Spatial-Temporal Gradient Analysis of Urban Green Spaces in Jinan, China Fanhua Kong ∗, Nobukazu Nakagoshi

Landscape and Urban Planning 78 (2006) 147–164

Spatial-temporal gradient analysis of urban green spaces in Jinan, China Fanhua Kong ∗, Nobukazu Nakagoshi

Graduate School for International Development and Cooperation, Hiroshima University, Kagamiyama 1-5-1, Higashi-Hiroshima 739-8529, Japan

Received 24 November 2004; received in revised form 27 June 2005; accepted 5 July 2005 Available online 23 September 2005

Abstract

In China, rapid urbanization has profoundly transformed the spatial pattern of urban land use, including urban green spaces. The government plans to optimize green spaces to integrate with urban development; this requires an understanding of the process of green space change. Quantification of green space patterns is a prerequisite to understanding green space changes, and is essential for monitoring and assessing green space functions. This paper presents a new method for quantifying and capturing changes in green space patterns, through a case study of Jinan City, China, during 1989–2004. Supported by GIS and remote sensing, the method comprises quantification of local area green spaces by the “moving window” technique (using FRAGSTATS), and a gradient analysis involving sampling from the urban center to the fringe. Results demonstrate that the significantly altered green space pattern could be quantified using landscape metrics in each local area. Gradient analysis undertaken in eight directions from the urban center reflects the changes in and effects of urbanization, and the implementation of government policy. In comparison with quantifying metrics in entire landscapes, this method more effectively links patterns and processes, and can establish an important basis for subsequent analysis of ecological and socioeconomic functions of green spaces. © 2005 Elsevier B.V. All rights reserved.

Keywords: Gradient analysis; Landscape metrics; Moving window; Remote sensing; Urban green spaces

1. Introduction but this increased to one in three by 1980 (World Com- mission on Environment and Development, 1987). It As a result of urbanization, the world’s popula- is expected that about 65% of the world’s population tion has become increasingly concentrated in cities. In will live in urban areas by 2025 (Schell and Ulijaszek, 1940, only one in eight people lived in an urban center, 1999). Population increases triggered the rapid growth of urban centers, and the environmental and socioeco- ∗ nomic consequences of this growth are profound; the Corresponding author. Tel.: +81 90 7508 8072; fax: +81 824 24 6987. increasing alienation between humankind and the nat- E-mail address: [email protected] (F. Kong). ural world is a particularly fundamental consequence

0169-2046/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.landurbplan.2005.07.006 148 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 of urbanization (Gordon, 1990). Urbanization has had landscape ecology also offers insights regarding the and continues to have a negative impact on green space optimization of the use of space vis-a-vis` environ- within cities. The impact of urbanization on urban mental conservation and improvement (Forman and green space is illustrated by the case of Mexico City, Godron, 1986; Dramstad et al., 1996; Jim and Chen, where the proportion of the city’s area that is made up 2003). of urban green space is falling by about 3.7% annually Landscape ecology deals fundamentally with how, (Nilsson and Randrup, 1997). when, and where spatial and temporal patterns influ- Urban green spaces can be defined as outdoor ence ecological processes, and how feedback from places with significant amounts of vegetation, and exist ecological processes influences ecological patterns mainly as semi-natural areas (Jim and Chen, 2003). (Turner, 1989; Urban et al., 1991; McGarigal and Cush- Urban green spaces are viewed as the last remnant man, 2002). Four basic elements are used to define of nature in urban areas (Beatley, 2000), and typ- landscape patterns: number, size, shape and juxtapo- ically perform important functions, including main- sition of patches; spatial patterns are quantified using taining biodiversity (Attwell, 2000), preventing soil landscape metrics. These are important contributors erosion (Binford and Buchenau, 1993), absorbing rain- to the interpretation of spatial patterns and ecological water and pollutants (Groot, 1994; Conine et al., 2004), processes (Gardner et al., 1987; O’Neill et al., 1988; and mitigating urban heat island effects (Stanners and Dunn et al., 1990; Wang and Zhang, 2001). Gradi- Bourdeau, 1995; Miller, 1997). Urban green spaces ent analysis, developed in the context of vegetation can also provide considerable socioeconomic benefits, analysis (Whittaker, 1967, 1975), has been used to such as providing amenity-recreation venues, reducing investigate the effects of urbanization on plant distri- work-related stress (Kaplan and Kaplan, 1989; Gobster bution (Kowarik, 1990; Sukopp, 1998) and ecosystem and Westphal, 2004), and increasing property values properties (Pouyat and McDonnell, 1991; Pouyat et al., (Geoghegan et al., 1997; Tyrvainen,¨ 1997; Morancho, 1995; Zhu and Carreiro, 1999). In recent research, Luck 2003). Rapid urbanization and increased leisure time and Wu (2002) and Zhang et al. (2004) used gradient make people more aware of urban green space, and analysis to study urban landscape patterns and the eco- there is an increasing realization that it is difficult to live logical consequences of the urbanization process by without some contact with nature. Even though they using a “split window” along their chosen transect. It become more urban in their way of life, the desire for remains difficult, however, to obtain spatial information contact with nature will continually increase rather than on each local area for the entire landscape. The prob- decrease (Miller, 1997). At the same time, governments lem of how to link pattern and process accurately in any are beginning to recognize the importance of healing local area has yet to be solved. In this study, instead of the rift between humans and nature. Green space is analyzing the whole or partial landscape pattern, gradi- becoming an important measure in judging the eco- ent analysis, supported by the “moving window” option logical sustainability of urban areas (Chiesura, 2004). in the FRAGSTATS program (Version 3.3) (McGarigal Planners and designers need efficient tools to quantita- et al., 2002a), was used to quantify the local landscape tively evaluate and compare the impact of alternative pattern across space (McGarigal and Cushman, 2002). plans and designs so that more informed development Quantified landscape metrics, combined with gradient choices could be made. analysis, were judged to be the most appropriate way The theory of landscape ecology has spawned sig- of relating the spatial pattern of urban green spaces to nificant innovation in landscape planning and design urbanization and underlying human processes, and of (Nassauer, 1999). The surge of interest in landscape determining their influence on ecological attributes of ecology is discernable in recent efforts to incorporate the environment. a landscape perspective into policies and guidelines Jinan City, a medium-sized provincial capital city for managing public lands (McGarigal and Cushman, representative of metropolitan areas in China, was cho- 2002). Landscape ecology is particularly well suited to sen as the study area. Since the Reform and Open Policy studying urban green spaces because both man-made and the Urban Land Reform were initiated in the 1980s, and remnant natural areas are considered (Design Cen- Chinese cities have been facing rapid urban expansion ter for American Urban Landscape, design brief, 2003); and development (Cheng and Masser, 2003). China’s F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 149 urban population in 2001 was 37.7% of the total popu- (Fig. 1). Jinan has a typical warm-temperate, semi- lation; the urban population is projected to reach 75% of humid, continental monsoon climate and well-defined the total by 2050 (Chinese Mayor’s Association, 2002; seasons. The mean annual temperature is 14 ◦C, and Zhang et al., 2004). This rapid urbanization has pro- the average mean precipitation 650–700 mm. Jinan foundly transformed the spatial pattern of urban areas. also has a special geological structure. Underground The government is now planning green spaces that will streams from Taishan flow along the limestone strata be integrated with urban development. This process to Jinan. The streams are halted to the north by igneous began in the 1980s, and the majority of cities imple- rocks and emerge in the form of numerous springs. mented green planning in the 1990s. Since the estab- There are at least 72 famous springs, and so it is lishment of its urban master plan (covering 1996–2010) known as the “City of Springs”. The natural vegetation in 1996, especially, the “Great Changes in Five Years” is deciduous broadleaf and evergreen coniferous policy in 1997: Jinan Wu Nian Da Bian Yang Fang An, forest. Human activities and influence have resulted in which states that “the local government and the citi- most of the original natural vegetation (such as Salix zens of Jinan City should fight their way to optimize babylonica L.) being partially or completely removed. the urban spatial pattern and improve the urban envi- As a result, the landscape that was once described in ronment in a period of about 5 years (1997–2001)”, the novel “Lao Can You Ji” as one in which “springs Jinan City has carried out five phases of greening work and willows are found in every courtyard” (Liu, (Jinan Planning Bureau, 2003), resulting in tremen- 1903) is all but gone. Today the dominant species are dous changes to urban spatial patterns. These changes Platanus orientalis L., Sophora japonica, Populus have resulted in both congruence and conflicts between tomentosa Carr., Platycladus orientalis Franco., and green space establishment and urban development; as bush-grass communities (Jinan Landscape Bureau, a result, the government has revised the current urban 2001). master plan. The revised plan will be in effect from Jinan City is the capital of Shandong province. 2004 to 2020. Over 2600 years have passed since the rise of the In this study, we integrated gradient analysis with Xihe culture. At present, the urban administration of landscape pattern metrics to characterize urban green Jinan City consists of six districts (Lixia, Licheng, space patterns quantitatively from 1989 to 2004. We Huaiyin, Tianqiao, Changqing, and Shizhong), three addressed the following questions: (1) How have urban counties (Pingyin, Shanghe, and Jiyang), and one city green spaces changed during the last 15 years? (2) How (Zhangqiu). Jinan has sprawled greatly in the last 50 did urban green spaces change according to distance years, expanding its total area to 8117 km2 (Fig. 1b); the from the urban center? (3) Can urbanization and the total population has also grown dramatically, increas- influence of government policies be reflected through ing from 3.19 million in 1952 to 5.75 million in 2002. the use of gradient analysis of urban green spaces? The areal extent of built-up areas has increased from We have tried to show that quantifying the pattern 24.6 km2 in 1949 to more than 190 km2 in 2003 (Jinan of urban green spaces through “moving window” and Statistics Bureau, 2003). spatio-temporal gradient analysis can effectively link Jinan City has a monocentric and irregular radio- pattern and process and will be an important prereq- sprawl pattern (Jiang and Zhang, 2003) (shown by the uisite for analyzing the ecological and socioeconomic sprawl of built-up area from 1956 to 2004, ring roads functions of urban green spaces. It may also provide and road network in Fig. 1c), which is the same as substantial information that is useful for eco-urban most medium-sized cities in China (Au and Henderson, planning. 2002; Pan and Zhang, 2002; Deng and Huang, 2003). With growth to the south constrained by hilly topography and to the north by the Yellow River (Fig. 1c), the 2. Study area Jinan Planning Bureau’s 2004–2020 Master Plan pro- poses to expand eastward, with the urban area expand- Jinan is located within Shandong Province, to the ing to the 3rd ring road; this will increase the urban north of Taishan and south of the Yellow River, at lat- area to 400 km2 (Fig. 1c). The area examined in this itude 36◦32–36◦51N, longitude 116◦49–117◦14E study includes the entire inner part of the second ring 150 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164

Fig. 1. The study area showing: (a) location of Shandong Province in China; (b) location of Jinan City; (c) urban area of Jinan City in 2004 as delineated in the urban master plan (2004–2020) and the monocentric spatial structure. road, and covers 149.2 km2, which is the core area 3. Data and methods in both the “past” and “future” Master Plans. In the 1996–2010 Master Plan and the “Great Changes in Primary data sources used in this study include: (1) Five Years” policy, the Jinan People’s Government 1989 spot images (resolution 10 m; 1 band), 1989 land- proposed a “One Ring, Three Greenbelts and Nine sat images (resolution 30 m, 7 bands), 1996 spot images Wedges” green network system and sought to build a (resolution 10 m, 1 band and resolution 20 m, 4 bands), “National Garden City”. In the second ring road area, an and 2004 spot images (resolution 10 m, 4 bands); (2) a “Inserting Green Wedges, Connecting Green Network, topographic map (1:10,000) created in 2000; (3) urban VerticalGreening” policy was implemented. In the past planning data and census data obtained from the Jinan 8 years (1996–2004), green space area has increased to Planning Bureau and the Statistics Bureau (auxiliary more than 1200 ha. information). F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 151

To obtain urban green space information for the ESRI) spatial analysis. To capture the synoptic features 3 years (1989, 1996 and 2004), the spot and landsat of the landscape, four landscape metrics were calcu- images were rectified and georeferenced to the Univer- lated (percent of landscape (PLAND), mean patch size sal Transverse Mercator (UTM) coordinate system by (MPS), patch density (PD) and landscape shape index using the ERDAS Imagine system (Version 8.5, ESRI, (LSI)), using FRAGSTATS (Version 3.3) (McGarigal Atlanta, GA 30329-2137, USA). The ERDAS image et al., 2002a). system was also used to perform a resolution merge To detect the gradient changes of the urban green of data from the 1989 and 1996 images, because the spaces, we first conducted a “moving window” analysis resolution of these images differed. The urban green supported by FRAGSTATS,which quantified the green space categorical maps were created by manual inter- space pattern using landscape metrics in the local area. pretation based on the ArcInfo (Version 8.2, ESRI, Both class- and landscape-level metrics were computed Redlands, CA 92373-8100, USA) platform, combined using a 500-m radius window size. Several trials were with field surveys and ground truthing as necessary. made to test the impact of window size on the vari- The urban green space data set was reclassified into ous landscape metrics, and to improve the smoothing nine types from the standard for classification of urban effect. Results with a 500-m radius window seemed to green space in China (Ministry of Construction, PR reveal fluctuations in most metrics. This may be viewed China, 2002), based on urban green space functions, as a good example of the effects of scale changes on land use type and ownership (Table 1; Fig. 2). These landscape metrics (Turner et al., 1989; Hunsaker et al., vector data were then converted to raster format with a 1994; Jelinski and Wu, 1996; Wu et al., 2000; Luck pixel size of 10 m × 10 m, using ArcMap (Version 8.2, and Wu, 2002). The window moved over the whole

Table 1 Reclassification of urban green space Original urban green Reclassified patch type Abbreviation Description space type (according to Ministry of Construction, PR China, 2002) Public park Public park PU Open to the public (includes community parks); provides education, pleasure and recreation; has natural and planted vegetation Plaza-green space PL Open to the public; provides open space, recreational opportunities; has planted vegetation, seldom trees, most is shorter shrubs and grassland; low diversity Nursery Nursery NU Propagating and cultivating vegetation, breeding and supplying saplings for urban greening Green buffer Green buffer GR Linear corridors protecting high-voltage transmission lines, screening wind and cleansing pollutants; with planted vegetation Attached green space Attached green space AT Attached to industrial, commercial, utility land etc.; has planted vegetation; low diversity Residential green space RE Attached to residential areas, including those planted and maintained by the individual residents; includes communal green space serviced the local community (excludes PU and PL). Provides aesthetic, amenity-recreation venues; limited plant diversity Roadside green space RO Linear corridors between sidewalks, curbs or island patches in crossroads; serve to buffer people from traffic, screen noise and solar radiation, etc.; has planted vegetation; limited plant diversity Other green space Riparian green space RI Linear corridors along watersheds; primarily natural habitat type; often high plant diversity Scenery forest SC Open to the public; serve to protect and preserve flora, fauna and providing scenic amenities; a mosaic of remnant or naturalized habitat types 152 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164

Fig. 2. Distribution of types of urban green space in 1989, 1996 and 2004.

Table 2 Deﬁnitions of landscape metrics (based on McGarigal et al., 2002b) Landscape metrics Abbreviation Description Units Range Class area CA CA equals the sum of the areas (m2) of all patches Hectares CA > 0, no limit of the corresponding patch type divided by 10 000 (to convert to hectares) Patch density PD The number of patches per 100 ha Number per PD>0, 100 hectares constrained by cell size Percent of landscape PLAND The proportion of total area occupied by a particular Percent 0 < PLAND < 100 patch type; a measure of landscape composition and dominance of patch types Mean patch size MPS The area occupied by a particular patch type divided Hectares MPS > 0, no limit by the number of patches of that type Largest patch index LPI LPI equals the area (m2) of the largest patch of the Percent 0 < LPI < 100 corresponding patch type divided by total landscape area (m2), multiplied by 100 (to convert to a percentage) Landscape shape index LSI The total length of edge involving the corresponding None LSI ≥ 1, no limit class divided by the minimum length of class edge for a maximally aggregated class, a measure of class aggregation or clumpiness Euclidian mean nearest MNN MNN equals the distance (m) mean value over all Meters MNN > 0, no limit neighbor distance urban green space patches to the nearest neighboring patch, based on shortest edge-to-edge distance from cell center to cell center Patch richness PR The number of patch types in the landscape; a None PR ≥ 1, no limit measure of diversity of patch types F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 153

Fig. 3. Grid map of the patch density metric (number of patches/km2) at the landscape level, obtained by the “moving window” method. The map for 2004 shows the overlay of 286 sampling points, distributed at 200-m intervals in eight directions from the urban center to the fringe. landscape, calculating the selected metric within the scape metrics fluctuate and display spatial signature window and returning that value to the center cell and from the urban center to the periphery. As per the result outputting a new continuous surface grid map for each explained in the following sections, the oversampling selected metric (McGarigal and Cushman, 2002). In in the city center will not affect the final conclusion, this research, eight landscape metrics were used to such as lower patch richness (PR) and patch density in quantify the urban green space pattern (Table 2). Three the center (Fig. 6), and the undersampling in the periph- grid maps showing the patch density metric are shown ery can still cover the typical areas, such as the east, in Fig. 3. Based on the grid maps of each landscape northwest and southeast (Figs. 4–7). metric, 286 spatial sample points were selected at a distance of 200 m in eight directions from the urban center (Fig. 3). Even though, radially oriented sampling would 4. Results result in an oversampling of the city center and undersampling of the periphery (Fig. 3), it could describe the 4.1. Synoptic analysis of green space pattern urban green space pattern better. This is because, firstly, Jinan is a single dominating center city with a radio- A comparison of data from the 3 years indicates a sprawl pattern (Fig. 1c). Secondly, the samples cannot significant overall increase in the area of urban green cover the whole study area but it is best to cover a larger space, as well as differing growth rates. The total green area, especially the typical areas. Therefore, we con- space area in 1989, 1996 and 2004 was 4302.2, 4337.7 ducted the eight-direction sampling to cover a higher and 5538.6 ha, respectively (calculated by class area percentage of the study area as well as the typical areas (CA) metric, Table 2); the growth rate for 1996–2004 (e.g., central business district in the center, mountain- (27.7%) was much higher than that for 1989–1996 ous areas in the south and southeast, and watershed area (0.8%). Among all of the green space types (Table 1), in the northwest) compared with a normal approach “attached green space”, “residential green space” and in two or four directions (Luck and Wu, 2002; Zhang “public park” increased the most. But “scenery for- et al., 2004). And finally, the “moving window” anal- est” decreased by about 66.9 ha from 1996 to 2004, ysis supported by FRAGSTATS combined with land- although it increased by 181.2 ha compared with 1989. scape metrics is a suitable approach for such analysis. Overall, the rank of percentage of landscape (Table 3a) For example, as shown in Figs. 4–7, the curves of land- did not change very much. In 2004 and 1996, the first 154 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164

Fig. 4. Class-level gradient changes in “residential green space” in 1989, 1996 and 2004, plotted in eight directions from the urban center to the fringe. Rows show gradient change along the four major axes [(A) south–north, (B) southwest–northeast, (C) west–east, and (D) southeast–northwest]; each column displays a different metric [(a) percent of landscape (%), (b) patch density (number of patches /100 ha), and (c) landscape shape index]. The values were obtained through sampling at 200-m intervals from class-level grid maps generated by the moving window method. F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 155

Fig. 5. Gradient changes in selected metrics at the landscape level in 1989, 1996 and 2004, plotted in four directions from the urban center to the fringe. Rows show gradient change along two major axes [(A) south–north and (B) southwest–northeast]; each column displays a different metric [(a) patch density (number of patches/100 ha), (b) landscape shape index, (c) euclidian mean nearest neighbor distance (m), and (d) patch richness]. The values were obtained through sampling at 200-m intervals from landscape-level grid maps generated by the moving window method. 156 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164

Fig. 6. Gradient changes in selected metrics at the landscape level in 1989, 1996 and 2004, plotted in four directions from the urban center to the fringe. Rows show gradient change along two major axes [(C) west–east and (D) southeast–northwest]; each column displays a different metric [(a) patch density (number of patches/100 ha), (b) landscape shape index, (c) euclidian mean nearest neighbor distance (m), and (d) patch richness]. The values were obtained through sampling at 200-m intervals from landscape-level grid maps generated by the moving window method. F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 157

Fig. 7. Gradient changes in the largest patch index metric at the landscape level in 1989, 1996 and 2004, plotted in eight directions from the urban center to the fringe. Each graph shows gradient change along one of four major axes [(A) south–north, (B) southwest–northeast, (C) west–east, and (D) southeast–northwest]. The values were obtained through sampling at 200-m intervals from landscape-level grid maps generated by the moving window method.

Table 3 Synoptic landscape characteristics of green spaces in Jinan City (explanation of landscape types provided in Table 1) Year Type

SC RE AT RO PU RI GR PL NU (a) Percent of landscape (%) 1989 10.303 (1) 5.093 (2) 2.833 (4) 2.503 (6) 2.769 (5) 3.041 (3) 1.119 (7) 0.307 (9) 0.872 (8) 1996 11.967 (1) 6.234 (2) 3.210 (3) 3.178 (4) 1.315 (6) 2.069 (5) 0.595 (7) 0.139 (9) 0.373 (8) 2004 11.518 (1) 8.046 (2) 6.218 (3) 4.319 (4) 2.839 (5) 2.295 (6) 0.912 (7) 0.494 (8) 0.489 (9) (b) Mean patch size (ha) 1989 102.466 (1) 1.979 (9) 2.322 (7) 2.146 (8) 15.885 (4) 4.629 (6) 9.817 (5) 15.273 (3) 43.370 (2) 1996 111.566 (1) 0.723 (9) 0.877 (8) 1.378 (7) 7.542 (3) 2.286 (6) 5.550 (4) 4.140 (5) 13.910 (2) 2004 81.818 (1) 1.369 (9) 1.999 (7) 1.727 (8) 10.860 (2) 3.200 (5) 2.896 (6) 3.878 (4) 10.413 (3) (c) Patch density (patch number/100 ha) 1989 0.101 (7) 2.574 (1) 1.220 (2) 1.166 (3) 0.174 (5) 0.657 (4) 0.114 (6) 0.020 (8) 0.020 (8) 1996 0.107 (6) 8.621 (1) 3.660 (2) 2.306 (3) 0.174 (5) 0.905 (4) 0.107 (6) 0.034 (7) 0.027 (8) 2004 0.141 (7) 5.879 (1) 3.111 (2) 2.507 (3) 0.261 (6) 0.717 (4) 0.315 (5) 0.127 (8) 0.047 (9) (d) Landscape shape index 1989 7.219 (6) 24.627 (1) 16.495 (3) 21.129 (2) 7.275 (5) 15.749 (4) 6.162 (7) 3.537 (9) 4.572 (8) 1996 5.792 (6) 42.597 (1) 28.215 (3) 36.986 (2) 10.132 (5) 22.364 (4) 5.751 (7) 3.802 (9) 4.960 (8) 2004 7.575 (8) 51.201 (1) 35.684 (3) 48.057 (2) 9.138 (6) 25.528 (4) 11.346 (5) 7.930 (7) 3.515 (9) 158 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 four were the same, but “riparian green space” and quence, the changing spatial signature of the different “nursery” decreased and were, respectively, replaced metrics provides information about residential devel- with “public park” and “plaza-green space”. Compared opment, and has implications for the study of the pro- with 1989, the decrease in the rank of “riparian green cess of urbanization and the influence of government space” was the most obvious, from third to sixth place. policy. “Residential green space” always had the highest value In general, in each of the 3 years, the percent- of landscape shape index and patch density and the age of landscape, patch density and landscape shape lowest mean patch size (Table 3b–d), indicating the index showed similar patterns from the urban center highest degree of fragmentation. On the other hand, to the fringe (defined by the second ring road), with “scenery forest” had the highest value of percentage of a low value in the urban center and two peaks on landscape and mean patch size and lower patch density both sides; an exception to this can be observed from and landscape shape index (Table 3a–c), suggesting southeast to northwest (Fig. 4D). The ring structure its dominance and high degree of aggregation. At the of Jinan City (Jiang and Zhang, 2003) is implied by same time, between 1989 and 2004, values for the patch this pattern, with the central business district in the density, landscape shape index and percentage of land- center and the residential areas on the two sides. In scape of “roadside green space” increased gradually, the southeast and northwest, the mountains and water- showing that connectivity increased and that a green shed, respectively, limit the residential area distribution space network is coming into being (Table 3a–c). (Fig. 1c). The different rates of increase in urban green space Close examination of the data reveals the unique and the change of landscape metrics were both related spatial signature of each metric in the different years to three developments: implementation of the “Great (Fig. 4). In the vicinity of the urban center, along the Changes in Five Years” policy from 1997, implemen- south–north direction (from −3to−1 km and from 2 to tation of the “urban master plan 1996–2010”, and ongo- 3 km), the southwest–northeast direction (from −4to ing urbanization. Jinan’s urban greening policy resulted −3 km and from 1 to 3 km), and the west–east direction in an increase in most green space types and changes (from −4to−1 km), values for percentage of land- in spatial patterns. Urban sprawl has encroached on scape showed a progressive increase between 1989 and green spaces in recent years, particularly on “scenery 2004 (2004 > 1996 > 1989). Values for patch density forest” and “riparian green space”. This type of syn- and landscape shape index, measured along the same optic analysis by landscape metrics can provide gen- directions and distances, showed a different tempo- eral information on urban green space changes. In ral pattern, in which 1996 > 2004 > 1989 (Fig. 4A–C). the absence of other information, however, it is dif- This indicates an increase in “residential green space”, ficult to link changes in green space patterns in local with a higher percentage cover and patch density, and areas accurately with the processes that produced these a degree of disaggregation. These increases suggest changes. Accordingly, in the next section, we describe that government policy is having an effect. Although it how gradient analysis was employed to address this was impossible for residential areas to sprawl in these problem. places, it was still possible to “insert green wedges”. This has been considered to be a good way to increase 4.2. Gradient analysis with class-level metrics green spaces, and was implemented by the government in the past. To reduce redundant data and information and sup- Near the fringe, in the north (Fig. 4A-a and port effective interpretation, “residential green space” A-c), northeast (Fig. 4B-a and B-c), east (Fig. 4C-a (defined in Table 1) has been evaluated in combina- and C-c) and northwest (Fig. 4D-a and D-c), the tion with three metrics (percent of landscape, patch percentage of landscape and landscape shape index density and landscape shape index). A comparison of displayed similar peaks and the same temporal the gradient change for each of these metrics for the 3 ranking: 2004 > 1996 > 1989. patch density had almost years studied, in eight directions from the urban cen- the same rank. This indicates that the urbanization ter, is given in Fig. 4. “Residential green space” was process is resulting in a greater number of residents closely associated with “residential area”; as a conse- living outside of the urban center, and has resulted F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 159 in the construction of more eco-communities. Eco- On the whole, when comparing 1989 with 1996 and communities have been established in an attempt to 2004, the patch density had a lower value, but euclidian realize sustainable community development, and are mean nearest neighbor distance (MNN) had a higher characterized by human-scale, sustainable settlement value and showed acute fluctuation in most places, based on ecological balance, community self-reliance, except for a slight change around the urban center and participatory democracy (Roseland, 2000). Chi- (Figs. 5 and 6). This indicates that in 1989, green spaces nese eco-communities share these aims, and although were unevenly distributed and had a lower degree of they have failed to achieve them completely, they often connectivity. The largest patch index (LPI) (Fig. 7) had offer advantages in terms of their location and environ- many more peaks in 2004 than it did in 1996 and 1989, ment, and are typically built in compliance with local demonstrating that the increase in urban green spaces urban planning policy (Huang and Bai, 2003; Cheng, is related to the “inserting green wedges” policy, with 2004). In the case of Jinan, the rate of forest coverage individual green space patches closing on each other was required to be not less than 30% (Jinan Planning and becoming more and more connected. When plot- Bureau, 2003). The significant peak in the east reflects ted, the patch richness of the whole landscape takes on the fact that urban sprawl is constrained by topography nearly an “M” shape, with a lower value near the urban and by the need to protect the spring water resources in center, higher values in intermediate areas, and lower the south. The “eastward sprawl” called for in the new values at the fringe. This indicates a monotone distri- Master Plan has triggered a rapid development of resi- bution of green space in the urban center. At the fringe, dential districts in recent years. Green space was often “scenery forest” has a higher area with some scattered designated to provide amenities to residents in the form “green buffer” or “riparian green space”. of recreational benefits. These areas were called “gar- In the south (between −3 and −2 km) and south- den districts”, “eco-districts” or “green-eco-districts”. east (between −7 and −5 km) (Fig. 7A and D), the The analysis showed a decline in the three met- largest patch index peaked in each of the 3 years stud- rics near the southwest fringes, where the temporal ied, reflecting the fact that the “Green Core” (com- rankings were: 1989 > 1996 > 2004 for percentage of posed primarily of “scenery forest” and “public park”) landscape and 1996 > 1989 > 2004 for patch density remained almost unchanged in each area. In addition, and landscape shape index (Fig. 4B). This is due to the in the southeast, the largest patch index value was dispersed urban sprawl that characterized these areas nearly 100% in each of the 3 years; this resulted from before 1989, and the “filling-in” process that occurred an increase in green space in the mountainous area subsequently, resulting in a higher residential density. before 1996. However, at the fringe of the two sides, However, near the south fringe (around −6 km), the the landscape shape index displayed peaks in 2004 temporal rankings were 2004 > 1989 > 1996; the peaks while rankings for patch density and euclidian mean in the southeast for the three metrics all suggest an nearest neighbor distance were 2004 > 1996 > 1989 encroachment on “scenery forest”, and indicate a grad- (Figs. 5 and 6). This indicates a higher disaggregation ually increasing population. Urban sprawl is beginning over time and illustrates that urban sprawl is enveloping to consume the green spaces (Fig. 1c). An increasing and encroaching on “scenery forest” and “public park”. number of people prefer to live close to green spaces, In 1989, in the north (between 4 and 7 km) and north- despite higher house prices and firm measures imple- east (between 3 and 7 km), these areas comprised a mented by the government to control such movements “Green Core”, which is evident in the significant peaks with the aim of protecting spring water sources. in the largest patch index, but these peaks had decreased by 1996 and the index remained low in 2004 (Fig. 7A 4.3. Gradient analysis with landscape-level and B). This represents the separation of green space metrics versus a dominant “Green Core”, and indicates that a period of intensive urbanization took place between In order to quantify changes in the spatial pattern 1989 and 1996. Subsequently, the process continued of green spaces over the study period at the landscape but slowed, causing separation of and encroachment level, five landscape metrics were combined with all on green spaces, especially on riparian green space. nine green space types (Figs. 5–7). This is also demonstrated by increases in the patch 160 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 density (Fig. 5A-a and B-a). In the west, from the and peaks on both sides. Close examination indicates urban center towards the west (up to −4 km), the largest distinctive synoptic characters and spatial signatures patch index and landscape shape index have a rank of in the local area for each of the 3 years studied. At 2004 > 1996 > 1989, while the euclidian mean nearest the landscape level (Figs. 5–7), comparison of 1989 neighbor distance rank is almost 1989 > 1996 > 2004 with 1996 and 2004 found the increases in patch den- (Figs. 6 and 7C). This accurately reflects the spatial sity, declines in euclidian mean nearest neighbor dis- growth of the “Green Core” and the decrease in distance and an increase in the number of peaks in the tance between individual green patches. largest patch index, which is indicative of growth in urban green space area and increased connectivity. Of all the landscape metrics examined, these three 5. Discussion and conclusions provided particular insight into the degree of spatial aggregation. The combined and integrated application of remote Overall, both the class- and landscape-level metrics sensing, landscape metrics, and gradient analysis rep- indicated that dramatic change occurred over the course resented an innovative approach for the study of spatio- of the study period, including the allocation of new temporal gradient change of urban green spaces. This urban green space units (shown by increasing patch study put forward three questions, and sought to answer density, landscape shape index, patch richness, and these by measuring the gradient change in urban green euclidian mean nearest neighbor distance), growth of space. In the process, we have demonstrated that urban- “Green Cores” (represented by increased largest patch ization and the influence of government policy can index, and euclidian mean nearest neighbor distance), be discerned through the quantification of the spatio- and a decrease in original green spaces (revealed by temporal gradient of urban green spaces. percentage of landscape).

5.1. Change in spatio-temporal pattern of green 5.2. Reﬂecting urbanization and the inﬂuence of space in Jinan City governmental policy

A significant change in Jinan City urban green This study explored the possibility that urbaniza- spaces over the past 15 years has been demonstrated tion and the influence of government policy can be through both synoptic and gradient analysis, at land- reflected in a gradient analysis of urban green spaces. scape and class level, using eight landscape metrics. In our study area, the process of urbanization has con- The increase in green space area was clearly demon- sisted primarily of the filling-in of existing urban areas, strated by the class area metric. To explain the differ- as well as a degree of urban sprawl on the fringe. ing rates of increase and the spatial pattern for each The urban sprawl of Jinan is clearly constrained by green space type, the percentage of landscape, mean its topography, but urban planning and policy deci- patch size, patch density, and landscape shape index (at sions also restrict and guide urban spatial growth and class-level) were compared (Table 3). Gradient analy- the direction in which development occurs. These in sis (shown in Figs. 4–7) was used to link the spatial turn affect the spatio-temporal pattern of urban green information with local areas, and made it possible to spaces. compare changes that occurred in specific metrics and Policies concerning urban development and urban particular locations over the course of the study period. greening appear to have had a significant impact on This enhanced our ability to link patterns and processes. the urban green space structure of the study area. At the class level (Fig. 4), the spatio-temporal sig- The Master Plan (1996–2010) and “Great Changes in natures for residential green space examined through Five Years” policy (adopted in 1997) clearly increased various metrics were shown to be distinctive but not the urban green space area and resulted in a marked unique. For example, comparison of the patch density, alteration in the urban green space structure, includ- landscape shape index, and percentage of landscape ing changes in “residential green space” in the vicin- over the study period suggested a largely monotonic ity of the urban center, increased patch density and gradient change with a low value in the urban center connectivity, and decreased euclidian mean nearest F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 161 neighbor distance (Table 3; Figs. 4–7). In response space on in-migration and associated land development to growing concerns about the undesirable impacts (Wu and Plantinga, 2003). The involvement of the gov- of sprawl on urban green spaces, especially in the ernment appears to be very important in optimizing south and southeast of the study area, a wide range green space patterns and preventing encroachment on of policy instruments have been developed to man- the remaining green spaces. age urban growth and to protect green spaces from development expressed as “conserve the springs”, 5.3. Importance of spatio-temporal gradient such as “programes for conserving springs” (Jinan analysis of urban green spaces Planning Bureau, 1996), “ordinance for protecting water resources of Jinan City” (State Council of Jinan Landscape ecology has emerged over the past City, 2001) and “regulation for conserving famous decade as the study of pattern–process relationships springs” (Jinan Landscape Bureau, 2004). The east- based on a patch mosaic model (Turner et al., 2001; ward sprawl allowed for in the new Master Plan has McGarigal and Cushman, 2002). Consequently, most triggered rapid development in the east, which may landscapes studied through landscape ecology have assist in reducing population pressure in the south. been defined by a patch mosaic map, with the land- At the same time, the regulation of urban greening scape quantified over the entire study area with met- (established in 1997) prescribes the level of greening rics. Quantitative information about how variables coverage required in new built-up areas (such as eco- vary through time and space is then lost, however communities). This has contributed to the increase of (McGarigal and Cushman, 2002). The averaging of “residential green space” in the east (Fig. 4). metrics over an entire study area may lead to incor- Urbanization has changed urban land use patterns, rect interpretations of the causal dynamics in the region resulted in the occupation of green spaces by urban (i.e., the synoptic changes reflected cannot be related development, and created new living conditions. The to specific locations or linked with a visualized spa- process of urbanization has made people more aware tial interpretation; Herold et al., 2002). To solve this of their living environment, which has resulted in problem, our study developed spatially explicit land- encroachment on green spaces. Evidence of this is pro- scape metrics through the use of the “moving window” vided by the increase in “residential green space” and method. This can provide rich quantitative information the landscape-level changes observed near the east, about the structure and pattern of urban green spaces at south and southeast fringes. Green space amenities the local landscape level, showing a better link between attract migrants (this is particularly the case in the pattern and process, and can effectively capture the south and southeast), and increasing numbers of peo- dynamic changes. ple prefer to live close to green spaces, despite the The quantified and spatially explicit urban green higher cost of living and the firm measures imple- space pattern is an important basic ingredient for ana- mented by the government to control such movements lyzing the ecological and socioeconomic functions of to protect spring water sources. At the same time, near urban green space. Green space amenities can affect the southwest fringes, changes in “residential green property prices (Geoghegan et al., 1997). Linking the spaces” (shown by percentage of landscape, patch quantified and spatially explicit urban green space density and landscape shape index) can be related to pattern with an economic model, such as the hedo- the dispersed urban sprawl that existed before 1989, nic price model combined with property characters, and the filling-in process that has occurred since then can help clarify the relationship between green space (Fig. 4). The increase of “residential green space” dur- and economic values (Geoghegan, 2002; Morancho, ing the study period in the north, northeast, east and 2003). Finally, developing an understanding of the northwest, and the reduction of the “Green Core” at the dynamic spatial patterns of green spaces can improve landscape level is indicative of different urban sprawl our ability to assess and create future planning scenar- processes. ios by combining appropriate spatial models, such as It is clear, however, that urban green spaces are the Cellular Automata model. Improving our ability to closely linked with urban policies and urban develop- relate pattern to process is essential for understanding ment. Urbanization has magnified the effects of green urban green space systems, and this cannot be done 162 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 effectively without first quantifying spatial patterns in Design Center for American Urban Landscape, Design Brief, 2003. local areas. The Diversity of Green Spaces. University of Minnesota. Avail- able at the following website: http://www.designcenter.umn.edu/ reference ctr/publications/designbriefs/pdfs/db2.pdf. Dramstad, W.E., Olson, J.D., Forman, R.T.T., 1996. Landscape Ecol- Acknowledgements ogy Principles in Landscape Architecture and Land-use Plan- ning. Harvard University Graduate School of Design, American We thank Prof. Yueguang Zong in Nanjing Univer- Society of Landscape Architects. Island Press, Washington, DC. sity for his valuable suggestion; Prof. Xiuzhen Li for Dunn, C.P., Sharpe, D.M., Guntenspergen, G.R., Stearns, F., Yang, valuable comments about selecting the landscape met- Z., 1990. Methods for analyzing temporal changes in landscape pattern. In: Turner, M.G., Gardner, R.H. (Eds.), Quanti- rics; Dr. Keiko Nagashima, Dr. Jhonamie A. Mabuhay, tative Methods in Landscape Ecology. Springer, New York, pp. and Mr. Martin Ward for linguistic assistance; Mr. 173–198. Chonggang Xu and Mr. Zaiping Xiong for valuable Forman, R.T.T., Godron, M., 1986. Landscape Ecology. Wiley, New advice regarding GIS technology; Mr. Haiwei Yin for York. data interpretation; Dr. Akira Kikuchi for helpful dis- Gardner, R.H., Milne, B.T., Turner, M.G., O’Neill, R.V., 1987. Nat- ural models for the analysis of broad-scale landscape pattern. cussions. This research was supported by the COE (The Landscape Ecol. 1, 19–28. 21st Century Center of Excellence) Program, Social Geoghegan, J., 2002. The value of open spaces in residential land Capacity Development for Enviromental Management use. Land Use Policy 19, 91–98. and International Cooperation in Hiroshima Univer- Geoghegan, J., Wainger, L.A., Bockstael, N.E., 1997. Spatial land- sity. Special thanks to the anonymous reviewers and scape indices in a hedonic framework: an ecological economics analysis using GIS. Ecol. Econ. 23, 251–264. the editor for their valuable comments to improve our Gobster, P.H., Westphal, L.M., 2004. The human dimensions of manuscript. urban greenways: planning for recreation and related experi- ences. Landscape Urban Plan. 68, 147–165. Gordon, D., 1990. Green Cities: Ecologically Sound Approaches to References Urban Space. New York Black Rose Books Press, Montreal. Groot, R.S.D., 1994. Environmental functions and the economic value of natural ecosystems. In: Jansson, A.M., Hammer, Attwell, K., 2000. Urban land resource and urban planting—case M., Folke, C., Constanza, R. (Eds.), Investing in Natural studies from Denmark. Landscape Urban Plan. 52, 145– Capital—The Ecological Economics Approach to Sustainabil- 163. ity. Island Press, Washington, DC. Au, C., Henderson, V., 2002. How migration restrictions limit Herold, M., Goldstein, N.C., Clarke, K.C., 2002. The spatiotempo- agglomeration and productivity in China? NBER Working Paper, ral form of urban growth: measurement, analysis and modeling. Beijing. Remote Sens. Environ. 86, 286–302. Beatley, T., 2000. Green Urbanism: Learning from European Cities. Huang, C.H., Bai, G.R., 2003. On the connotation and indexes of Island Press, Washington, DC. residential eco-community. Human Geogr. 18 (1), 53–56 (in Chi- Binford, M.W., Buchenau, M.J., 1993. Riparian greenways and water nese). resources. In: Smith, D.S., Hellmund, P.C. (Eds.), Ecology of Hunsaker, C.T., O’Neill, R.V., Jackson, B.L., Timmins, S.P., Levine, Greenways. University of Minnesota Press, Minneapolis, pp. D.A., Norton, D.J., 1994. Sampling to characterize landscape 69–104. pattern. Landscape Ecol. 9, 207–226. Cheng, J.Q., Masser, I., 2003. Urban growth pattern modeling: a Jelinski, D.E., Wu, J., 1996. The modifiable areal unit problem case study of Wuhan city, PR China. Landscape Urban Plan. 62, and implications for landscape ecology. Landscape Ecol. 11, 199–217. 129–140. Cheng, S.D., 2004. Concept and practice of ecological community. Jiang, L.G., Zhang, Z.L., 2003. Analysis on the ring structure of Eng. J. Wuhan Univ. 37 (3), 83–97 (in Chinese). urban land use in Jinan City. Areal Res. Dev. 22 (4), 73–76 (in Chiesura, A., 2004. The role of urban parks for the sustainable city. Chinese). Landscape Urban Plan. 68, 129–138. Jim, C.Y., Chen, S.S., 2003. Comprehensive greenspace planning Chinese Mayor’s Association, 2002. The Report of 2001–2002 Chi- based on landscape ecology principles in compact Nanjing city, nese Urban Development. Xiyuan Publishing House, Beijing (in China. Landscape Urban Plan. 65, 95–116. Chinese). Jinan Landscape Bureau, 2001. Planning of Landscape and Green Conine, A., Xiang, W.N., Young, J., Whitley, D., 2004. Planning for Space System in Jinan City. Jinan Landscape Bureau and Jinan multi-purpose greenways in Concord, North Carolina. Landscape Institute of Urban Planning and Design. Local Record Press, Urban Plan. 68, 271–287. Jinan (in Chinese). Deng, F.F., Huang, Y., 2003. Uneven land reform and urban Jinan Landscape Bureau, 2004. Regulation for conserving famous sprawl in China: the case of Beijing. Prog. Plan. 61, 211– springs. Local Record Press, Jinan (in Chinese). 236. F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 163

Jinan Planning Bureau, 1996. Master Plan of Jinan (1996–2010)— Pouyat, R.V., McDonnell, M.J., Pickett, S.T.A., 1995. Soil charac- Programes for Conserving Springs. Local Record Press, Jinan teristics in oak stands along an urban–rural land-use gradient. J. (in Chinese). Environ. Qual. 24, 516–526. Jinan Planning Bureau, 2003. Great Changes of Jinan in Five Years. Roseland, M., 2000. Sustainable community development: integrat- Jinan Press, Jinan (in Chinese). ing environmental, economic, and social objectives. Prog. Plan. Jinan Statistics Bureau, 2003. Jinan Statistical Yearbook. China 54, 73–132. Statistics Press, Beijing (in Chinese). Schell, L.M., Ulijaszek, S.J., 1999. Urbanism, Health and Human Kaplan, R., Kaplan, S., 1989. The Experience of Nature. Cambridge Biology in Industrialized Countries. Cambridge University Press, University Press, Cambridge. Cambridge. Kowarik, I., 1990. Some responses of flora and vegetation to urban- Stanners, D., Bourdeau, P., 1995. Europe’s Environment—The ization in central Europe. In: Sukopp, H., Hejny, S., Kowarik, I. Dobris Assessment. Office for Official Publications of the Euro- (Eds.), Urban Ecology: Plants and Plant Communities in Urban pean Communities, Luxembourg. Environments. SPB Academic Publishing B.V., The Hague, The State Council of Jinan City, 2001. Ordinance for Protecting Water Netherlands, pp. 45–74. Resources of Jinan City. Local Record Press, Jinan (in Chinese). Liu, E., 1903. Lao Can You Ji. The Commercial Press, Shanghai. Sukopp, H., 1998. Urban ecology—Scientific and practical aspects. Luck, M., Wu, J., 2002. A gradient analysis of urban landscape pat- In: Breuste, J., Feldmann, H., Uhlmann, O. (Eds.), Urban Ecol- tern: a case study from the Phoenix metropolitan region of USA. ogy. Springer, Berlin, pp. 3–16. Landscape Ecol. 17, 327–339. Turner, M.G., 1989. Landscape ecology: the effect of pattern on pro- McGarigal, K., Cushman, S.A., 2002. The Gradient Concept of Land- cess. Ann. Rev. Ecol. Syst. 20, 171–197. scape Structure: Or, Why Are There So Many Patches. Avail- Turner, M.G., Gardner, R.H., O’Neill, R.H., O’Neill, R.V., 2001. able at the following website: http://www.umass.edu/landeco/ Landscape Ecology in Theory and Practice. Springer–Verlag, pubs/pubs.html. New York. McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., 2002a. Turner, M.G., O’Neill, R.V., Gardner, R.H., Milne, B.T., 1989. FRAGSTATS: Spatial Pattern Analysis Program for Categori- Effects of changing spatial scale on the analysis of landscape cal Maps. Computer Software Program Produced by the Authors pattern. Landscape Ecol. 3, 153–162. at the University of Massachusetts, Amherst. Available at the Tyrvainen,¨ L., 1997. The amenity value of the urban forest: an appli- following website: www.umass.edu/landeco/research/fragstats/ cation of the hedonic pricing method. Landscape Urban Plan. 37, fragstats.html. 211–222. McGarigal, K., Ene, E., Holmes, C., 2002b. FRAGSTATS (Version Urban, D.L., O’Neill, R.V.,Shugart, H.H., 1991. Landscape Ecology. 3): FRAGSTATS Metrics. University of Massachusetts-Produced Bioscience 37, 119–127. Program. Available at the following website: http://www.umass. Wang, Y.Q., Zhang, X.S., 2001. A dynamic modeling approach to edu/landeco/research/fragstats/documents/fragstats documents. simulating socioeconomic effects on landscape changes. Ecol. html. Model. 140, 141–162. Miller, R.W., 1997. Urban Forestry: Planning and Managing Urban Whittaker, R.H., 1967. Gradient analysis of vegetation. Biol. Rev. Greenspaces, second ed. Prentice Hall, Inc., Upper Saddle River, 42, 207–264. New Jersey. Whittaker, R.H., 1975. Communities and Ecosystems. MacMillan, Ministry of Construction, PR China, 2002. Standard for Classifica- New York. tion of Urban Green Space, CJJ/T 85-2002, J185-2002. Ministry World Commission on Environment and Development, 1987. Our of Construction Press, Beijing (in Chinese). Common Future. Oxford, New York. Morancho, A.B., 2003. A hedonic valuation of urban green areas. Wu, J., Jelinski, D.E., Luck, M., Tueller, P.T., 2000. Multiscale Landscape Urban Plan. 66, 35–41. analysis of landscape heterogeneity: scale, variance and pattern Nassauer, J.I., 1999. Culture as a means of experimentation and metrics. Geogr. Inform. Sci. 6, 6–19. action. In: Weins, J.A., Moss, M.R. (Eds.), Issues in Landscape Wu, J., Plantinga, A.J., 2003. The influence of public open space on Ecology. International Association for Landscape Ecology, Fac- urban spatial structure. J. Environ. Econ. Manag. 46, 288–309. ulty of Environmental Sciences, University of Guelph, Guelph. Zhang, L.Q., Wu, J.P., Zhen, Y., Shu, J., 2004. A GIS-based gradient Ontario, Canada, pp. 129–133. analysis of urban landscape pattern of Shanghai metropolitan Nilsson, K., Randrup, T.B., 1997. Proceedings of the Eleventh World area, China. Landscape Urban Plan. 69, 1–16. Forestry Congress, Antalya, Turkey, 13–22 October. Zhu, W., Carreiro, M.M., 1999. Chemoautotrophic nitrification in O’Neill, R.V., Krummel, J.R., Gardner, R.H., Sugihari, G., Jackson, acidic forest soils along an urban-to-rural transect. Soil Biol. B., DeAngelis, D.L., Milne, B.T., Turner, M.G., Zygmunt, G., Biochem., 1091–1100. Christensen, S.W., Dale, V.H., Graham, R.L., 1988. Indices of landscape pattern. Landscape Ecol. 3, 153–162. Fanhua Kong is currently a PhD candidate of the Graduate School Pan, Z., Zhang, F., 2002. Urban productivity in China. Urban Studies for International Development and Cooperation at Hiroshima Uni- 39 (12), 2267–2281. versity. She was also a researcher invited by the 21st Century Center Pouyat, R.V., McDonnell, M.J., 1991. Heavy metal accumulations of Excellence (COE) program of Hiroshima University. Her recent in forest soils along an urban-rural gradient in southeastern New work focuses on the landscape ecology and urban green space plan- York, USA. Water Air Soil Pollut. 57, 797–807. ning and management in Northern China. Between 2000 and 2003, 164 F. Kong, N. Nakagoshi / Landscape and Urban Planning 78 (2006) 147–164 she was invited to the Institute of Applied Ecology, Chinese Academy got his Doctor of Science degree from Hiroshima University. His of Sciences (CAS) in Shenyang for research on landscape patterns professional activities include landscape ecology and EIA. His area and forest fires in the Da Hinggan Mountains. of professional interests is landscape of Asian–Pacific region. Any other item that he deems important of interesting is culture/urban Nobukazu Nakagoshi is a professor of Graduate School for Inter- landscape. He serves IALE as one of the vice presidents and also national Development and Cooperation, Hiroshima University. He Japan Assocation for Landscape Ecology as the president.