, more densely in the river valley area As for a landscape patch, are its area and cir- and more sparsely in the southern mountainous area, but cumference suitable for the formula(7)? That is, has most sparsely in the northern mountainous area. Indus- landscape shape the fractal characteristic? Related re- trial area far from settlements is mainly distributed in search showed that any type landscape has the fractal the river valley area and terrace & gentle slope area, characteristics(BAI, 2000; XU et al., 2000). more densely in the river valley area, but more sparsely in the terrace & gentle slope area. All the water area is Taking the formula(7) logarithmic transformation, we obtain: distributed in the river valley area. ln[A(r) ] = D~ln[P(r) ] + C 3 THE FRACTAL CHARACTERISTICS OF LAND- SCAPE MOSAIC (8) It is obvious that as long as we establish the regressive model like formula(8) with the area and circumference of the patch, we can get regressive coefficient 2/D1 and Research findings show that landscape mosaic is the most typically fractal entity in nature(BAI, 2000; XU et hl., 2000), and various landscapes can be quantified by the fractal theory. MANDELBROT(1982) D1., where Dt represents the complexity and stability of landscape type. The larger D~, the more complicated the landscape type is. When D~ is equal to 1.50, the landscape type is in the state of probability similar to studied the structures of fractal entities and put forth a Brown movement, which is in the most unstable state. model: The more closer D~ is to 1.5, the more unstable the (5) landscape type is(PEARCE, r is 1995). According to the steps mentioned above, we D is fractal dimension. DONG establish the fractal structure models of various land- Lian-ke used formula (5) to build the fractal structure scapes and get their dimensions(Table 4). 1) According to complexity (D1), various land- ES(r) ]~- [V(r)]ff where S(r) is surface area, measurement scale, V(r) is volume, model which is for n dimensions Euclid space as the 1992; ZHAO et al., Table 4 The fractal m o d e l s a n d d i m e n s i o n s of different types of l a n d s c a p e Landscape type Fractal model(regression equation) Fractal dimension(D~) Sample n u m b e r ( n ) Correlation( R 2) Farmland In A, r) = 1. 6 2 4 7 1 n P ( r ) - 1. 1668 1. 2310 1231 0. 9352 Orchard In A, r) = 1. 5 6 6 1 n P ( r ) - 0 . 7431 1. 2771 707 0. 9535 0. 9267 Grassland In A r) = 1. 7 2 6 4 1 n P ( r ) - 2 . 1339 1. 1585 1041 Woodland In A r) = 1 . 5 1 n P ( r ) - 0. 3928 1. 3333 169 0. 9293 Scattered In A r) = 1. 57311riP(r) - 0 . 8164 1. 2714 119 0.9115 woodland In A r) = 1. 5 9 0 2 1 n P ( r ) - 1. 0455 1. 2577 212 0. 9304 Bare land lnA, r) = 1. 4 9 1 n P ( r ) - 0. 7539 1. 3423 160 0. 9244 Water area lnA, r) = 1. 8 1 0 2 1 n P ( r ) - 2 . 9351 1. 1049 54 0. 9159 Urban area lnA, r) = 1. 6 0 9 9 1 n P ( r ) - 1. 0011 1. 2423 400 0. 9589 Rural Settlement lnA, r) = 1. 6 2 4 7 1 n P ( r ) - 1. 1668 1. 0793 283 0. 9355 Industrial area far from settlement In A, r) = 1. 8 5 3 1 1 n P ( r ) - 2 . 3133 372 XU Jian-hua, LU }'an, A1 Nan-shan et al. scapes are ranked as follows: water area (1. 3423)> of farmland, woodland (1. 3333)> orchard settlement are distributed in the area with a slope of 0 - woodland (1. 2714)> bare (1. 2771)> scattered land (1. 2577)> rural orchard, scattered woodland and rural 2 °. Among those landscapes, water area is mainly the settlement ( 1. 2423) > farmland ( 1.2310) > grassland Huanghe River and the rest are direct results of human (1. 1585)> urban area(1. 1049)> industrial area far activities. Then it is indicated that the area, which is the from settlement(1. 0793), but if they are ranked in the most suitable for human production and living, is most terms of stability, the conclusion is just adverse. intensively affected by human activity. Seventy percent of 2) Water area is the most unstable landscape be- grassland, woodland and bare land as well as 30% of cause Xigu District is located in the transitional region farmland are distributed in the area with a slope of 25 - of three natural regions, and it is very complicated as a 90 °. It shows that landscapes affected less intensely by result of effects of unstable rainfall and surface water human activities are just in the area with steeper slope, so supplies. the conclusion can be made that the structure of land use 3) That the dimension of woodland landscape is is not rational in the region. In the area with a slope of approximate to that of water landscape shows that they 2 - 6 °, orchard, rural settlement, water area and scat- are similar in the complexity and stability as well as dis- tered woodland cover a larger area. In the area with a The reason is that there are no slope of 6 - 15 °, orchard, rural settlement and scattered large-scale wood-planted conditions and even small- woodland are distributd widely. In the area with a slope scale woodland is also based on the water supplies. of 15 - 25 ° there is mainly farmland. tribution pattern. 4) The complexity of scattered woodland is less than that of woodland and its stability is very weak, r i , . I --M because scattered woodland is mainly artificial, and due I~1 2 5 - 9 0 ° to the difficulty and hard conditions of planting trees in k~l 15 - 2 4 o the region. 5) The complexity of farmland is higher than that ml6_15 of grassland, but has a weaker stability. This can be o F:I 2 - 6 ° i i explained that although the distribution pattern of arti- ,~ I I 1-10_2 o ficial farmland is more complicated than that of natural grassland, its stability is weaker than grassland's sta- 0 ~ o o 8 bility. 6) The complexity of rural settlement is higher than those of urban area and industrial area far from settlement, but lower in stability. Thus it is proved that rural settlement is lack in planning and there exist serious Fig. 4 Constitution of landscapes at different slopes in Xigu District irrational approvals and construction. From Table 5, we can see that most of landscapes 4 ANALYSIS ON SPATIAL STRUCTURE OF LAND- are distributd in the area with a larger slope, and SCAPE MOSAIC patches with a slope of above 25 ° account for 50. 9% of all the patches. The number of patches is the maximum 4. 1 Slope Analysis in the area with a slope of 0 - 2 ° and the minimum in the area with a slope of 2 - 6 °. The number of patches The distribution characteristics of landscapes at increase with the slope from 2 ° to 90 ° , but diversity different slopes in Xigu District of Lanzhou City are decrease with the slope. Dominance in a slope of 0 - 2 ° shown in Fig. 4. Above 80% of water area, urban area is higher and increases with the slope as urban land- and industrial area far from settlement and above 30% scape is made from several larger and continuous A Study on Landscape Mosaic Structure in Urban-Rural Area in Northwest of China With RS and GIS 373 Table 5 Quantitative characteristics of landscape mosaic structure in different slope area Slope Area (ha) Patch number Diversity index (H) Dominance (D) Fragmentation ( F ) 0 -2 ° 11966.3 3288 2.036 6.062 27.477 2 -6 ° 1462.5 611 1.828 4.587 41.777 6 -15 ° 2088.4 1217 1.779 5.325 58.275 15 - 25 ° 2303.8 1297 1.323 5. 845 56.299 25 - 90 ° 18477.3 1355 1. 026 6. 186 7.333 patches. The fragmentation is the smallest in the area nance is much smaller. That human consciously improve with a slope of 25 - 9 0 °, because grassland, bare land natural conditions such as artificial irrigation projects and woodland are distributed here continuously. In the also contribute to enrich landscapes. slopes of 6 - 15 ° and 15 - 2 5 °, the fragmentation is the scape diversity and fragmentation obviously increase in largest since there are so many small patches. the area with an altitude above 2000m because of vertical Rainfall, land- zone law. However, in the area with an altitude of 1800 4.2 2000m, the number of landscape types is the smallest and Altitude Analysis consequently dominance is the largest because of less inStatistical results of landscapes located at different fluence of human activities. altitudes show that above 80% of water area, urban area and industrial area far from settlement and above 4. 3 Comprehensive Analysis of Slope and Altitude 40% of orchard and rural settlement is distributed in the There are also Fig. 5 shows that most landscapes are distributed above 50% of orchard, grassland, scattered woodland at a relatively small area and above 40% of patches area with an altitude below 1600m. and bare land that are located in the area with an alti- concentrates in the area with a slope of 0 - 2 °, about tude of 1 6 0 0 - 1800m. In the area with an altitude of 6 5 % o f patches at an altitude below 1800m, 2 5 % o f 1 8 0 0 - 2000m grassland and farmland are widely dis- patches are located at the 1st and 2nd terraces of the tributed, which account for above 20% of the whole Huanghe River, namely the river valley area and 40% region. Above 60% of grassland and 40% of farmland of patches at the 4th and 5th terraces, namely terrace & are distributed in the area with an altitude of above gentle slope area. Generally distribution frequency of 2000m. The quantitative characteristics of landscapes at most types of landscapes is higher in the area with low different altitudes are shown in Table 6. In the area with altitude and gentle slope. an altitude of 1600 - 1800m, the area of each landscape lower in the area with higher altitude and steeper slope. Accordingly frequency is is the largest and the number of patches is the maxi- But woodland landscape is mum. Diversity(H) begins to increase and ends to de- mostly distributed in the southern mountainous area with crease with the altitude. Dominance is the strongest in higher altitude and steeper slope. an exception, which is the 'area with an altitude of 1 6 0 0 - 1800m and the weakest below 1600m. Fragmentation in the area with 5 THE MAIN FACTORS AFFECTING AND CON- an altitude above 2000m is the largest and much larger TROLLING LANDSCAPE SPATIAL PATTERN below 1600m. Except for area and dominance, other three dictators are the smallest in the area with an alti- Using the spatial analytical functions(Overlay, tude of 1800 - 2000m. The reason is that the area with Buffer and so on ), we find that the regional landscape an altitude below 1800m has a gentle slope and is af- pattern is mainly affected and controlled by climatic fected intensely by human activities, so diversity and conditions, river corridor, topographical pattern and fragmentation are much larger and accordingly domi- human activities. XU Jian-hua, 374 LU Yah, AI Nan-shah et al. Table 6 Quantitative characteristics of landscape mosaic structure in different altitudes area Altitude(m) Area (ha) <1600m Patch n u m b e r ( P ) Diversity(H) Dominance(D) F r a g m e n t a t i o n ( p a t c h e s / k m 2) 8423.66 1078 1.978 5.005 12.797 16242.31 I811 1.305 6.197 11.150 1800 - 2 0 0 0 m 6390.51 543 0.784 5.514 8.497 > 2000m 5241.97 786 1.177 5.490 14.994 1600 - 1 8 0 0 m 50 5O .q 40 3o 30 20 ~q / ¢ 4O m z m z = ~q 2o / 10-- 10 / / 2 6° 0.2 ° (a) 6-15 ° 15-25 ° 25-90 ° 1600-1800m 1500-1600m , S 1800-2000m >2000m Altitude ( b ) Altitude distribution frequency Slope SIopedistribution frequency Fig. 5 The distribution frequencies of landscapes at different slopes and altitudes mountains, and the dominant sort of species is spruce. 5.1 Sixty percent of woodland in the whole district con- Climate centrates here. From the macro-regional background, Xigu District is located at the transitional region of the three climatic 5 . 2 River Corridor and Landform regions, namely eastern monsoon region, northwestern arid region and the Tibet Plateau region. And this dis- The regional landscape spatial pattern is absolutely has a typical characteristic of continental and controlled by river corridors and landform, by whose semiarid climate. The natural landscape formed in the effects the four sub-regions are formed, namely the river climatic valley area, trict conditions is semi-arid grassland(FANG, the terrace & gentle slope area, the so grassland is a dominant landscape in the southern mountainous area and the northern mountain- whole district. Within the district the spatial differences ous area. This objectively shows the spatial pattern of of climatic conditions are very distinct, especially ver- landscape. The river valley area naturally becomes the tical climate changes in the mountainous areas. In the center of human activities because of its flat land, northern mountainous area there is enough sunlight and convenient traffic and closer to water resources, so intensive various landscape patches are distributed evenly in the 1994), evaporation, 320mm/a), but sparse rainfall(290 - where it is very difficult to cultivate the area and landscape diversity index is large and domi- only nance is weak. In the terrace & gentle slope area due to district and is platter land, relatively lower altitude and more conve- mostly(above 60% ) irrigation land depending on lift nient irrigation conditions, obviously there concentrate irrigation from the Huanghe River. In the southeast farmland, orchard and rural settlement. In the southern Guanshan mountainous area temperature is relatively mountainous area and northern mountainous area, on low, but annual rainfall is as much as above 500ram. account of the effects of steeper slope , higher altitude So there are some forests in the shade hillsides of and of being far away from water resources, their land- farmland by natural rainfall. makes up of 2. 81% of the So whole farmland A Study on Landscape Mosaic Structure in Urban-Rural Area in Northwest of China With RS and GIS scape types are not so many as those of the two forenamed areas. Landscape diversity is smaller and dominance is stronger. In the northern mountainous 375 woodland or grassland. The other is to scientifically plan rural settlement and industrial area, especially industrial area far from settlement. area dominates and intensely distributes grassland, but arid farmland, grassland and woodland in the southern 6.1 Converting Cultivated Land with a Slope above 25 ° mountains. into Woodland and Grassland By overlaying slope map and landscape mosaic 5.3 Human Activities and Traffic Corridors map, we find that there is I775.54ha of farmland, All the landscapes in the district are affected or accounting for 29. 87% of the whole farmland in the being interrupted to some extent by human activities, district, among which farmland, rural northern mountainous areas with a slope above 25 °. settlement and industrial area far from settlement, From points of view of environment conservation and scattered woodland as well as some woodland are all rational land use, all the farmland should be converted completely artificial landscapes. However traffic corri- into woodland or grassland. According to land suitabil- dors including railway, highway, rural roads and etc. ity evaluation, we give the two suggestions that in the are special artificial landscapes and they are the link- southern mountainous area, the farmland on the hill- ages of various spatial locations of human activities. sides facing to the Sun should be converted into grass- orchard, urban area, which is distributed in the southern and And they also affect the spatial pattern of other land and the farmland on the hillsides not facing to the man-made landscapes(CHEN et al. 1996). At a range Sun should be converted into grassland; of 5km to each side of railways(Lanzhou- Urumqi northern mountainous area, all the farmland should be railway and Lanzhou-Xining converted into grassland. railway) and national and in the highways(103 and 312 national highways) there intensely distributed urban area, rural settlement and in- 6 . 2 Planning Rural Settlement and Industrial Area dustrial area, which respectively account for 9 8 . 8 9 % , 64. 46% and 87.82% of the whole area of their same Our research has proved that rural settlement and types. And 88.06% of orchard is intensely distributed industrial area far from settlement are distributed dis- at a rang of 5km to each side of railway, national persedly and are lack in planning and there exist serious highway and county-rural road. The three landscapes of irrational approvals and construction. We make sug- farmland, orchard and rural settlement interlace fre- gestions from the problems mentioned above that rural quently in the district. construction land and industrial area far from settlement should be planned and managed scientifically. In the 6 TWO POINTS ON LANDSCAPE OPTIMIZATION terms of relational laws, rural settlement and industrial area far from settlement should be built or rebuilt com- AND MANAGEMENT pactly in the planned area. The aim to study regional landscape is to open out the factors, which affect and control the landscape REFERENCES pattern, and their mechanism by analyzing the structure characteristics of landscape mosaic in the different regions in order to provide the scientific bases for optimizing and managing landscapes. According to the above study, we stress two basic problems on optimizing and managing landscapes. One is to convert cultivated land in the mountainous area with a slope above 25 ° to BAI Lian-li., 2000. Fractal geometry applications in description and analysis of patch patterns and patch dynamics[J]. Ecological Modelling, 132:33 - 50. CHEN Li-ding, FU Bo-jie, 1996. Analysis of impact of human activity on landscape structure in the Yellow River delta - - a case study of Dongying region[J] . Acta Ecologica Sinica, 16(4): 334-344. (in Chinese) XU Jian-hua, LU Yan, AI Nan-shah et al. 376 DENESH M, 1995. Practical applications of geographic information systems and digital elevation models in water resources development. In: '95 Proceedings of Regional Conferenceon Water Resources Management, Conference Secretariat[ M ] . Isfahan: Isfahan University of Technology, 301 - 310. DONG Lian-ke., 1991. Fractal Theory and Its Application[M] . Shenyaqg: Science Press of Liaoning. (in Chinese) FANG Chuang-lin, XU Jian-hua., 1994. Gansu Local Geography [M]. Lanzhou: Gansu Education Press. (in Chinese) FORMAN R, Godron M, 1986. Landscape Ecology [M]. New Land Mosaics, the Ecology of Landscapes and Regions[M]. Cambridge: Cand~ridge University Press. MANDELBROT B B, 1982. in Landscape Ecology[M]. New York: Springer-Verlag. XIAO Du-ning, 1991. Landscape Ecology: Theory, Method and Application [ M ]. Beijing: China Forestry Press. (in Chinese) XU Jian-hua, AI Nan-shan, JIN Jiong et al., 2000. A fractal study on the mosaic structure of the landscape of northwest China - - taking the drainage area of Heihe river as an exampie[J]. Arid Zone Research, 18(1): 3 6 - 3 9 . (in Chinese) ZHANG Jin-tun, QIU Yang, ZHENG Feng-ying., 2000. Quantitative methods in landscape pattern analysis[J]. Journal of York: Wiley & Sons. FORMAN R, 1995. 76. TURNER M, R H GADNER (eds.), 1991. Quantitative Methods The Fractal Geometry of Nature [M]. New York: W H Freeman. PEARCE M C, 1992. Pattern analysis of forest cover in southwestern Omtario[J]. The East Lakes Geographer, 2 7 : 6 5 - Mountain Science, 18 (4) : 346 - 352. (in Chinese) ZHAO Yong-ping, WANG Yi-mou, 1995. Graphic fractionation and its application based on quantitative research of desertification[J]. Journal of Desert Research, 15(2): 175 - 180. (in Chinese)