6.2 Study area, materials, and methods
In this chapter, to further illustrate how the local government and residences work against depopulation in Kitakyushu, the study unit is street. Fig. 6-3(b) shows the distribution of streets in Kitakyushu in 2000. The data was downloaded from the Portal Site of official Statistics of Japan (https://www.e-stat.go.jp/). The total streets of Kitakyushu in 2000 is 1521.
6.2.2 Materials
Different from Chapter 3 and 4, where the dependent variable is the population change ratio, in this chapter, the number of population change (NPC) was selected as the dependent variable. In chapter 3 and 4, as the magnitude of population varies greatly between cities, the population change ratio is more suitable than NPC as dependent variable for it following a normal distribution. However, within a city, the NPC is more normally distributed than the population change ratio. In addition, there are some streets where there have been no residents, streets where there were residents and now there are no residents, and streets where there were no residents and now there are residents. Hence, the NPC was a more suitable choice as dependent variable.
The study period was selected as from 2000 to 2015. The compact city issue and its accompanied policies and strategies in Japan were first raised in 2003 and Kitakyushu is one of the first Japanese cities to push for compact city. We used the data in 2000 to represent the situation of the city before the compact city issue, and the data in 2015 is the closest to the current situation. To explore the inner correlates of NPC, we selected in total 29 variables from demographic system and urban vitality system. Moreover, the variables were divided into 2 kinds of situation including before "compact city"
and variation since "compact city". The demographic factors were collected from the population census data in 2000 and 2015 from the Portal Site of official Statistics of Japan (https://www.e-stat.go.jp/). The elevation data of Kitakyushu was collected from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation Model Version 3 (GDEM 003), it was downloaded through the USGS EARTHEXPLORER website (https://earthexplorer.usgs.gov/). The average elevation and average slope of each street were calculated using ArcGIS 10.8. The other urban vitality factors were based on the building data in Kitakyushu and calculated using ArcGIS 10.8.
Table 6-1. Description of explanatory variables for urban shrinkage in Kitakyushu.
System Situation Variable Abbreviation
Demographic factors
Before
"compact city"
Total population TP
Aging population ratio APR
Underage population ratio UPR
Total households TH
House ownership ratio HOR
Total employed staff and workers TE
Proportion of employees in secondary and tertiary industries
PESTI
Variation since
"compact city"
Change ratio of total population CRTP
Change ratio of aging population CRAP
Change ratio of underage population CRUP
Change ratio of total households CRTH
Change ratio of total employed staff and workers CRTE Change ratio of proportion of employees in
secondary and tertiary industries
CRPESTI
Urban vitality factors
Before
"compact city"
Average building age ABA
Building coverage ratio BCR
Average floor area ratio AFAR
Total floor area of commercial buildings TFACB Total floor area of residential buildings TFARB Total floor area of educational welfare facility
buildings
TFAEB
Total floor area of office buildings TFAOB
Average elevation AE
Average slope AS
Variation since
"compact city"
Change ratio of building coverage ratio CRBCR
Change ratio of floor area ratio CRFAR
Increase floor area of commercial buildings IFACB Increase floor area of residential buildings IFARB Increase floor area of educational welfare facility
buildings
IFAEB Increase floor area of office buildings IFAOB CHAPTER SIX: CORRELATION ANALYSIS OF POPULATION CHANGE
AND URBAN VITALITY IN KITAKYUSHU
6.2.3 Methods and research flow
Fig. 6-4 Research flow of evaluating urban regeneration against depopulation in Kitakyushu.
In this chapter, the methods have mentioned were designed to model the correlation between number of populations change and its potential related demographic and urban vitality factors. We first applied both global and local Moran’s tests to reveal whether the spatial dependency of depopulation in Kitakyushu. Because the global Moran’s test result showed the overall spatial autocorrelation was not strong that the GWR and SGWR regression analysis may not suitable for this case. However, the local Moran’s test showed that there were several streets clusters in Kitakyushu. So, we extracted those clusters and applied zonal statistics and correlation analysis to evaluate the urban regeneration in those regions.
The Moran’s tests were applied to find whether the distribution of shrinking streets is concentrated in space. Spatial autocorrelation analysis, including Global Moran’s I statistics and Local Moran's I statistics were performed to reveal the spatial dependence of population change. Global Moran’s I, which is a rational number ranged from -1 to 1 after normalized variance was selected for spatial autocorrelation analysis for its widely used in revealing the global spatial autocorrelation. Global Moran's I >0 means positive spatial correlation, the larger the value, the more obvious the spatial correlation; Global Moran's I <0 means negative spatial correlation, the smaller the value, the greater the spatial difference; otherwise, Global Moran's I = 0, indicates a random space.
However, the Global Moran’s I could only reflect on the spatial autocorrelation but do not identify the location and type of spatial clusters. The Local Moran’s I can be applied to identify the local differences and similarities among neighboring streets. The Local Indicators of Spatial Association (LISA) can be determined using Local Moran's I. Generally, four clustering/outlier types are classified using the Local Moran’s I including (1) high-high cluster (HH); (2) high-low outlier (HL); (3) low-high outlier (LH); (4) low-low outlier (LL). HH and LL reflect the positive spatial correlation; HL and LH reflect the negative spatial correlation. Therefore, both the Global and Local Moran’s I statistics for analyzing the correlation of population variation between each street. In this chapter, as the size and scale of the streets were quite different and the streets were not normally distributed, we selected
the adaptive bi-square as the kernel type, and the k was selected as 30. In this chapter, HH refers to an above-average number of populations increase in a local street, with the same characteristic exists in its neighboring streets. HL refers to an above-average number of populations increase in a local street, whilst the population decreased in the neighboring streets. LH refers to the shrinkage in a local street, whilst an above-average number of populations increases in the neighboring streets. LL refers to the shrinkage in a local street, and its neighboring streets have the same characteristics. Thus, we used ArcGIS 10.8 to statistic the Global/Local Moran’s I, which shows the spatial relationship between the population change in a street and its neighbors for Kitakyushu.
Then the Pearson Correlation analysis was conducted to reveal the correlation between NPC and each explanatory variable. In statistics, the Pearson correlation coefficient is a statistic that measures linear correlation between two variables X and Y. It has a value between 1 and −1, where 1 is total positive linear correlation, 0 is no linear correlation, and −1 is total negative linear correlation. And it can be calculated through Eq. 6-1.
𝜌𝑋,𝑌=𝑐𝑜𝑣(𝑋, 𝑌)
𝜎𝑋𝜎𝑌 (Eq. 6 − 1)
where cov is the covariance, X is NPC, Y is the explanatory variable in Table 6-1, 𝜎 is the standard deviation.
CHAPTER SIX: CORRELATION ANALYSIS OF POPULATION CHANGE AND URBAN VITALITY IN KITAKYUSHU