GIS-based spatial analysis for supporting the location of welfare facilities for the elderly using open data

(1)

GIS‑based spatial analysis for supporting the location of welfare facilities for the elderly using open data

著者游寧龍

著者別表示 You Ninglong journal or

publication title

博士論文本文Full 学位授与番号 13301甲第4743号

学位名博士（学術）

学位授与年月日 2018‑03‑22

URL http://hdl.handle.net/2297/00051480

doi: 10.3390/ijerph14101102

(2)

Dissertation

GIS-Based Spatial Analysis for Supporting the Location of Welfare Facilities for the Elderly Using

Open Data

Graduate School of

Natural Science & Technology Kanazawa University

Division of Environmental Design

School ID No.: 1524052008

Name:

游寧龍 (YOU Ninglong)

Chief advisor: Prof. Zhenjiang Shen

February, 2018

(3)

(4)

Abstract

This Ph.D. research aims to use the GIS-based spatial analysis and open data to support the location of welfare facilities for the elderly by two aspects of the evaluations of spatial distribution for welfare facilities for the elderly and explorations of possible solution for difficulties in implementing the location. It contributes to help the government sectors identify the location more accurately and alleviate the difficulties in implementing the location of welfare facilities for the elderly. The proposed methods in this dissertation are improved based on the needs to enhance the location support of welfare facilities for the elderly in the Tokyo metropolitan area.

In the context of obvious unbalanced relationship between supplies from welfare facilities for the elderly and demands of elderly population who cannot acquire services, where to locate welfare facilities for the elderly has been a problem long troubled the government sectors. A growing number of programs have been formulated to support the location of welfare facilities for the elderly, which focused on specific welfare facilities for the elderly. The spatial distribution of welfare facilities for the elderly cannot but be enhanced step by step due to the difficulties in the implementation of the location. Thus, how to support the location of welfare facilities for the elderly more effectively by spatial distributive evaluations and implementary explorations is still a challenge. Based on this, we first make an overall analysis of the distributive relationship between the spatial distribution of welfare facilities for the elderly and population by building a coupling degree model. It addresses the needs of improving the evaluation of spatial distribution for facilities at a relatively micro level. The result may help the government sectors analyze the reasonable degree of the relationship between the distributions of welfare facilities for the elderly and population.

Secondly, in order to further analyze the spatial distribution of specific welfare facilities for the elderly and support their locations, we continue to calculate a beds-needed index for assessing the allocation of special elderly nursing homes by introducing a parameter-improved floating catchment area method. It addresses the needs to improve the focuses on the access and potential demand of elderly population in assessing the allocation. The result may help the government sectors assess suitable

(5)

areas for locating the SENHs in the short-term construction.

Thirdly, besides the evaluations for the current situation, the possible solutions to ease the difficulties in the location are also needed to be explored. We then propose an integrated framework which contains two models of measuring the spatial equity and evaluating the potential to identify potential facilities that can be transformed into elderly day care centers. It addresses the needs to ease the difficulties in implementing the location due to available sites shortage. The result may help the government sectors investigate the potential facilities that can be used to transform into elderly day care centers to ease the difficulty.

Finally, another way of redevelopment to deal with the difficulty of a shortage of available sites is proposed. We use an integrated approach which contains two frameworks of identifying potential sites using a four-step identification method and establishing the priority of identified sites using a multi-criteria evaluation, to evaluate potential sites for redevelopment. It addresses the needs to consider how to support the location by redevelopment which have the condition to implement in priority that can be used to locate the facilities. The result may help the government sectors explore the potential sites of redevelopment where the location of facilities can be implemented.

As proved by published and submitted research papers consisting in this dissertation, the GIS-based spatial analysis and open data with these two aspects are useful for supporting the location of welfare facilities for the elderly. The results are able to evaluate the suitable sites and explore possible solutions to support the location of welfare facilities for the elderly based on evaluated coupling degree, shortage areas, or prioritized situations in the present or future urban areas.

Keywords: location support, spatial distributive evaluation, explorations of possible solution, coupling degree model, parameter-improved floating catchment area method, spatial equity measurement, multi-criteria evaluation, Chome

(6)

Acknowledgements

The three year’s doctoral study and life in Urban Planning Laboratory, Kanazawa University is unforgettable for me. I have not only learned abundant scientific knowledge but also how to do the actual research.

The first acknowledgement must go to my supervisor Prof. Zhenjiang Shen. It is impossible for me to finish this dissertation without his guidance, suggestion, and patience during the process of preparation and writing. His teaching will benefit my whole life.

Then, I am very grateful to my examination committee members, Profs. Ito Satoru (Kanazawa University), Ikemoto Ryoko (Kanazawa University), Nishino Tatsuya (Kanazawa University), and Kawakami Mitsuhiko (Kanazawa University), for their careful reading and valuable comments on this dissertation. Without their useful help, I would not finish this research work smoothly.

Furthermore, thanks to all my teachers, classmates, and friends from Kanazawa University and other universities in China. I gained deep friendships and diverse knowledge from the daily activities and communications with them. Thanks to Prof.

Hulao Fang, I may not be able to join this big family without his help and recommendation.

I also acknowledge that this research was supported by the Ministry of Education, Culture, Sports, Science, and Technology in Japan with financial support in my last year.

That is meaningful for my study and life in Japan.

Finally, the special acknowledgement must go to my parents for their endless encouragement and support in my three years’ doctoral study and life. I am so proud of being their son, and I will continue to improve myself in the future life.

YOU Ninglong February, 2018

(7)

Chapter 1 Introduction

1.1 Research Background

Where to locate welfare facilities for the elderly has been a problem long troubled many government sectors due to the phenomenon of unbalanced relationship between supplies from facilities and demands of elderly population who cannot acquire services is obvious in urban development (Wenting Zhang et al., 2016). The strategic location of welfare facilities for the elderly is to provide services for demand elderly population.

Coordinating the demands of welfare facilities for the elderly through rational spatial distribution planning is urgent in development strategies in many cities, especially large cities. The government sectors need to locate the welfare facilities for the elderly in suitable areas, which is not only benefit for demand elderly population to use the facilities easily but also the government sectors to allocate limited budget optimally for realizing maximum service value.

Supporting the location of welfare facilities for the elderly is a critical method to deal with the unbalanced relationship between supplies from welfare facilities for the elderly and demands of elderly population who cannot acquire services. It has two layers of meaning: (1) identify the unbalanced areas in current situation, (2) explore solutions to alleviate this imbalance. However, how to support the location of welfare facilities for the elderly more effectively for improving the rationality of their spatial distribution with the developing of analytic methods and opening of urban data is still a challenge. GIS-based spatial analysis has been used by many government sectors increasingly to support the location of public facilities in the past decades, and related techniques in the spatial analysis were developed gradually. Meanwhile, with the opening of urban data constantly, the potential of optimizing the location of welfare facilities for the elderly from more valuable perspectives and the demands of innovative analysis methods are enhanced. Many researchers from different disciplines (including geography, urban and regional planning, transportation, etc.) studied for new methods, models, and algorithms in supporting the location of facilities, especially for welfare facilities for the elderly. Their achievements provide critical conceptual and technical supports for this Ph.D. research.

(11)

In addition, Japan is the country with the most rapidly aging effect (Tatsuya Nishino, 2015). It is much more critical for realizing maximum service value through the rational locations in the context of an obvious unbalanced relationship between supplies from facilities and demands of elderly population who cannot acquire services, especially in large cities like Tokyo. The government sectors should consider and pay more attention to the actual demands of elderly population, quality of available facility services, and possible optimization solutions, based on the current situation and characteristics of specific welfare facilities for the elderly. It is known that the enhancement of spatial distribution of welfare facilities for the elderly is a stepwise process with a limited government budget. Related studies should further try to improve the spatial analysis for supporting the location of welfare facilities for the elderly on the basis of understanding the existing short-comings in location assessments and implementations. Therefore, there have the needs and conditions to explore more effective spatial distributive evaluations and implementations to support the location of welfare facilities for the elderly.

1.2 Research Purpose

Welfare facilities for the elderly are facilities that provide various services for the elderly (65 years old or older), which is an important part of social welfare facilities in Japan. In this dissertation, their locations are identified and analyzed using GIS-based spatial analysis. The GIS-based spatial analysis refers to a set of methods which is devised to support the location of welfare facilities for the elderly from a perspective of spatial data analysis using GIS functions, which contains two aspects of evaluating current spatial distributions and exploring possible solutions for the difficulties in implementing the location. The available data used in these aspects are open data. Open data refers to a dataset which stores spatial data (such as the geographic locations of welfare facilities for the elderly, land use and building data, actual street network and so on) and non-spatial data (such as feature data of welfare facilities for the elderly, demographic data, social-economic statistics data and so on). These data are published in official websites of government sectors and searching websites which can be freely collected and used by government officers, urban planners, and researchers.

This Ph.D. research aims to use the GIS-based spatial analysis and open data to

(12)

support the location of welfare facilities for the elderly by the evaluations of spatial distribution for welfare facilities for the elderly and explorations of possible solution for difficulties in implementing the location. It contributes to help the government sectors identify the location more accurately and alleviate the difficulties in implementing the location of welfare facilities for the elderly.

Firstly, we make an overall analysis of the distributive relationship between the spatial distribution of welfare facilities for the elderly and population. It reveals unreasonable current situations of the spatial distribution of welfare facilities for the elderly in parts of the metropolitan area. In the second part, we focus on a specific welfare facility for the elderly (special elderly nursing homes) and assessing their locations by adopting improved methods and specific parameters, based on the short-comings in existing location assessments. In the third and fourth parts, we further explore two possible solutions (evaluations of potential facilities that can be transformed into elderly day care centers, and potential sites of redevelopment where the location of welfare facilities for the elderly can be implemented) to alleviate the actual difficulty of a shortage of available sites in the implementation of the location. It is expected to help the government sectors identify the location more accurately and alleviate the difficulties in the implementation of the location.

1.3 Literature Review

1.3.1 Researches for Distributive Evaluation in Supporting the Location of Welfare Facilities for the Elderly

According to existing researches, the issue of analyzing the location of welfare facilities for the elderly has received much attention. With the development of geographic information technology and mathematical method in recent years, most of studies adopted the spatial analysis methods based on the GIS functions (such as density analysis, neighborhood analysis, network analysis, interpolation analysis, weight analysis and so on) to explore the distributive evaluation in supporting the location of welfare facilities for the elderly. These studies can be generally classified into three categories according to their various methods and models which had been used in the spatial analysis.

An extensive literature on evaluating the relationship between the spatial

(13)

distribution of welfare facilities for the elderly and demand population exists, which have been proved as an effective way to provide basis for location of welfare facilities for the elderly. For example, Yasuhiro Nohara & Eiji Satoh (2017) evaluated the environment of services for the elderly using parameters of the service-providing area and the cover rate of elderly population at a sub-regional block level (500m*500m mesh); Rosilawati Zainol & Christopher J. Pettit (2016) examined the relationship between the distribution of the elderly and health care facility using local and global indicators at a block level; Michael P. Johnson et al. (2005) presented location models for senior centers based on the data of related cost and available budget at level of block;

Y Cheng et al. (2011) analyzed the spatial distribution of population of the elderly and facilities by developing various indicators of demographics of the elderly population and characteristics of residential care industry at a district level; Lee R Mobley et al.

(2006) examined the relationship between the elderly and primary care services by analyzing a set of data relating to the factors of demand, supply, and intervening at the level of census tract, etc.

Many researches focused on the accessibility in evaluating the location of welfare facilities for the elderly with an increasing attention on distance and convenience between the supply and demand. For example, Sekhar Somenahalli & Matthew Shipton (2013) examined the accessibility between the distribution of the elderly and services by quantifying twenty variables in the GIS at a suburb level; Shawky Mansour (2016) assessed the spatial pattern of service distribution at a subnational level, using parameters of the distances from demand populations to facilities and the ratios of facility to population in calculating the accessibility; Zhuolin Tao et al. (2014) measured the accessibility between the residential care facilities and demand locations at a sub-district level, and further optimized the measurement by adopting the weighted average accessibility score.

What’s more, some researches considered the location of welfare facilities for the elderly by analyzing the spatial equity from a perspective of social science. For example, Jian Sun et al. (2017) described the equity status of the distribution of high-technology medical equipment by calculating a concentration index at level of city; Shangyi Zhou et al. (2013) assessed the spatial equity of public-and-community facilities for the elderly using the weighted population data, types and densities of facilities at a sub-district

(14)

level; Luis Rosero-Bixby (2004) proposed comprehensive indices of the distance to the closest facility, size, proximity, and characteristics of demand population to evaluate the spatial equity in access to health care facility for the elderly at each 17,000 enumeration unit.

Different from the general population, spatial distributive evaluations for the location of welfare facilities for the elderly are more important for the demand elderly population due to their declining health status and movement (Shangyi Zhou et al.

(2013)). Meanwhile, the limited government budget of most cities dictates the construction of welfare facilities for the elderly can be finished but step by step. An effective identification for the location is more needed to improve senior citizen services gradually from the most needed areas. As showed by existing studies, the operational methods in spatial distributive evaluations for supporting the location of welfare facilities for the elderly have not been developed comprehensively, especially in the adjustments for specific facilities. Most studies assessed the location using various methods from a relatively general view. In other words, the welfare facilities for the elderly usually referred to a category of aged care or health care facilities which their demand population was represented by elderly population in broad terms. Furthermore, the basic units of spatial distributive evaluations that impact the location of welfare facilities for the elderly are mostly relatively large administrative divisions in the city, such as the (sub-)district, block, or census tract, etc.

However, there is a lack of research focusing on the impact of special factors on the location of specific welfare facilities for the elderly. It should be considered to meet the demands of detailed management for a variety of welfare facilities for the elderly in development strategies, based on the characteristics of facility and demographics of demand elderly population. Furthermore, the present analyses of the location of welfare facilities for the elderly mainly adopted relatively large administrative divisions as the basic units. Few studies focused on evaluating the location of welfare facilities for the elderly at a relatively micro level, which is unsuitable for the specific implementation of spatial planning and management.

1.3.2 Researches for Discussing Possible Solutions in Implementing the Location Using Open Data

(15)

As described in existing researches, various data sources had been adopted in exploring the possible solutions for implementing the location. Among them, solutions using open data have attracted much attention (Balena, P. et al., 2013). Open data is accessible, free, and public information (Liu et al., 2015), which is used to fill the gaps that the original datasets used in urban spatial planning and studies were solely provided by the government sectors and other authorities. Many new analytical methods based on open data were developed as well (Ge Zhang et al., 2015). One good application is the explorations of possible solution to support the location of health care facilities for the elderly. For example, Abdisalan M Noor et al. (2009) used public health facility database in proposed global positioning systems for assessing the geographic access to health care; H. Hazrin et al. (2013) mapped the spatial patterns of health clinics in GIS environment using the available data which were collected from various government sectors and handheld GPS; Koutelekos J. et al. applied location-allocation models to analyze the spatial distribution of health infrastructure using available data including population, road, land use, personal data of the people, etc. published by local authorities; Magnus Ryden (2011) mainly used open satellite data and available data from public websites to analyze the strategic locations of field hospital.

Meanwhile, many other studies applied open data in different fields of facility planning to discuss the possible solutions, such as traffic facilities and transportation (Greg Rybarczyk & Changshan Wu., 2010, Eric M. Delmelle et al., 2012, N. B.

Hounsell et al., 2016, Anastasia A. Lantseva & Sergey V. Ivanov., 2016, Chih-Hao Wang & Na Chen, 2015, Bilal Farhana & Alan T. Murray, 2008), children facilities (Ahmet Ozgur Dogru et al., 2017), emergency facilities (Rifaat Abdalla, 2016), energy facilities (Alexis Comber, 2015), and commercial facilities (Dyah Lestari Widaningrum, 2015), etc.

These methods and data have been used by many technology sectors and researchers which were known as a significant role in the spatial planning and management of urban facilities. However, on the overall situation, how open data can be integrated into spatial analysis effectively has not been discussed in related literature adequately (Ge Zhang et al., 2015), especially for the utilization of special open data in analyzing the location and exploring possible solutions for alleviate the difficulties in implementing the location of specific welfare facilities for the elderly. Therefore, this

(16)

dissertation will use diverse open data (including the formats of .shp, .csv, .xls, etc.) from different sources in the spatial analysis for supporting the location of welfare facilities for the elderly. Many analysis and statistical software (such as ArcGIS, Micro Excel, SPSS, etc.) will be used to analyze and calculate the data for supporting the implementation of welfare facilities for the elderly more effectively.

Based on above two aspects of existing researches, this research work uses the smallest geographical division in Japan (the Chohchohme, Chome, for short) as the basic unit for locating and supporting welfare facilities for the elderly. The Chome is the smallest administrative division in city which its average land area is 10-20 ha. Most administrative divisions (such as municipality, ward, etc.) and urban spheres (such as spheres of welfare, daily life, etc.) are consisted of the Chome. We not only evaluate current spatial distributions of overall and specific welfare facilities for the elderly, but also explore possible solutions to promote their implementations, based on their characteristics and demographics of demand elderly population by introducing special open data.

1.4 Organization

The main content of the dissertation is organized into four parts (see Fig. 1.1). The first component is the analysis of the distributive relationship between the welfare facilities for the elderly and population using the coupling degree model and kernel density analysis. Next, a method of parameter-improved floating catchment area is introduced to calculate the beds-needed index for evaluating allocation of spatial elderly nursing homes. The third component suggests integrating two models of spatial equity measurement and potential evaluation into one framework to investigate the possibility of transforming potential facilities into elderly day care centers. Lastly, methods of four-step identification and multi-criteria evaluation are used to explore the possibility of using potential sites of redevelopment to locate welfare facilities for the elderly.

Among them, part 1 and part 2 belong to the evaluations of spatial distribution for facilities. Part 3 and part 4 further to make explorations of possible solution to alleviate the difficulties in implementing the location. The data used in this dissertation are open data. All parts are organized to realize the purpose of helping the government sectors identify the location of welfare facilities for the elderly more effectively and alleviate

(17)

the difficulties in implementing the location in urban areas.

Fig. 1.1 The framework of this Ph.D. research content

1.4.1 Analyzing the Distributive Relationship between the Welfare Facilities for the Elderly and Population

Chapter 2 introduces the coupling degree model and kernel density analysis to analyze the coupling relationship between the spatial distributions of welfare facilities for the elderly and population, using open data of information points of public facilities and polygons of the Chome. This study reveals characteristics of the spatial distributions of welfare facilities for the elderly and population, and their coupling relationship at a relatively small basic unit, the Chome. The results show there exist areas where the distributive relationship between welfare facilities for the elderly and population are relatively unbalanced, and it is intimately connected with coupling degree. It may help the government sectors analyze the reasonable degree of the relationship between the distributions of welfare facilities for the elderly and population.

For this chapter, I have published one paper on the No. 39 Information, System, Utilization, Technology Symposium (Architectural Institute of Japan (AIJ)) (J-stage), and the title is “Spatial Structure of Public Service Facilities and Its Coupling Relationship with Population Distribution based on GIS: Case Study of Central Tokyo,

(18)

Japan”.

1.4.2 Assessing the Allocation of Special Elderly Nursing Homes

After the analysis of welfare facilities for the elderly, Chapter 3 focuses on special elderly nursing homes (SENHs) and assesses its allocation. Departments of social welfare and public health need dependable evaluations for improving reasonability of periodical construction of the SENHs. The chapter has the purpose to evaluate allocation of the SENHs that may help departments assess suitable areas for locating SENHs. Present evaluations use spatial units of sphere of daily life, the ward, and sphere of welfare to estimate the ratios of beds to elderly population in Tokyo.

Therefore, we compute beds-needed index (BNIS, for short) via adopting a parameter-improved floating catchment area method (PI-FCA, for short) in a relatively micro spatial unit (called Chome). In method of PI-FCA, service areas (catchment areas) are formed on the basis of criterion of the average residents who serviced by the targeted facility and their capacity, demand elderly population is aged population needing health care from care level 3 to level 5, and they are revised by the coefficient of potential demand of elderly population by organizing the multivariate linear model in the next step. The improved results were gained by the method of PI-FCA. Lastly, the chapter maps spatial placements of BNIS’s degree, to give a basis to allocate the SENHs. It meets demands of the government sectors, and can be lightly applied by the other kinds of health care facilities, especially for the elderly. For chapter 3, I have published one manuscript on International Journal of Environmental Research and Public Health (SSCI/SCIE), and title is “Assessing the allocation of special elderly nursing homes in Tokyo, Japan”.

1.4.3 Investigating the Possibility of Transforming Potential Facilities into Elderly Day Care Centers

This chapter proposes an integrated framework that contributes to investigate the potential facilities that can be used to transform into elderly day care centers (DC centers) in supporting the location of welfare facilities for the elderly. To address the needs to ease the difficulty in implementing the location due to available sites shortage, based on the usage of other potential facilities, efficiently. This framework contains two

(19)

different models: measuring the spatial equity of DC centers and evaluating the potential of potential facilities. The models consider the facility’s scale index, density index in service area, index of demand population, and travel mode-based distance index in measuring the spatial equity of DC centers, and the facility’s capacity index, index of demand population in evaluating the potential. Results are analyzed and visualized in the geographic information system (GIS), and show diverse indices of spatial equity of DC centers and priorities of potential facility which are accompanied by dynamic variations in demand elderly population. The proposed framework could be a useful reference to help the government sectors investigate the potential facilities that can be used to transform into DC centers to ease the available sites shortage.

1.4.4 Exploring the Possibility of Using Potential Sites of Redevelopment to Locate Welfare Facilities for the Elderly

Redevelopment is a major way of urban renewal that responses to the changes of socio-economic indicators such as the aging. It has received increasing attention in large cities like Tokyo. The promotion of urban redevelopment contributes to promote the construction of welfare facilities for the elderly due to its economic and social benefits.

It is a benefit for supporting the location of welfare facilities for the elderly in urban areas to give priority to improve elderly services in the selected sites of redevelopment.

However, there still has challenges to improve the efficiency of promoting redevelopment based on the private-public collaboration, where information is needed that can be referenced for site selection of redevelopment at the Tokyo metropolitan level. Therefore, an integrated approach is devised based on two frameworks. The first identifies potential sites for redevelopment by adopting a four-step identification method: define input sites; examine demographic conditions; check physical environmental conditions; and verify growth potential. The other establishes the priority of identified potential sites with a set of criteria, which are selected based on the demands of private and public sectors, and weighted by learning from existing single-project assessment systems, reviews of literature, and discussions with experts.

This approach responds to the current needs and can be easily applied in other cities.

In the following chapters, methods of coupling degree model, parameter-improved floating catchment area method, spatial equity measurement, potential evaluation, and

(20)

multi-criteria evaluation, etc. will be set as the specific spatial analysis methods in this research work, and related open dataset (including spatial data, such as the geographic locations of facilities for the elderly, land use and building data, actual street network, etc., and non-spatial data, such as feature data of facilities for the elderly, demographic data, social-economic statistics data, etc.) will mainly be used in the case study of Tokyo metropolitan area in Japan. It is important for understanding how to support the location of welfare facilities for the elderly using these methods.

(21)

Chapter 2 Analyzing the Distributive Relationship between the Welfare Facilities for the Elderly and

Population

2.1 Introduction

Public facilities are generally used to meet people’s daily requirements with government investments. They belong to social service facilities and mainly include medical facilities, educational facilities, welfare facilities, cultural facilities, etc., which are significant to improve residents’ quality of daily life. Especially for welfare facilities for the elderly, the problem of balancing the supplies from facility and demands of population is becoming urgent in the context of increasing aging population. The quantity and quality of facility services that can be served to the public are influenced by the spatial distributions of facility and population. Therefore, this chapter aims at building a coupling model between spatial distributions of public facilities and population to measure their matching degrees based on the GIS, and gives special concerns to welfare facilities for the elderly. The study area is the Tokyo metropolitan area.

Evaluation for spatial distribution of public facilities is an important issue in urban development strategies in many cities. Take the Beijing city, China as an example.

With the new round revision of Beijing’s urban master plan has been spread out in 2014, the evaluation of implementation status and exploration of promotional strategy for public facilities have been a core content of constructing livable city. In view of current development situation, differences between spatial distribution of the housing and its surrounding public facilities are increasingly prominent. Difficulties in the usage of public facilities with Beijing’s characteristics have become the focus problems of people’s livelihood. For example, many public facilities are existing problems of uneven distribution and insufficient supplies, especially for medical and basic welfare facilities for the elderly at a community level (Jinfeng REN et al., 2012). They have own peculiarities in spatial distributions, variation trends and associations with population.

In this aspect, Tokyo is more mature in the planning and management of public facilities.

In 2013, the number of hospital per 100,000 people in Tokyo is 1.6 times of Beijing, and

(22)

the numbers of primary and middle school per 100,000 people of right age are more than Beijing, etc. (Wenzhong ZHANG, 2016). However, there still exist the mismatches between public facilities and population in Tokyo to a certain extent. The analysis of distributive relationships between public facilities (especially for the welfare facilities for the elderly) and population is a benefit for the rational locations and spatial planning of public facilities.

The purpose of public facilities is to provide services and pursue to maximize public benefits. The maximization of public benefits is mainly reflected in the minimum consumption of residents to a facility which is related to the service population and average distance to a facility (Changkui SHI, 2015). The measure of coupling fuses these two core elements. It expresses the internal consistency of two geological matters in one space and the degree of interaction (Chenghao YANG, 2013). At present stage, coupling analysis has been used in urban planning fields. Miaoxi ZHAO et al. (2014) discussed the coupling relationship between information technology level and public transportation; Feng HAN et al. (2009) focused on coupling relationship between the traffic pattern and city structure, and analyzed associated factors; Søren Kjær Foged (2016) analyzed the internal relation of outsourcing the public service in the city and the size of population.

Based on relevant literatures, the analysis of coupling relationship has been applied in parts of urban spatial planning and management. But studies for spatial distribution of facilities from perspectives of population and distance were relatively less. Even existing in studies, scholars were mainly focusing on the relatively macro level of city.

It lacks views at relatively medium or micro levels, which caused inaccurate evaluations for location guidance. Therefore, this chapter takes central Tokyo as a case study and evaluates spatial distribution of public facilities. Based on this, we build a coupling model using the Chome as the basic unit of population distribution for discussing the coupling relationship between spatial distribution of public facilities and population.

This chapter will provide a certain scientific basis for measuring reasonable degree of public facilities (especially for welfare facilities for the elderly) in Tokyo.

2.2 Study Area and Data

This chapter selected central Tokyo (units of 23 wards) as the case study. 23 wards

(23)

are the central of politics, economy and culture of Tokyo, which cover an area of 621.81 km². The population of 23 wards is over 9 million people and their distributions have changed with the reform of urban functions and structure. The spatial distributions of public facilities and population will be firstly described. Secondly their coupling relationships will be analyzed by building the coupling degree model to evaluate the implementation effect.

The data has been used in this chapter mainly comes from the data and materials that official issued by the government sectors and departments: (1) The information points (shape file) of medical facilities (hospital and clinic), primary schools, welfare facilities for the elderly, cultural facilities (library, gallery, and gym) and urban parks in the 23 wards from the national land data published by the Ministry of Land (Fig. 2.1).

There are 27526 points in total and each point contains the information of name, location and category. We can acquire the number of each public facility in each Chome through the data collection. (2) The land use map (shape file) of 23 wards from national land data was published by the Ministry of Land. It will be used to make comparison between population distribution and its land use type, which has significant impacts on evaluation of coupling relationship. (3) Finally we use the administrative divisions of ward and the Chome respectively (shape file) from portal site of official statistics of Japan. We convert the polygon of Chome into centroid in order to make visualization analysis. Useful information in these data are mainly including the land area and population of the Chome, which are required parameters in calculating the coupling degree.

(24)

Fig. 2.1 Spatial layout of points of public facility

2.3 Methodology

2.3.1 Kernel Density Analysis

Kernel Density Analysis is a significant method of spatial elements analysis in the ArcGIS. It mainly used to calculate the concentration of specific elements within their scope of neighborhood. This method takes the location of a specific point as the center and makes distribution of its properties in the range of specified thresholds. It means in a circle with the radius r, the density is biggest in the center and gradually weakened outwards (G. Modica et al., 2012). Finally the density map of points will be formed by the same way of superposition and smoothing process in a certain area.

2.3.2 Coupling Degree Model

Coupling relationship expresses the internal consistency of two geological matters in one space. Hence, we need to estimate the degree of interaction and influence from each element through the calculation of coupling degree. In the process of modeling between public facilities and population, two parameters of service population and the distance to a facility will be mainly used (Huitong WANG et al., 2014). Based on this, we build the basic model for calculating the coupling degree as follows (Eq. 2.1):

(25)

C_c = P_c× D_c (2.1)

C_c is the coupling degree. P_c is the population parameter. D_c is the distance parameter.

This chapter selects the Chome in 23 wards as the basic unit of population distribution, and their surrounding public facilities will have great impacts on residents’

quality of daily life directly in the Chome. Each ward is divided into several Chome which means the service capabilities of public facilities in each Chome are determined by the proportion of population in each Chome to the total population of its ward.

However each kind of facility has many species which is hard to determine serviced areas. Therefore, we adopt P_m which denotes the population of Chome with highest service capability in whole city and treated as the benchmark 1 (P_m≥ P_ch) (Joseph AE et al., 1982). We can get population parameter by Eq. 2.2:

P_c = ^P^ch

P_m (2.2) P_ch is the population of each Chome. P_m is the population of Chome with the highest service capability.

Due to the location of public facilities are relatively random in the ward and Chome, and the housing distribution in Chome is difficult to measure. Hence, we adopt the variable of average distance to represent the distance from housing to facility.

Meanwhile, research objects in this chapter refer to a large class of public facility in general which focus on an overall relationship with population. Therefore, in the calculation of distance parameter at the level of Chome, we assume that facilities are distributed randomly. Based on this, we use the average distance under the condition of random distribution as one of variables in distance parameter D_c (Marcon E et al., 2003; Canfei HE et al., 2007). The distances are related to the land area of Chome and number of each facility (Eq. 2.3).

D_average= √^A_N^c

f⁄2 (2.3)

(26)

D_average is the average distance to each facility. A_c is the land area of each Chome. N_f is the number of each facility. In practices, each facility has its own appropriate service radius according to relevant specifications and plans, and it will cause inconvenience to residents when the distance exceeds the suitable radius.

Therefore, we involve the ratio of service radius into above average distance from the house to facility to represent the distance parameter (Eq. 2.4):

D_c = 2R_m √^A_N^c

f

⁄ (2.4) R_m is the suitable service radius of each public facility. A_c is the land area of Chome. N_f is the number of each facility in the Chome. The longer distance, the lower is the distance parameter. Above all, we build the coupling degree model between public facilities and population is following by putting Eq. 2.2 and Eq. 2.4 into Eq. 2.5:

C_c =^2R^m^P^ch

Pm√^Ac Nf

(2.5)

C_c is the coupling degree of each Chome. R_m is the suitable service radius of each facility. P_ch is the population of Chome. P_m is the population of Chome with highest service capability. A_c is the land area of each Chome. N_f is the number of each facility in the Chome.

2.4 Results

2.4.1 Spatial Distribution Analysis of Public Facilities

After making a classification of information points of public facilities, we analyze the kernel density using ArcGIS10.0, respectively (Fig. 2.2). Medical facilities are existing agglomeration phenomenon in certain degree; wards with higher density of distribution are mainly including the Chuo-ku, Chiyoda-ku, Shibuya-ku, and Shinjuku-ku, which took 800 meters as the radius of searching density. They are radiating out from center to periphery. From the land look, the density of medical facilities is higher in commercial area, especially for the proportion of daily medical services like dentistry and internal medicine are relatively high. Primary schools have

(27)

been evenly spread in various land use areas and its spatial distirbution presents an obvious decentralized structure with multiple cores. These cores are distributed in the regions outside the core areas. Denser regions of primary schools are mainly located in high-rise residential areas, quasi industrial areas, and a few commercial areas. The degree of balanced distribution of cultural facilities is highest and spatial structure is scattered dot form distribution. Its clustering is not obvious and distribution is consistent in various land use areas. The spatial structure of welfare facilities for the elderly is similar with primary school which presents a decentralized structure with multiple cores.

Denser regions of welfare facilities for the elderly are mainly located in high-rise residential and quasi industrial areas. Urban parks are easier clustered in outside regions, such as the Itabashi-ku and Ota-ku. Core areas like Bunkyo are reflecting a small amount of clustering. Parks are mainly located in low-rise residential and industrial areas which basically achieve high space coverage for urban areas.

(28)

Fig. 2.2 Results of kernel density analysis for public facilities

2.4.2 Population Distribution

At the macro level, the population is centered in the Chiyodai-ku, Chuo-ku and Taito-ku and the number increases gradually from center to outside wards. A larger proportion of the population is concentrated in the Edogawa-ku, Adachi-ku, Nerima-ku, Setagaya-ku and Ota-ku, which is closely related to the current life mode of separation of workplace and residence in Tokyo. Making a comparison with land use, populations are mainly clustered in residential areas, and different categories of residential area reflect various characteristics of density. Wards of the Adachi-ku, Katsushika-ku and Edogawa-ku are strongly correlated with industrial area. Most of them are high-rise residential area and a large number of populations have gathered in order to meet the living demands of employment. In the western regions such as the Nerima-ku, Suginami-ku, Setagaya-ku and Ota-ku, etc., where gathered a large number of populations in the land use type of low-rise residential area. However in central regions like the Toshima-ku, Shinjuku-ku and Shibuya-ku, the lands are mainly used in high-rise housing construction, but population density is lower due to reasons of near the core commercial areas, high housing prices, etc. At the medium level, the feature of population distribution in the Chome is basically similar with the macro level. The Chome with higher degree of population agglomeration are basically distributed evenly

(29)

in their wards, but they generally appear a trend of increasing gradually from internal to external in urban areas (Fig. 2.3).

Fig. 2.3 Results of population distribution analysis in the ward area of Tokyo

2.4.3 Coupling Relationship between Facilities and Population

With the urban development and space extension of Tokyo, the distributions of public facilities and population basically gathered in regions outside the urban core areas and they are gradually extended. The harmonious development of these two ones is significant for improving residents’ quality of daily life and optimizing spatial structure of urban. Facilities tend to cluster moderate in areas with high population density, which is benefit for more residents’ daily needs. They should adapt to the features of population distribution and present moderate scattered layout to achieve efficient coverage for residents’ life. From this current status, public facilities were in agreement with population as a whole. But whether their relationships matched at the level of Chome, and where the mismatches occurred. Further research is still required to evaluate these questions in the Chome.

In practice, various facilities have different suitable service radius because of their own properties. This chapter selects suitable service radius of each facility according to planning indices from several actual cases. In addition, we select medical facilities, primary schools and welfare facilities for the elderly as case studies to make deeper analysis on coupling degree with population due to the depth of relations to daily life

(30)

and available data. Firstly, we put organized data into Eq. 2.5 to calculate the coupling degree. Then, we pick those Chome which coupling degrees are less than 0.5 as areas that the coordination relationship is low. Using the software of SPSS22.0 and ArcGIS10.0 to calculate overall and average degree of these facilities and show their spatial distributions. The schematic diagrams of coupling are showed in Fig. 2.4.

Overall, the coupling degrees between public facilities and population are relatively high. The average coupling degree of medical facilities, primary schools and welfare facilities for the elderly are 2.41, 1.51 and 1.90 respectively. Medical facilities’ coupling degree is highest and primary school is relatively low. Comparing to medical facilities, the spatial planning of welfare facilities for the elderly still has spaces to improve. From perspective of spatial structure, the Chome with relative low coupling degrees of welfare facilities for the elderly are mainly located in city center and sub-center. But the land uses of these districts are commercial purpose that their welfare facilities for the elderly are mainly providing services for peripheral elderly population. Hence, these Chome are not treated as the main areas of coupling evaluation. Focusing on others non-commercial use areas, the Chome with relative low coupling degree of medical facilities are clustered in parts of residential areas in the Setagaya-ku, residential and industrial areas in the Adachi-ku and Katsushika-ku, and industrial areas in the Shinagawa-ku and Ota-ku. The Chome with relatively low coupling degree of primary schools are mainly clustered in parts of residential and industrial areas in the Adachi-ku, and industrial areas in the Edogawa-ku and Koto-ku. The Chome with relatively low coupling degree of welfare facilities for the elderly are mainly clustered in parts of residential areas in the Itabashi-ku, residential and industrial areas in the Adachi-ku, and industrial areas in the Edogawa-ku. These regions are existing mismatches to varying degrees and the couplings of different pubic facilities are distinguishable (Fig. 2.5). It will disturb the comfort and accessibility of residents’ daily life, especially for the elderly population.

(31)

Fig. 2.4 Coupling relationship between public facilities and population

Fig. 2.5 The number of Chome in each ward which the coupling degree is lower than 0.5

2.5 Conclusions and Discussions

Spatial distribution analysis of public facilities and its coupling degree with population is one of methods for evaluating implementation and developing optimization strategies of public facilities, especially for welfare facilities for the elderly.

Results show, (1) Spatial distribution of welfare facilities for the elderly in Tokyo are relatively balanced basically and clustering in local areas to some degree; (2) There are differences of clustering areas among various facilities, which welfare facilities for the elderly are mainly concentrated in residential and industrial areas; (3) The distribution

(32)

of population in Tokyo reflects a trend of increasing gradually from internal to external in urban areas, but it is more balanced at the level of Chome. Meanwhile, except for residential areas, industrial areas present a certain size of population agglomeration; (4) At the Chome, the coupling relationship between welfare facilities for the elderly and population still leaves spaces to improve, mismatching phenomenon mainly appear in residential and industrial areas of the outskirts of urban, and there are close correlation with density of welfare facilities for the elderly.

This model also can be applied in other cities to evaluate the allocation of public facilities based on population distribution, and find areas do not match. Case study of Tokyo provides useful inspirations to manage public facilities. A relatively balanced allocation of facilities, proper numbers and be constructed in time will have positive effects on the improvement of coupling degree, especially for the construction of welfare facilities for the elderly. Compared to the basic unit of Chome, spatial distribution of public facilities in other cities can be considered to optimize from a relatively micro-perspective, which has closer relations to residents’ daily needs.

(33)

Chapter 3 Assessing the Allocation of Special Elderly Nursing Homes

3.1 Introduction

The phenomenon of aging effect in Japan (especially in large cities) is very serious in the world (Ihara, K., 1997). In Tokyo, the existing beds in the special elderly nursing homes, shortly for SENHs, are obviously fewer than demand elderly population (Japanese Nursing Association, 2017). It has become a very serious trouble which caused the social welfare and public health departments (departments) in the ward area of Tokyo formulated programs of senior citizen health and welfare, shortly for programs, to promote establishment of the SENHs in areas where the services are not enough, such as sphere of daily life, the ward, and sphere of welfare. Those programs were firstly formulated in 2006 and revised per three years. For determining the priority to locate the SENHs more accurately in the recent construction, these departments require more dependable evaluations to assess location of the SENHs. But, the present evaluations in programs have no adequate guiding function for providing useful information for the departments. Thus, this chapter focuses on improving the present assessment for allocating the SENHs, by proposing a new index of beds-needed index to the SENHs (called BNIS), and mapping spatial placements of BNIS’s degree for evaluating allocation in the ward area of Tokyo using the open data at level of Chome.

The SENHs provides nursing (or health) care for aged people who need care from level 3 to level 5, called ER3–5, especially for bedridden. Aged people who need nursing care in Japan are divided into 7 different levels, including support 1-2, and care 1, 2, 3, 4, 5, according to their health status. A higher level means they have a higher requirement for nursing. Normally, the elderly people who in levels from care 3 to care 5 (the corresponding government sectors make the examination), has the right for using the SENHs. They are usually treated as “last habitat”, to guarantee good health in aged people’s later years. This group of elderly people (ER3–5) tends to select the closer SENHs which have conversant environments. The phenomenon is mainly because their poor physical and rough living situations. In Tokyo, allocating the SENHs at levels of the sphere of daily life, the ward, and sphere of welfare, formulated by the departments, is mainly based on two methods of calculating the ratio of beds to elderly population

(34)

(called BPR method), and demand elderly population (population of ER3–5). BPR method is generally used at sphere of welfare (consist of many successive wards) or one ward. The demand elderly population is analyzed in sphere of daily life which consists of several Chome. These methods formed a procedure from top to down, by using these sphere of daily life, ward, and sphere of welfare, but lacks of evaluating the BPR at a specific level of the Chome (Chome is the lowest administrative division in Tokyo).

That is one of main origins of this research, which even the smallest spatial unit (sphere of daily life) is yet too dissimilar or large for evaluating an allocation in the implementation at present stage (this spatial unit was generally based on the spatial distribution of local residents with relatively homogeneous demographic or socioeconomic features). Therefore, in this chapter, the Chome is used as the basic unit for estimating proposed BNIS. Then, it maps spatial placement of BNIS’s degree.

The BNIS means quantity of shortage available beds, which is the demand elderly population, requires using the SENHs, but cannot acquire corresponding health care.

Spatial placements of BNIS’s degree, the distribution of ER3–5 that has the demands for using the SENHs, will be formed by the Chome with similar BNIS. The spatial mode is not just limited by Chome (which means its administrative boundary) due to the spatial scale is generally smaller than service areas (or called catchment areas). This has an implication that, its location is not just restricted within scope of Chome which its BNIS is relatively high, but also the neighboring Chome within catchment areas.

Assessing the location of SENHs in the Chome is treated as a supplement. That means the SENHs can be better located where the BNIS’s degree are relatively the highest.

Moreover, BNIS’s degree is represented by the population of ER3–5, who cannot acquire services from the SENHs. It has two groups. First is the demand elderly population that more than SENHs’ capacity in scope of catchment areas. Second is the demand elderly population that not in the catchment areas. A high BNIS means that, it is much more significant for the government sectors or the other social welfare organizations to build the SENHs. The data about the ER3–5 and their characteristics of demographic and socioeconomic at the Chome provide the help and possibility to calculate the proposed BNIS within the relatively micro urban units in urban area.

For original BPR method, aged population parameter is adopted as demand elderly population in denominator. Parameter of population of ER3–5 is able to replace general

(35)

number of the elderly for improving the accuracy of BPR method (Mitchell, O.S. et al., 2006). However, two implicit shortcomings are still present. Firstly, it makes assumption that, demand elderly population in ward or sphere of welfare has same accessibility to arrive the SENHs, in other words, spatial distance connecting the ER3–5 and SENHs (convenience to acquire health care) is not taken into account in the round.

Then, the potential demand of elderly population for using the SENHs is not considered adequately; due to the actual demands of elderly population sub-groups very differ according to demographic and socioeconomic features. Though the departments have adopted partial features of ER3–5 (such as dwelling situation, family structure, etc.) to assess SENHs’ allocation by ways of onsite visits or questionnaire survey. But, it can gain only part of data from the demand elderly population, because of limitations in the general field surveys. At the same time, it is still hard for the related government sectors to combine and use those data in calculating BPR through a simple way using the BPR method. It will benefit for developing the methods which can add the deliberation of spatial distance connecting demand elderly population and SENHs, and further improve demand elderly population (ER3–5) via considering the parameter of potential demand of elderly population.

Several literature suggested using and improving the two-step floating catchment area method, to consider more about the spatial distance connecting demand and supply in calculating the more rational proportion of serviced population and facility’s capacity (Luo, W., 2004, Gu, W. et al., 2010, Wan, N. et al., 2012, Yang, D.H. et al., 2006). The presuppose behind this “two-step” of the method is the basic spatial unit to estimate the ratio of beds to elderly population is much bigger than service areas (or catchment areas) of targeted facilities. For the chapter, original method of two-step floating catchment area is introduced for calculating the ratio of beds to elderly population in two kinds of spatial units, which are both smaller and bigger than catchment areas (the original method of two-step floating catchment area is abbreviated “FCA”). FCA focuses on the present shortcomings of the BPR method. It is used as a supplement to calculate ratio of beds to elderly population at ward, sphere of welfare, and even the Chome. The mentioned catchment areas are service areas with defined spatial distances. In order to decide a facility’s catchment area, several researches adopted the coverage areas through determining diverse travel times and distances according to the data of actual

(36)

street network (Cheng, G. et al., 2016, Luo, W. & Qi, Y., 2009, Dai, D., 2010, Tao, Z. et al., 2014); and some researches use various distances to form different service areas, based on the features of different specific facilities, to determine the catchment area (Gupta, K. et al., 2016, Cutumisu, N. & Spence, J.C., 2012, Higgs, G. et al., 2015).

Most researches considered the catchment area on the basis of the capacities and types of facilities, and also including the density of serviced population. In the chapter, catchment areas are formed based on different spatial distances in the second step based on criterion of the average demand elderly population that serviced by corresponding SENHs and SENHs’ capacity as well. This capacity is represented by the existing available beds of each SENHs.

The demand elderly population (ER3–5) in method of FCA, its calculation is not only affected by the health status of the ER3-5, but also their demographic and socioeconomic features (Cheng, Y. et al., 2011, Ding, Q.X. et al., 2016). It is a critical consideration to improve the FCA in constructing the welfare facilities for the elderly.

The core question is that, how to transform the data of demographic and socioeconomic features into coefficients, to enhance the consideration of demand elderly population (La Rosa, D., 2014, Higgs, G. et al., 2015). It is benefit for meeting the elderly’s needs.

At the same time, there have internal relations among different variables in the real-world data of demographic and socioeconomic features of demand elderly population (Simizutani, S. & Inakura, N., 2007). Combining these variables simply, ignoring correlations in computing coefficients is not accurate. Therefore, this chapter proposes a new multivariate linear model for calculating influence of each variable on demand elderly population that needs to use the SENHs. Its result can called as the coefficient of potential demand of elderly population. The model is the method for enhancing original parameter of demand elderly population (ER3–5) in the FCA.

This proposed FCA was introduced firstly to improve original assessments like gravity model. This model takes urban spatial distance connecting demand and supply into account by the distance decay (Field, K., 2000, Shen, Q., 1998). With the development of analytical methods, most current studies address to enhance original method of FCA via defining facilities’ catchment area, due to this method ignores the differences of accessibility (Cheng, G. et al., 2016, Cutumisu, N. & Spence, J.C., 2012, Dai, D., 2010, Tao, Z. et al., 2014, Luo, W. & Qi, Y., 2009, Higgs, G. et al., 2015, Gupta,