The Location Evaluation and Relocation of Community & in-Home Care Centers: A Case Study of Fuzhou, China

ABSTRACT This study is a case study on evaluating locations of 46 community and in-home care centers (CHCC) in four central districts in Fuzhou, China. The goals are to evaluate the location suitability of the current centers, eliminate the unsuitable locations and recommend suitable alternatives. In particular, this study tried to explore the feasible approaches for urban planners to enhance the process of planning and implementation participated by multiple stakeholders, with less decisions made by the care providers and the government departments. The suitability analysis of the current CHCC locations was implemented in two steps. First, the inappropriately sited centers were identified by the multiple indexes of legality, vulnerability and convenience through a Fuzzy Analytic Hierarchic Process analysis and were emulated and visualized by GIS. Second, the ill-located centers recommended more suitable plots identified by p-center or coverage location models according to the type of these centers. As a result, five centers are perfectly located, while 19 centers need to be relocated, as they were located on sites that were either inconsistent with the appropriate land use or in the risks of natural hazards. The rest half (22 centers) need improvements to different degrees to meet all the relevant planning requirements.


Introduction
China is currently ageing more rapidly than the rest of the world. According to the latest 7th national census (standard time 0:00, November 1, 2020), the population aged 60 and older in China already make up 18.70% of the population, whereas that number make up 12.3% of the global population in 2018 (WHO 2019). In order to equip individuals, families and societies to address social and economic challenges of rapid ageing population, the central government established a primary"ageing-in-place" care system consisting of community care, in-home care and institutional care for seniors since 2009 (Lin 2011) and urged municipal governments to launch community and in-home care centers (CHCC) on a tight schedule. As community-based care centers, CHCC aims to sustain the independence of senior residents with physical limitations or limited functioning due to memory loss. Providing comprehensive full-day care, including adult day care and in-home care services, CHCC inevitably plays a key role in the primary care system. CHCCs in China are usually TOT (transfer-operationtransfer)-type PPP (Public-Private-partnership) projects, in which the land/space for CHCC are generally provided by the local government. The past few years have thus witnessed huge growth in the number of CHCC in many China cities. However, these centers were built by care providers introduced to each community with the funding from the municipal Civil Administration Department. Due to the absence of planners in the location finding stage and the instructive and descriptive nature of the current CHCC-related regulations and guidelines, many centers are roughly located on inappropriate sites. In some extreme cases, CHCCs are found in buildings without necessary permits and even risking their users' lives. This study thus aims to explore the feasible approaches for urban planners to assist decisions on siting CHCCs, or other public facilities, participated by multiple stakeholders, such as communities, the care providers and the relevant government departments without making all the decisions for them.

Location modelling in location science
Based on the 19th century-founded location theory, the location science includes both analysis and modelling. The CHCC location modelling belongs to the equity-oriented model among the four general classes of location models: allocation, equity, stochastic and competitive. As Murray (2010) pointed out, the equityoriented model has two general approaches: center and covering. The center models are developed to solve the p-center (PC) problem, with the intent of locating p facilities in order to minimize the maximum distance a demand point was from its closest facility (Hakimi 1964;Handler 1990;Mladenovic, Labbe, and Hansen 2003), whereas covering models are built to solve the maximal coverage location (MCL) problem, where p facilities are to be sited in order to maximize demand served within the stipulated standard (Church and ReVelle 1974;Schilling, Jayaraman, and Barkhi 1993). Nevertheless, in reality, site selection decision is a complex problem with many factors that go into it. International organizations such as WHO (2007), ADI (1999) and various national and regional government departments have already identified many demographic and socioeconomic factors specific to care center locations in urban areas, such as land use, community demographics, natural and environmental hazards, transportation network and traffic. Despite the fact that planners and government officials have long tried to increase the effectiveness and accessibility of the care system, most of their efforts have failed to even bring about a precise methodology or definition of the most efficient structure for delivering care facilities or determine and understand the relative influence or weight of the major factors affecting their locations. There are probably two major reasons. First, the classic location model is a single objective model, which can only achieve one objective at a time. With the location of the facilities and demand allocation being mutually restrictive, the determination of location is often the result of multiple objectives, but there is no one location model that addresses all these issues. Second, some location-related factors are incontrollable. As Reinke (1988) pointed out, "Health system research is complicated in another way; because it is carried out in the community, it includes multiple variables not easily controlled."

GIS in location evaluation and planning
Though not ideal nor "optimal", GIS has been used for the visual interactive siting of facilities and subsequent evaluation. Generally, three strategies are used to deal with these problems. First, redefining the problem more precisely. For instance, to choose the best fit location model for CHCC, Li et al. (2017) divided CHCC into three types: rehabilitation care centers, medical care center and day care center, and found that covering model fits the rehabilitation and the medical centers the best, whereas center model fits the day care centers the best (Li, et al. 2017(Li, et al. , 2019. Second, using decision model to weight factors differently, normally by Analytic Hierarchic Process (AHP) and Fuzzy AHP (FAHP) analysis. For instance, to make the decision where to locate a new healthcare project for the metropolitan city of Turin in Italy, Caprioli and Bottero (2021) compared AHP and FAHP methods and found that AHP should be preferred in the case of the high relevance of all the criteria considered, whereas FAHP introduces more uncertainty in criteria weights assignment and reduces the burden of many spatial analyses with GIS. Wilson et al. (2019) use an inductivehierarchical vulnerability index construction model to identify nursing home vulnerability attributable to location using a triangulated approach that includes historic natural hazards, community vulnerability and nursing home attributes. Third, the very recent heuristic modelling, by defining a bi-level (Dan and Marcotte 2019) or multi-level problem (Ortiz-Astorquiza, Contreras, and Laporte 2019), in which the first level considers the facilities' location and the other ones consider the optimization of other aspects.
Hence, the aim of this study is to explore, from different perspectives, the problems of CHCC location by describing the different planning efforts and methods used for achieving better CHCC location and delineating the CHCC service area. This study uses the Fuzhou metropolitan area as a case study to examine the relationship between existing CHCCs' locations and the factors that are identified by the current planning regulations. GIS suitability analysis was used to find out whether existing patterns of location are representative of what prior research suggests. This understanding is important for planners, care providers, civil administrators and all those who want to enhance accessibility of aged population segments to provide valuable support for CHCC location planning and decision-making and key to ensuring operational efficiency and equity in the provision of services.

Methodology
The suitability analysis of the current CHCC locations was implemented in two steps. First, the inappropriately sited centers were identified by the multiple indexes of legality, vulnerability and convenience through a Fuzzy Analytic Hierarchic Process (FAHP) analysis and were emulated and visualized by GIS. Second, the ill-located centers recommended more suitable plots identified by p-center or coverage location models according to the type of these centers.

Conceptual framework
To accurately conceptualize location suitability according to land use, natural hazard frequency and the accessibility to facilities, the conceptual framework outlined by Wilson et al. (2019) is adapted in this study ( Figure 1). To compose the composite index, three types of index were considered: legislation index (LI) of land use, hazard index (HI) of natural or environmental hazards and convenience index (CI) measuring the accessibility to relevant resources.
Findings from various research indicate that communities and their residents are vulnerable to a variety of natural hazards and environmental health risks (UN 1998). The older adult residents are considered medically vulnerable and therefore more susceptible to the impacts of hazards than their younger peers (Hames et al. 2016). As such, the current planning regulation recognize the vulnerability of CHCC by emphasizing that the land used by CHCC should be geologically stable, not in danger of floods and away from pollution and/or noise sources, the manufacture and/or storage of dangerous goods and municipal roads with heavy traffic. The suggested features include adequate sunlight and good ventilation and easy access to facilities or venues for daily life and social activities. Factors in the relevant planning laws and regulations issued by the central and local governments were identified and filled into this framework (Table 1), which later would be processed by FAHP analysis.
Data Source: Fuzhou municipal government and various national regulations.

Legislation index
LI consists of two mandatory factors: the land type (LT) and the land acquirability (LA). The former one is used to define the location searching areas, whereas the latter one is used after suitable locations are found. In the current regulatory plans, there are two types of lands suitable for CHCC to use: either lands closely related to residential functions or lands closely related to public management and public service. According to China's current land use classification, three types of lands are suitable for CHCC: ① reserved, vacated, and vacant residential lands (R21, R22); ②share land with nursing homes (A6); and ③ share land with hospitals (A51). Moreover, although residential lands mixed with commercial (B1) and office (B2), such as commercial and residential lands (RB), are not recommendable sites to build elderly supporting facilities, the current building on them can be transformed and reused as CHCC. Therefore, six abovementioned recommended land types are used as mandatory factors in the optimal location evaluation. In case that some community cannot find any suitable land for their CHCC under this condition, lands of land types B1, B2 and RB can be used as a compensation.
The land acquirability (LA) deals mostly with the economic aspects of land use, including renting or purchasing the occupancy rights. This is a crucial factor for site selection, since it is usually the local government who provides spaces according to the care givers' need. To "build CHCC software and soft power, supplemented by hardware and/or facility construction", the local government would invest most on the light assets, e.g. the services and management, rather than on the heavy assets, e.g. the ownership of the building.

Hazard index
HI consists of both large-scale disaster (LSD) and the environmental health risk (EHR). Located on the southeast coastline of China, the most frequent hazard of our study area is flooding brought by the monsoon and tropical storms. Before going into the East China Sea, the Min River (闽江), the largest river in Fujian, had brought rich soil to its alluvial plain where Fuzhou (Foochow), the capital city of Fujian Province, has been siting since 202 BC. Fuzhou thus has relatively low altitudes in her urban areas and used to be covered with numerous streams. In recent year's urban sprawl, some of the streams have been filled in and had been causing floods and waterlogging problems from time to time. Other geological hazards include landslides and mudslides. It is generally required by the current regulation to build CHCC at least 50 m away from a (hidden) geological disaster point. Nevertheless, in case that some community might not find any suitable land for their CHCC under this condition, on the condition to intervene the Water Resources Department, some waterlog risk areas could be used. In addition, it seems noteworthy that the altitude data is excluded in the GIS analysis, due to the fact that lands on steep slopes have been identified as non-construction lands and, care centers in urban areas are required to be located on relatively flat sites with slopes less than 10%.
With a forest coverage rate of 58.36%, ranking second (after Hangzhou) among capital cities nationwide, the environmental health risks (EHR) is relatively low in Fuzhou. Pollution index is a mandatory index. In urban areas, pollution mainly come from industrial and municipal infrastructure facilities. The sewage treatment plant (U21) is one of the primary urban pollution source. CHCC should be located more than 300 m away from these buildings. In addition, it is necessary to locate care centers in the downwind direction.

Convenience index
CI consists of both mobility index (MI) and facility index (FI). The MI includes factors that encourage both short-(SDM) and long-distance mobility (LDM). To encourage SDM, the main entrance of the care centers should be opened to urban branch roads rather than arterial roads with busy traffic to prevent neighbourhood traffic accidents. To encourage LDM, the CHCC should be connected to public transport, with a closest bus/ metro station within 500 m.
The FI, on the other hand, refers to urban and community facilities including evacuation shelters, general hospitals/clinic, community health service centers/ stations, urban and community open spaces and neighbourhood stores. Studies in different countries all suggest that the public facilities frequently used by the seniors should be located within their walking range, i.e. 300 to 500 m, to meet their physical and psychosocial needs. Considering the current regulations, 500 m service radius of public facilities in the community is selected. Nevertheless, a service radius of 300 m could be considered when condition permit, and it is used as a buffer in the ArcGIS calculation in this study for the suitable plots.
① General hospitals/clinic As seniors' hospital visits are usually accompanied by the caregivers, a driving time of 30 min is chosen for medical services in general hospitals as previous studies suggested . This distance of 20 km at an average speed of 40 km/h would have covered the whole Fuzhou urban areas.
② Community health service centers (CHSC) and community health stations (CHS) CHSC (or community hospital) and CHS both provide community residents with general practice medical cares, including vaccine, rehabilitation and health education services, and in different scales. CHSC, usually occupying an independent plot, can be found in each sub-district, whereas CHS are usually found in commercial buildings in each communities.
③ Parks, community open space and evacuation shelters The accessibility and service radius of open spaces is also an optional factor. They come in variety forms of community parks, including street and sidewalk pocket parks (both referred to as dots in the regulatory plan without specified size), community playground, city parks, greenways/greenbelts and beaded parks. Greenways and beaded parks have grown in great numbers lately in Fuzhou, connecting community parks to form a continuous, crisscrossed urban green

The study areas and data sources
Fuzhou has long been in the club of ageing societies, with residents aged 60 and above making up 16.76% of the 8.29 million municipal population (more than 3.01 million in our study area) and the population aged 65 and above making up 11.72%. Thus, there is a huge demand of CHCC in Fuzhou. The per capita disposable income of all residents in Fuzhou is approx. 2100 USD in the first quarter of 2021 and that of urban residents is approx. 2600 USD, which indicates an economic growth strong enough to support the development of CHCC. Central Fuzhou as the study area is chosen for three considerations. First, as one of the statesponsored 203 CHCC pilot cities/areas since 2017, Fuzhou has already archived the full coverage of CHCC in the urban area with at least one CHCC per constituency. Second, Fuzhou is a livable provincial capital, which has a comprehensive middleand-up development level in China. Third, only the study areas are under the administration of the municipal government, whereas other areas are all under the administration of the district government. Therefore, the study area, rather than the whole city, could a representative case with applicable for other city in China.
In our study area, there are currently 39 constituencies, which is the administration unit for National Congress Reprehensive election, in Gulou (鼓楼), Taijiang (台江), Cangshan (仓山) and Jin'an (晋安) Districts in central Fuzhou and 267 communities within its jurisdiction. By the end of January 2020, there are in total 46 district-level care centers (Figure 2) in these four urban districts, all of which are located in mature communities with dense population and high ageing rate compared to the newly developed urban districts.
The key data used in this study come mainly from two government departments: Fuzhou Urban Planning Bureau and Fuzhou Civil Administration Department. Planning data mainly obtained from the regulatory planning and various specific planning includes comprehensive transportation planning, drainage special planning, drainage and waterlogging special planning, geological disaster prevention scheme, and etc., whereas CHCC data include the address, building area, number of bed and other relevant data obtained from Fuzhou Civil Administration Department and all verified by site visits. The population data are obtained from the sixth and seventh national census published data, statistical yearbooks and working reports, official website of Fuzhou of Statistics Bureau and other open resources.
By visiting these care centers, data of attributes of the total 46 current CHCC, including the exact location, floor area, center types, building types and the community and the surrounding neighbourhood the CCHC serves, were obtained by site visits done between May and June 2019. Generally, there are two center types: day centers (17, 37%) with no beds for seniors to stay overnight and long-term rehabilitation/medical care centers (29, 63%) with beds for seniors to stay overnight. Moreover, all of the 46 CHCCs are found in re-used buildings, transformed from commercial, residential or other community facilities, and only seven centers are located in individual buildings and all the rest share buildings with other facilities. Thus, generally there are three building types ( Figure 2): ① independent building/land (7, 15%); ②sharing a building with other community facilities (11, 24%); ③ occupying a building remodelled from other land uses, which is not consistent to the appropriate land types (28, 61%).
It is a common practice to convert a CHCC from other land use as long as it is consistent to the suitable land types, as the local government and the care providers both invest on the services and management, rather than on the ownership of the building. Nevertheless, as no planners had involved in the site selection procedure so far, the legislation and the safety of some of the land use remain questionable. Moreover, inappropriate locations have also resulted in centers with limited service hours restricted by their commercial landlords or overall insufficient community services as centers occupied spaces for other community facilities.
Eighteen centers are found located in spaces that violate higher-level planning and have to be relocated to recommended sites as soon as possible. According to the Fuzhou's Geological Hazard Prevention and Control Plan (2019), there are 50 geological disaster (hidden hazards) sites in Fuzhou, among which 17 disaster sites are in the urban area. Besides, the "Specialized Plan for Drainage and Waterlogging Prevention" calculated the high-risk areas of waterlogging once in 50 years. And there are 125 industrial lands in the forms of industrial parks in Fuzhou's urban areas, which are mainly first class (M1). Although they have little noise, no pollution and no potential safety hazards to the neighbourhood nearby, the residential areas inside M1 are unsuitable for CHCC. The "Specialized Drainage Plan" shows that there are five sewage treatment plants in the study area. In addition, in some cases that a care center occupies the site of public facilities other than CHCC, the services of the public facilities would inevitably fail to meet the requirements of the relevant specifications and regulations, and it needs to be relocated.

Evaluation methods
To validate the location finding strategy in the current urban planning system, each index in the framework of this study corresponds to the actual detailed regulatory planning data as shown in Figure 3.
A combination of vulnerability filtering and FAHP was used to evaluate the 46 current CHCCs. As the LI and the HI are both mandatory, all the sites either with the inappropriate land type or in danger of large-scale disaster (LSD) and/or the environmental health risks (EHR) are mapped out on the ArcGIS's data platform. FAHP was implemented to give weights to each optional CI, as it introduces more uncertainty in criteria weights assignment and reduces the burden of many spatial analyses with GIS (Caprioli and Bottero 2021). GIS is applied to implement different categories of indicators on the map according to the weight, mapping out locations according to their degree of suitability for CHCC to be relocated.
FAHP analysis is applied in the index weighting phase, which is useful for differentiating the importance level of each CI compared to others in achieving the evaluation objective. A panel of experts, including five chief planners/senior architects, three managers of adult care centers and four civil administrators, were consulted for filling in the pairwise comparison matrixes. At the sub-criteria level, each expert answered a specific questionnaire containing pairwise comparison questions related to their field of expertise. The priority of each element is extracted by the matrix of the eigenvectors, which numerically represent the synthesis of the experts' preferences. Since in a FAHP, the preference judgements are expressed through triangular fuzzy numbers, i.e. a range of confidence in the assignment of the values in the pairwise comparison, these preference judgements were converted into triangular fuzzy numbers to fill out the matrix of triangular fuzzy numbers, simply assigning to m the value expressed by the expert, l = m-1 and u = m + 1, respectively. Similarly, all pairwise comparison matrices with the triangular fuzzy numbers were performed, the eigenvector calculation and the estimation of the priority vector. The results of these procedures are shown in Table 2.

Relocation methods
To relocate the CHCC on unsuitable sites, with the maps produced by the FAHP, two location models are used according to the center types, as covering model fits the long-term rehabilitation and the medical centers the best, whereas center models fit the day centers the best (Li et al. 2017(Li et al. , 2019. The coverage location model is implemented on the population heat map. The distribution density of those aged 60 or over in each constituency was calculated by its census population data, the average ageing rate and distributed evenly to residential communities. The aged population distribution density is regarded as the demand point of elderly care services. For CHCC with beds, the goal is to maximize the population each center serves.
The p-center location model takes the constituency boundary as its service scope. By calculating the time it takes to get the CHCC by walking along the road, the location goal is to minimize the maximum distances in the coverage area. This model is more suitable for those day centers where the seniors all live at walking distance.
Each center can choose from the results that suit itself, as we aim to facilitate the decision-making process with multiple stakeholders without making decisions for all. Moreover, these centers are constantly changing their services from time to time. Judging from different field visits, most of the early centers opened in 2017 were day centers. The majority changed into long-term care centers with overnight services to increase revenue. As such, each center is not assigned a location model according to its current service type. This is one of the major differences compared to previous location finding studies.

Current CHCC with land use problems
Comparing the location of existing 46 care centers in Fuzhou's four districts with the suitable lands we found, five CHCCs are located on suitable sites, whereas 18 centers are located in spaces that violate higher-level planning and have to be relocated to recommended sites as soon as possible. The rest 23 centers could stay in their current location on the conditions of taking necessary remedial measures according to the relevant regulations. The existing problems of the current centers can be divided into five categories (Figure 4): ① the compatibility of land use of 18 centers are inconsistent with higher-level planning; ② the entrances of 13 centers are opened to traffic arterial in the risk of traffic accidents; ③13 centers are located on urban waterlogging high-risk areas; ④ 27 centers are far from either CHSC or disaster shelters, which result in inaccessibility and inconvenience; ⑤ 23 centers lack open space in the vicinity which connect to the community.

FAHP results
The result of FAHP on subsidiary CI shows that, as shown in Table 2, the experts rated that the priority of the FI is much higher than that of the MI at Level 2. At Level 3, compared to LDM, the importance of SDM seems predominant. The Community health care centers/stations are regarded the most important in the FI, which is followed by the availability of facility within walking distance (SDM) and the community open space.
In the Choice phase, all standardized maps are aggregated among each other using a weighted sum function to obtain the overall suitability map, according to the following equation: where Sj represents the suitability for pixel j, Wi the weight of factor i and Xi the standardized index score of factor i. The aggregation consists of two parts, both for the values obtained with FAHP. The According to Fuzhou's regulatory detailed planning, there about 3,000 eligible plots in the study area, with land areas ranging from 369 to 582,477 m 2 . Combined with the HI map, which marked out all the plots in risk of hazards such as flooding, waterlog, landslides and mudslides, the suitability shows that a substantial portion of the territory analyzed is vulnerable/sensitive and unsuitable for hosting CHCC, as well as the suitable area map. For the case study under analysis, the suitability map implemented with FAHP gives us the degree of land suitability with highest medium and low values in the central Fuzhou. Apart from the specific results of this case study, the suitability map highlights a large number of most suitable areas (green color) with high value of suitability, suitable areas (yellow) with median suitability and need relatively small number of improvements, and those tolerable areas (orange) with low suitability and need relatively large number of improvements. Locating of the current 46 CHCCs on the suitability map make it possible to tell what to do with them: whether these centers stay where they are, or need improvements, or need to be moved to suitable new locations. The results are shown in the optimal suitability map ( Figure 5).
The optimal CHCC locations show that eight constituencies either have no suitable type of land for CHCC or most of their lands are located in vulnerable/sensitive areas. To deal with this problem, two of the subsidiary LI are relaxed. First, lands of types B1, B2 and RB can be used as compensations. It is possible to use architectural design methods such as adding a new exit for the CHCC, which can avoid conflicts between different functions. Second, on the condition of the Water Resources Department intervening and approving, some waterlog-risk areas could be used. The compromised new evaluation is done for all areas, but the results of only those eight constituencies with no suitable lands otherwise were kept and combined with the optimal suitability map ( Figure 5) to form a revised suitability map ( Figure 6). Still, there are 19 centers, as we can see in the revised suitability map, that are not located on suitable lands, which need to be relocated according to the points of needs in the Population Heat Map.

Relocation
The results of coverage and p-center location models are shown in Figures 7 and 8, respectively. Among them, 13 centers are inconsistent with the land use planned and six are located on lands in risk of hazards (excluding abovementioned eight constituencies). Figure 7 gives the suggestion of new locations (in red) for those long-term centers with beds for seniors to stay overnights, whereas Figure 8 gives the suggestion of new locations (in green) for those day centers. As the centers are still evolving themselves, we cannot make the suggestions according to their current types.

Discussion
This paper is different from pervious studies in terms of objective, method and data. It aims to explore feasible methods to facilitate the location finding process with  multiple stakeholders rather than making planning decision for them. In Methodology, the two levels of factors that it use are introduced: mandatory and advisory. The violation of mandatory factors ends any further discussion, and it will hugely eliminate the location searching possibilities. FAHP can be applied to determine the weight for each advisory index afterwards. Regarding data, it analyzes actual planning data. In recent years, the China government had made great efforts to make more data more accessible by academic request, but there exists barriers, even between government departments. For example, realtime population data can be obtained only in the public security system, and urban and rural planning data are in the planning system. It is thus rare for urban planning research to analyze actual planning data, which separate the research from the reality and thus difficult to apply.

Suitability maps
Suitability maps represent an easy way to show the results of the methodology applied, and they allow all stakeholders to participate in the debate, thanks to the easy-visual GIS representation. This is important in the localization debate as it brings into play different points of view with strong conflictual positions, which can be managed and discussed in the decision-making process starting from clear and not questionable results. The visualization of the results may also be useful in other decision problems, such as for example the undesirable sites (such as waste incinerators or landfills) where the phenomenon of social opposition, often called NIMBY (Not in My Back Yard), is particularly strong and disruptive.

Possible improvement measures
Eighteen out of the current 46 (39%) care centers have land use problems. A possible but not feasible solution to these problems is to change the regulatory plans in accordance with the procedures for attached public facilities in such small scale. Among them, five centers are located on comprehensive land with residential functions. They can have a separated new entrances added or improve local environments depending on the specific situation.
Among the current care centers, 21 (46%) and 6 (13%) centers are located too far from CHSC/CHS and evacuation shelters, respectively. These problem can be solved by associated medical teams and thoughtful emergency evacuation plans, respectively.
The main entrances of 13 care centers are directly opened to city main road or need to cross other plots to enter. If possible, we suggest to modify these entrance together with the internal layout.
There are 14 centers (30%) located in waterlogging high-risk areas. Measures can be taken after applying for assessment by the relevant government departments, such as the Water Resource Department or the Construction Bureau.
There are 23 centers (50%) that do not have adjoining green spaces or play grounds. In case that it has to cross the city's main and secondary roads to access the site, the care center's location is not improvable. Nevertheless, in case that the road those aged have to cross is a branch road or a below-grade road, the care centers are fine to use under the guidance of the caregivers.

Limitation of this study
The granularity of data on population, road network and actual traffic has prevented from accurately calculating all the accessibility for each facility. Nevertheless, in this research project, we started to establish cooperation with relevant government departments and some of the service suppliers, who both require us to stand in the perspective of the whole picture and expect effective and quick responding tools for decision-making.

Conclusions
This study discusses in depth how to evaluate and improve the current CHCC siting in urban areas according to land use, natural hazard frequency and the accessibility to facilities. In particular, this study tried to explore the feasible approaches for urban planners to enhance the process of planning and implementation participated by multiple stakeholders through sharing the rich fruits of location science with communities, the care providers and the relevant government departments, without making all the decisions for them.
The suitability analysis of the current CHCC locations was implemented in two steps. First, the inappropriately sited centers were identified by the multiple indexes of legality, vulnerability and convenience, which is summarized from the relevant policies, regulations and research in accordance with the current urban planning system through FAHP analysis and were emulated and visualized by GIS. Second, the ill-located centers recommended more suitable plots identified by p-center or coverage location models according to the type of these centers. As a result, five centers are perfectly located, while 19 centers need to be relocated, as they were located on sites that were either inconsistent with the appropriate land use or in the risks of natural hazards. The rest half (22 centers) need improvements to different degree to meet all the relevant planning requirements.
This research method quantifies the policy and normative clauses and integrates the academic achievements in the current planning system. The case studies is proved to be and can be used to improve the effectiveness of planning implementation.
To some extents, this method reflects the idea of "multi-plan integration", which can be applied to other land uses in China, especially those of the subcontracted urban functions, adapting to the trend of more and more mixed use of land uses today, making urban planning more feasible and effective.