Introduction

Innovation, the main driver of regional economic development, has become a popular research topic (Capello and Lenzi, 2015; Lema et al. 2015; Horbach and Rammer, 2022; Crescenzi et al. 2007). Regional innovation has evident spatial correlation (Wang et al. 2022; Capello and Lenzi, 2019; Zhang, 2013; You et al. 2021), and the spatial pattern of regional innovation is an important component of regional innovation research (Jörn et al. 2022; Ma et al. 2018; Liu and Qu, 2021; Fu et al. 2016).

Current researches on the spatial patterns of innovation usually take countries or provincial, municipal and county administrative regions as spatial units to portray regional innovation. Moreno et al. (2005) explored the spatial distribution of innovation activities and the technology spillovers in knowledge creation and diffusion in 175 regions of European countries and found a correlation between R&D expenditures and agglomeration economies. Hou et al. (2016) used provinces as spatial units to calculate the innovation capability index and used cluster analysis and other methods to divide the country into four levels. Cao and Qin, (2012) used patent data to measure urban innovation capabilities and concluded that the distribution of regional innovation capabilities in China exhibits the same spatial pattern as regional economic development. Lv and Li (2010) established an innovation capability evaluation mechanism based on factors such as knowledge, technology, institutions, and services to explore the spatial patterns of China’s innovative urban system. Jiang (2014) found that the distribution of the innovation outputs in Zhejiang province, China is “high in the north and low in the south, high in the east and low in the west”.

The above studies treat the spatial units as a whole, overgeneralize the experience of selected units to an entire area, and analyse the spatial patterns of innovation in the study area from a macro perspective (Liu, 2010; Wang et al. 2021; Gonçalves and Almeida, 2009), fail to consider the internal spatial distribution characteristics of regional innovation (Ma et al. 2017; Wang and Sun, 2017). This is not conducive to making specific plans in line according to the distribution of innovation capabilities within the city, such as promoting the implementation of innovation supporting resources, increasing the influence of innovation hot spots, and enhancing the development vitality of innovation cold areas.

To overcome the limitations of using administrative divisions as spatial units in studying the spatial patterns of regional innovation, the urban technological innovation capability was evaluated from a customized smaller-scale geospatial unit. Duan et al. (2015) used postal units as spatial units, studied the spatiotemporal evolution patterns of urban innovation in Shanghai and Beijing. Relying on the geographical coordinates of patents, Hu et al. (2022) divided the spatial units, designed the innovation capability and innovation structure indexes of the spatial units, and revealed the spatial patterns and evolutionary characteristics of the industrial innovation in Shenzhen. These studies improve the accuracy and provide new insights and methods for research on the spatial patterns of regional innovation.

In this study, we take Jiangsu province as the research area and 2019 as the time node. Using the method proposed by Hu et al., we performed geocoding and spatial unit division based on the addresses of the principal patent applicants, calculated the innovation capacity index of each industry within each spatial unit. The Moran index and standard deviational ellipse were used to depict the spatial patterns of innovation in Jiangsu province macroscopically (overall view) and meticulously (local view). It fully reveals the spatial patterns of technological innovation capability in Jiangsu province, which breaks through the limitations of only using provinces, cities and counties as the overall evaluation in previous articles. While taking into account the overall innovation characteristics, it focuses on the spatial distribution of innovation within the city, which helps accurately target urban innovation clusters and innovation -deficient areas and provides a scientific basis for the development of urban innovation capabilities in Jiangsu province.

Data and methods

Study area and data

Jiangsu province, the study area of this paper, is one of the most active and innovative regions in China. Patent data, as one of the main results of innovation activities, can reflect the innovation capacity of a region (Acs and Audretsch, 1989; Cheng and Kung, 2019; Klein, 2020). From the perspective of invention and technology, there is a close relationship between patents and innovation. Patents represent the technological output of the innovation output capacity, and the information obtained by analysing patents has an approximate 85% probability of being representative of the characteristics of the innovation in the corresponding industry (Wang, 2009). This study is conducted based on the patent data of 2019 to exclude the impact of the COVID-19 pandemic, using the number of granted patents as an indicator of technological innovation capability to discuss the technological innovation capability based on patent data analysis.

Patent data are obtained from the Wanfang Data Knowledge Service Platform, an authoritative platform in China. Each piece of patent data includes the abstract, type, application date, classification number, applicant/patentee, inventor/designer and address of the principal applicant which describes the spatial location information of the patent. According to the types of patents and the Industrial Classification for National Economic Activities (GB/T 4754-2011), the patents are classified into three categories, including 2089 primary industry patents, 279,403 secondary industry patents, and 113,882 tertiary industry patents. Because primary industry patents are few, this paper only used secondary and tertiary industries patent data for research.

Spatial processing of data

The addresses of the principal applicants in the patent information are geocoded to obtain spatial coordinates by using the Baidu Maps application programming interface. The spatial coordinates in the Mars coordinate system obtained from Baidu Maps are converted into spatial coordinates in the World Geodetic System 1984.

The patent data after geocoding is represented as a large number of discrete points in space. In this paper, the research area is divided into areal spatial units (Van Groeningen et al. 1997; Li et al. 2020) and then discrete point data are assigned to the faceted spatial units according to the method of Hu et al. (2022). According to the distance between the patent points and the grid, the patent points are allocated to the nearest grid. The size of the grid divided by the space is calculated using Formula 1:

$$I = \sqrt {\frac{{2Area}}{{Num}}}$$
(1)

In Formula 1, I represents the side length of the grid element, Area represents the area value of the study area, and Num represents the number of discrete points. Through calculation and rounding, the side length of grid cells is 750 meters.

Spatial analysis methods

Technological Innovation capability index of a spatial unit

In this paper, the technological innovation capability index of a spatial unit is used to represent the technological innovation capability of the spatial unit (Hu et al. 2022). The higher the spatial technological innovation capability index is, the higher the technological innovation capability of the spatial unit, and vice versa. The technological innovation capability index of a spatial unit is the logarithm of the total number of patents in the spatial unit. The spatial technological innovation capability index is calculated using Formula 2.

$$IC = ln\left( {\mathop {\sum}\limits_{i = 1}^n {P_i + 1} } \right)$$
(2)

In Formula 2, IC is the spatial technological innovation capability index, Pi is the number of patents in the spatial unit, and n is the number of spatial units.

Spatial autocorrelation

In this paper, spatial correlation is analysed by calculating the global Moran index and the local Moran index. The global Moran index can be used to assess the degree of correlation in the overall space. Its value is between −1 and 1. A positive value indicates a positive correlation, a negative value indicates a negative correlation, and the value of 0 indicates random distribution (no correlation). The closer the absolute value is to 1, the stronger the correlation (Molaei Qelichi et al. 2017).

Local Moran index is introduced to evaluate local correlation (Pavlov and Koroleva, 2014). The calculated local Moran index values can be used to generate an agglomeration map. If the value is positive, the attribute values of the same class are similar; if the value is negative, the attribute values of different classes are similar.

Spatial distribution characteristics

The standard deviation ellipse is used to assess the degree of dispersion of the spatial distribution of data (Furfey, 1927). The centre of the ellipse is the weighted average centre of the spatial data, and the weight of each unit is the technological innovation capability index of each unit. The long axis of the ellipse indicates the direction of the data distribution, and the short axis indicates the range of the data distribution.

Spatial patterns of technological innovation capability

Overall spatial pattern of technological innovation capability in Jiangsu

The introduction of overall technological innovation capability

Based on the technological innovation capability of each industry in the spatial units, the distributions of the technological innovation capability index of the secondary and tertiary industries in Jiangsu province was plotted, as shown in Fig. 1.

Fig. 1: Distribution of the technological innovation capability index of the secondary and tertiary industries in Jiangsu province.
figure 1

The map of (a) overlays the vector data of the municipal administrative districts of Jiangsu province and the raster data of the technological innovation capability index of the secondary industry. The map of (b) shows the same content related to the tertiary industry.

Full size image

According to the magnitude of the technological innovation capability index, technological innovation capability is divided into five levels—namely, low, relatively low, medium, relatively high, and high—by using the natural discontinuity point classification method. The technological innovation capability index is relatively evenly distributed among the cities. The regions with low technological innovation capability are distributed most extensively, and the regions with high technological innovation capability are distributed least extensively.

Overall, the spatial pattern of the innovation capabilities of the secondary and tertiary industries in Jiangsu province is strong in the south and weak in the north. In this study, Nanjing, Wuxi, Changzhou, Suzhou, Nantong, Yangzhou, Zhenjiang and Taizhou belong to southern Jiangsu, and the remaining five cities belong to northern Jiangsu. The overall distribution of the secondary and tertiary industries in southern Jiangsu shows the spatial distribution characteristics of “one core and one belt” with Nanjing as the innovation core, and Suzhou, Wuxi and Changzhou as the southeast-northwest-trending innovation agglomeration belt. The overall distribution of the technological innovation capability index in northern Jiangsu is relatively scattered, and the technological innovation capability index values are mostly distributed in the central area of each city and are mostly low and medium. Based on the patent data of the secondary and tertiary industries in Jiangsu province in 2019, the technological innovation capability index is calculated, and then the global Moran index is calculated. The global Moran index values of the innovation capabilities of the secondary and tertiary industries in Jiangsu province are 0.315 and 0.328, respectively, indicating that the overall technological innovation capability of Jiangsu province shows a positive spatial correlation and that the spatial correlation of the tertiary industry is higher than that of the secondary industry.

The gathering of technological innovation capability

The method described in Section “Spatial processing of data” is used to divide the grids, calculate the local Moran index based on the technological innovation capability index, and explore the agglomeration pattern of technological innovation capability in the province, as shown in Fig. 2.

Fig. 2: Agglomeration pattern of the technological innovation capability of the secondary and tertiary industries in Jiangsu province.
figure 2

The map of (a) overlays the vector data of the municipal administrative districts of Jiangsu province and the raster data of the agglomeration pattern of the technological innovation capability of the secondary industry. The map of (b) shows the same content related to the tertiary industry.

Full size image

The local Moran index is used to assess the degree of similarity in local areas. The secondary and tertiary industries in Jiangsu province show very similar agglomeration patterns. Their high-, medium- and low-value regions exhibit roughly similar distributions, with most medium- and high-value regions concentrated in Nanjing, Suzhou, Wuxi and Changzhou, and a few medium- and high-value regions distributed in Xuzhou, Lianyungang, Huai’an, Yancheng and Suqian, showing obvious spatial heterogeneity. The agglomeration of the innovation capacity of the secondary and tertiary industries roughly shows dotted, patchy and zonal distributions. Patchy and zonal distributions dominate in southern Jiangsu, which show an extensive distribution of high-value regions and an obvious H-H agglomeration pattern, indicating that agglomerations are widespread and relatively high level in southern Jiangsu. In contrast, northern Jiangsu is dominated by dotted distributions, with patchy distributions in local areas, limited H-H agglomeration, and no significant spatial correlation in some areas.

The innovation capacities of the secondary and tertiary industries show relatively high-level agglomeration in the Suzhou-Wuxi-Changzhou area and Nanjing, and relatively low-level agglomeration in Yangzhou, Taizhou and Nantong. The H-H agglomeration groups present relatively obvious characteristics:

  1. (1)

    These groups occur mainly in the Suzhou-Wuxi-Changzhou area in a zonal distribution. Therefore, the Suzhou-Wuxi-Changzhou area is viewed as the innovation core of the secondary and tertiary industries in southern Jiangsu. These industries are concentrated in Kunshan; districts of Gusu, Wuzhong and Industrial Park in Suzhou; districts of Huishan, Xishan, Liangxi and Xinwu in Wuxi; and districts of Wujin, Tianwu and Zhonglou in Changzhou.

  2. (2)

    There is a concentrated distribution of H-H agglomeration groups in the districts of Gulou, Xuanwu and Qinhuai in Nanjing, indicating that Nanjing is the secondary core of the innovation capacity of southern Jiangsu.

  3. (3)

    There are only a few H-H agglomeration groups in Yangzhou, Taizhou, Zhenjiang and Nantong, and these agglomeration groups are located mainly in south-central Yangzhou, central Taizhou, northeastern Zhenjiang and southwestern Nantong. The number of H-H agglomeration groups in these four regions is far less than that in the Suzhou-Wuxi-Changzhou area and Nanjing. This finding indicates that the agglomeration level of technological innovation capability in these four regions is lower than that in the Suzhou-Wuxi-Changzhou area and Nanjing.

Nanjing is the core of the H-L agglomeration pattern in southern Jiangsu, and the Suzhou-Wuxi-Changzhou area is in second place. Similar to the H-H agglomeration groups, the L-H agglomeration groups are distributed mainly in Nanjing and the Suzhou-Wuxi-Changzhou area, and sporadically distributed in Yangzhou, Taizhou and Nantong. The L-H agglomeration units are accompanied by the H-H agglomeration units. The higher the number and concentration of H-H agglomeration units are, the higher the number of L-H agglomeration groups. The L-L agglomeration pattern is rarely seen in Figures.

The differences of technological innovation capability in industries

Based on the technological innovation capability index of the secondary and tertiary industry, the difference of the technological innovation capability of the secondary and the tertiary industry in Jiangsu Province is explored by calculating the difference of the index, as shown in Fig. 3.

Fig. 3: Distribution of the difference in the technological innovation capacity index of secondary and tertiary industries.
figure 3

The map uses the secondary industry technological innovation capability index minus the tertiary industry technological innovation capability index to reflect the spatial differences in the technological innovation capability of the secondary and tertiary industries.

Full size image

From the overall situation, the difference in the technological innovation capability index of the secondary and tertiary industries is divided into three levels. Greater than zero means that the innovation capacity of the secondary industry in the region is higher than that of the tertiary industry, and less than zero indicates the opposite situation, while equal to zero means exactly equal or null.

From a numerical point of view, there are more areas greater than zero than areas less than zero, indicating that the innovation capacity of the secondary industry in Jiangsu Province is generally higher than that in the tertiary industry.

From the perspective of spatial distribution, there are more areas greater than zero in many cities, which are relatively concentrated than the areas less than zero, which are more closely distributed around the positive concentration area. This shows that the development of the secondary industry innovation industry in many cities in the province is better and centralized.

The tertiary industry is developing around the secondary industry innovation concentration area, and the overall urban technological innovation capability is mainly driven by secondary industry innovation. However, there are exceptions. One is Nanjing, where the number of areas greater than zero and less than zero is equivalent, indicating that the innovation and development level of the local secondary and tertiary industries is basically the same, moving hand in hand around the city centre. The other is Suzhou and Wuxi, where there are more widely distributed areas of less than zero, indicating that the two cities have stronger technological innovation capability in the tertiary industry, which will have more development potential in the future.

Spatial pattern of technological innovation capability in southern Jiangsu

The overall spatial analysis

Because southern Jiangsu has active innovation activities and complex and distinct spatial distribution patterns, this paper further analyses the spatial pattern of the innovation capacities of the secondary and tertiary industries in southern Jiangsu.

Based on the technological innovation capability index values of the secondary and tertiary industries in southern Jiangsu, the standard deviation ellipse and the average centre are used to investigate the spatial distribution centre and development direction of the innovation capacity of the study area, as shown in Fig. 4.

Fig. 4: Spatial distribution pattern of the technological innovation capability index of the secondary and tertiary industries in southern Jiangsu.
figure 4

The map of (a) enlarges the technological innovation capability index of southern Jiangsu in Fig. 1a and overlays the vector data of the major railway networks in Jiangsu province and the standard deviation ellipse. The map of (b) shows the same content related to the tertiary industry.

Full size image

The elliptical centre of the tertiary industry in southern Jiangsu is more obviously deviated to southeastern Jiangsu than that of the secondary industry. From the perspective of spatial distribution characteristics, the difference between the long and short axes of the standard deviation ellipse of the technological innovation capability of the secondary and tertiary industries is relatively large. This finding indicates that technological innovation capability has obvious directionality (southeast-northwest-trending). The short axis indicates a strong cohesion and concentrated index values. The area of the standard deviation ellipse of the tertiary industry is smaller than that of the secondary industry, indicating that the units with technological innovation capability in the tertiary industry has a smaller distribution range.

The spatial distribution of the railway network in southern Jiangsu shows that there are railways in areas with relatively concentrated innovation capacity and that the areas with high innovation capacity are mostly distributed along the railway lines or at the nodes of the railway network. In the spatial distribution characteristic of “one core and one belt” the belt is roughly distributed along the railway line.

One core—Nanjing

Specifically, the technological innovation capability indices of the secondary and tertiary industries in Nanjing are shown in Fig. 5. Compared with other cities in Jiangsu province, Nanjing has fewer regions with low innovation capabilities and more regions with medium and high innovation capabilities. Nanjing’s innovation-intensive districts are concentrated in districts of Gulou, Xuanwu and Qinhuai in the city centre, and extend to the districts of Jianye, Yuhuatai, Jiangning and Qixia. In Pukou and Luhe districts, a stripe-shaped region with medium-high innovation capabilities and of a certain scale is formed in the area close to the city centre, and a few regions with medium and low innovation capabilities are evenly distributed in the areas far from the city centre. Lishui and Gaochun districts have a few regions with high innovation capabilities in the centre, and regions with medium and low technological innovation capability are sporadically distributed in the other districts.

Fig. 5: Distribution of the technological innovation capability index of the secondary and tertiary industries in Nanjing.
figure 5

The map of (a) enlarges the technological innovation capability index of the secondary industry of Nanjing city in Fig. 4a and overlays the vector data of the municipal administrative districts of Nanjing. The map of (b) shows the same content related to the tertiary industry.

Full size image

One belt—the Suzhou-Wuxi-Changzhou area

Figure 6 shows the innovation capacity index of the secondary and tertiary industries in the Suzhou-Wuxi-Changzhou area. In the area, Suzhou has the highest innovation capacity, while Wuxi and Changzhou have higher regional innovation capacities in regions near Suzhou. Regions with low and medium innovation capacities are commonly seen in this area, while regions with high innovation capacities are rarely seen. The innovation capacity of the tertiary industry in this area is higher than that of the secondary industry, forming a southeast-northwest-trending active innovation zone.

Fig. 6: Distribution of the technological innovation capability index of the secondary and tertiary industries in the Suzhou-Wuxi-Changzhou area.
figure 6

The map of (a) enlarges the technological innovation capability index of the secondary industry of Suzhou-Wuxi-Changzhou area in Fig. 4a and overlays the vector data of the municipal administrative districts of Suzhou-Wuxi-Changzhou area. The map of (b) shows the same content related to the tertiary industry.

Full size image

Suzhou has the highest overall technological innovation capability, is the core of innovation spillover in the “one belt”, which drives the innovative development of two neighbouring cities, namely, Wuxi and Changzhou. The technological innovation capability index of Suzhou shows the highest values in districts of Gusu, Xiangcheng, Wuzhong, Industrial Park and eastern Huqiu district. The high values of the regional technological innovation capability index are densely distributed in northern Wujiang district and southern Taicang. There is a significant high-value area of technological innovation capability in northeastern Kunshan. The technological innovation capability index values show a relatively even distribution in Zhangjiagang and Changshu, with low and medium values in most parts of these cities and a few high values distribute in the central parts.

Regions with medium and high innovation capacity in Wuxi are distributed mainly in districts of Liangxi, Xinwu and Huishan, and extend to Xishan city and Binhu district. Zhonglou and Tianning districts, southern Xinbei district and eastern Wujin district in Changzhou have high-innovation areas, while Jintan district and Liyang city in Changzhou have low concentrations of regions with innovation capacity, and the innovation capacity index values in these regions are generally low and medium.

Conclusion and discussion

Conclusion

This study replaced administrative regions with spatial units based on geographic coordinates of patent data, calculated the innovation capacity index of secondary and tertiary industries within each spatial unit, further investigated the spatial patterns of the technological innovation capability in Jiangsu province. The innovation capacities of the secondary and tertiary industries in the province show two distribution patterns. Patchy and zonal distributions dominate in southern Jiangsu and the overall distribution show the spatial distribution characteristics of “one core and one belt” with Nanjing as a core and Suzhou-Wuxi-Changzhou area as the southeast-northwest-trending belt, while dotted distributions dominate in northern Jiangsu.

The innovation capacities of the secondary and tertiary industries in Jiangsu province exhibit positive spatial correlations, and the innovation regions of the tertiary industry greatly influence each other. The number of H-H agglomeration units is the largest among the four types of agglomeration units in Jiangsu with obvious characteristics, that each city has the most H-H agglomeration units, forming a number of high-level innovation activity concentration areas, especially the Suzhou-Wuxi-Changzhou area, followed by Nanjing, and the agglomeration level in other areas is not high, and the others are hardly seen in Figures.

In the majority of cities, the overall urban technological innovation capability is mainly driven by secondary industry innovation. But the technological innovation capability of the secondary and tertiary industries is about the same in Nanjing, and the tertiary industry is comparably dominant in Suzhou and Wuxi.

In southern Jiangsu, the spatial pattern of technological innovation capability is formed with Nanjing as the innovation core and the Suzhou-Wuxi-Changzhou area as an agglomeration belt with a high technological innovation capability. This pattern is related to the spatial distribution of the railway network.

From an industrial perspective, the technological innovation capability of the secondary industry is relatively low compared with the tertiary industry, and thus less spillover effect on neighbouring regions and a limited impact range, so the industry needs further development.

From a spatial perspective, the innovation capacities of the Suzhou-Wuxi-Changzhou area and Nanjing are stronger than those of Zhenjiang, Yangzhou, Taizhou and Nantong. The innovation capabilities of the secondary and tertiary industries have relatively obvious directionality (southeast-northwest-trending), relatively obvious cohesion and concentrated index values. The average centre of the tertiary industry is closer to Suzhou, so Suzhou more greatly affects the innovation activities of the tertiary industry. In Nanjing, the innovation development of the secondary and tertiary industries is relatively balanced, and there are comparable numbers of regions with high levels of innovation capacity in city centre, and low, medium levels of innovation capacity in other places. In the Suzhou-Wuxi-Changzhou area, Suzhou is the core of innovation spillover to promote the technological innovation capability development of other cities. In this area, the number of regions where the innovation capabilities in the tertiary industry is high exceeds the number of regions where the innovation capabilities in the secondary industry is high, and the development of industrial innovation is higher-level and more promising.

Discussion

As the main research result of innovation activities, patents are an effective means to measure the innovation capacity of cities. Most of the innovation capacity measured by patent data can reveal the technological innovation capability in a certain region, such as the researches done by Cao and Qin (2012) and Gonçalves and Almeida (2009). However, patent data cannot fully reflect the technological innovation capability of cities, mainly because simple patent counts do not take into account differences in the quality and economic impact of patents on actual innovation.

Besides, this paper uses only patents as the research object, and does not consider the relationship between scientific research papers and innovation capability. The reason is that the papers have not been truly transformed into actual innovative capabilities can that apply in new inventions or innovative technologies. And the geographical positionings of the papers are often concentrated in universities, thus spatial analysis through theirs publishing unit will lead to inaccurate conclusions. About how to improve, scientific research papers can be taken into account as potential technological innovation capability, so as to strengthen the reference index of technological innovation capability and improve the comprehensiveness of the paper.

About the space unit size determination, the experience of spatial sampling optimization is referenced (Van Groeningen et al. 1997; Li et al. 2020; Hu et al. 2022) to guarantee the uniformity of spatial unit distribution. In order to ensure the accuracy and rationality of the grid size, this study has made a distribution map of the technological innovation capability index of secondary industry in Jiangsu Province to compare, where the grid side is too small (500 meters) and the grid side is too large (1000 meters), as shown in Fig. 7.

Fig. 7: Distribution of the technological innovation capability index of the secondary industry with different grid side lengths.
figure 7

The map of (a) overlays the distribution of the technological innovation capability index of the secondary industry with a grid side length of 500 meters. The map of (b) shows the same content with a grid side length of 1000 meters.

Full size image

Judging from the map effect, if the size of the grid is too small (500 m), the distribution of the technological innovation capability index may be difficult to recognize. On the contrary, when the size of the grid is too large (1000 m), the distribution of the technological innovation capability index may lack of accuracy.

Judging from the results presented, in southern Jiangsu, the same conclusion can be drawn from the 500-meter grid unit, but the innovation activities in northern Jiangsu are scattered, and the concentration tendency of scattered high-innovation areas is not obvious. If the 1000-meter grid unit is used for analysis, it can be difficult to identify the distribution pattern of innovation activities within the region where it is small but has a concentrated distribution of technological innovation capability, which makes it difficult to conclude the specific “one core and one belt” model in Nanjing and Suzhou-Wuxi-Changzhou area.