**4. Discussion**

In this paper, the intensity of the urban landscape for seven metropolitan areas in Poland was assessed. Additionally, the landscape of RM in Germany was examined as a comparative area for US-ZM—the only polycentric agglomeration in Poland. The use of the ULII index showed that the landscape of the US-ZM differs from the landscape of other metropolitan areas in Poland (higher average values of ULII). At the same time, the landscape of the US-ZM shows some similarities to the landscape of RM (similar values of ULII). Moreover, MAW also shows some similarities to the US-ZM in terms of landscape diversity. What is more, the study showed that urban landscapes occur only in the central zones of the metropolitan areas that were delineated on the basis of official documents. In turn, rural and transitional landscapes predominate as far as percentage of their surface is concerned.

The similarity of the US-ZM and RM landscapes may result from the similar factors that influenced development of these metropolitan areas. In both cases, the mining industry was a factor in the location and development of cities [49]. Moreover, the spatial nature of the agglomeration, polycentric in the case of US-ZM and RM, and monocentric in the case of other metropolitan areas, has an influence on the landscape differentiation and similarities which occur [50]. The value of ULII could also be influenced by a degree of industrialization. High values of ULII occur in former industrial districts such as Nowa Huta (Kraków), Bałuty, Fabryczna (Łód ´z), Psie Pole, Kowale (Wrocław) and Młyniska (Gda ´nsk).

The value of ULII could be influenced by many different factors. The initial comparison of the spatial distribution of ULII indicates that among these factors transport network systems and their density, and the spatial organization of metropolitan areas (polycentric or monocentric) can be distinguished [51]. The occurrence of raw materials is also of grea<sup>t</sup> importance as they stimulate the economic development of metropolitan areas (this is in the case of US-ZM and RM) [52], and their primary and secondary functions [53,54]. Furthermore, the natural environmental conditions are crucial [55]. They may be favorable (hydrographical network, access to the sea, favorable topoclimatic conditions) or unfavorable (high mountain areas, boggy areas, proximity of protected areas) for the spatial development of metropolitan areas [56–58]. The crucial factor that should be taken into account when interpreting the value of ULII is the process of changes on the border of metropolitan areas in official administrative documents. This process is connected with decisions about incorporating particular municipalities that are of different types: rural, urban or urban–rural. Furthermore, economic potential [59], spatial policy and land

prices [60–62] may have an influence as well. Many aspects of urbanization are interpreted through globalization [63].

Nevertheless, the most visible aspect is the relations between the values of ULII and transport network systems and industrial districts. However, this issue requires further recognition in our study area through the statistical analysis of the correlation that occurs between these variables and the construction of a regression model that would allow the occurring relationships to be quantified. Such analyses were conducted by Conway and Hackworth in the Greater Toronto Area [34]. A detailed analysis of landscape urbanization in the context of driving forces in our study area will be the subject of a separate study by the authors. Nevertheless, some basic conclusions could be drawn. The analysis of the hexagons level reveals a star-shaped pattern of the urban landscape in MAW. Additionally, in US-ZM and RM the intensity of the urban landscape is visible in a linear pattern. This is due to the emergence of the urban sprawl process along the transport network which is typical of other metropolitan areas [64–66]. In MAW, the urban landscape is visible along the A2 motorway, and the S8 and S7 expressways along the east-west and northsouth routes. The main railway lines to the capital run parallel to the road network. In US-ZM and RM, the intensity of the urban landscape increases along the east-west line, which is connected with the railway between Gliwice and Mysłowice and the A4 motorway in Poland, and the A2 and A42 motorways in Germany. The relationship between the development of the road network and the processes of suburbanization has been confirmed by Garcia-Lopez [67] who, based on the example of Barcelona, stated that the construction of new road infrastructure generates urban sprawl processes. Baum-Snow [68] puts forward the thesis in a similar way, examining the relationship between suburbanization and road development in the United States. Moreover, it is worth noting that the ULII is influenced by the types of urban coverage, which include residential buildings related to urban sprawl processes, as well as industrial areas and large-surface communication junctions. Thus, the transport network influences the level of ULII both directly and indirectly.

Similar research has already been carried out in Olsztyn and Sieradz (Poland) [25,26]. The areas classified as urban in Olsztyn cover only 24% of the city's area. In turn, in Sieradz non-urban areas also prevail. The percentage of agricultural areas within the boundaries of metropolitan areas in Poland was studied by Sroka et al. [36]. Their study shows that 49.9% of the areas administratively belonging to metropolitan areas are occupied by farms. Similar comparative studies conducted for RM and US-ZM showed that agricultural areas cover 39.2% of the RM area and 42.7% of the USZM [38]. These values are lower than the results presented in this paper (81% of seven metropolitan areas in Poland are covered with rural landscapes, 52% in RM, and 55% in US-ZM, respectively). However, the differences result from the different classifications of rural landscapes. In this paper, rural landscape includes arable land, pastures and forests. Hence, this landscape type in our paper has a broader scope. The applied approach allowed the assessment that only in the US-ZM and RM is it possible to distinguish the core of a metropolitan area, in which urban landscapes have the largest percentage among all the analyzed areas. A comparison of landscape types in municipalities and hexagons shows differences in the accuracy of the results. In accordance with the adopted criteria, there are no urban landscapes in KMA, LMA and WMA at the level of municipalities, while at the level of hexagons their occurrence is visible. On average, the share of rural and transitional landscapes is higher at the municipality level than at the hexagon level. Hexagons illustrate with greater accuracy the spatial distribution of landscape types enabling the justification of the existing layout of urban landscapes and further spatial analyses, e.g., related to the study of urban development in terms of driving forces. Moreover, research at the hexagon level is objective and based on mathematical logic, as opposed to the level of boundaries of municipalities and metropolitan areas, whose scope is determined administratively.

As already mentioned in the introduction, many authors use the CORINE database for landscape analysis [30–32]. Nevertheless, it is worth noting that despite many advantages, this database has some limitations [69]. However, this database is imperfect in small-scale

studies, while in landscape analysis it is an appropriate source [33]. Admittedly, there are many other databases that may be used in such studies. For instance, in Europe, higher resolutions have Urban Atlas maps—a project developed as part of Global Monitoring for Environment and Security. Nevertheless, maps are created only for selected areas around large cities, so they do not cover the entire EU territory [69]. Another example of the increase in the spatial resolution are the "fourth-level" CLC maps, however, they were prepared only in some countries and classifications used in them are inconsistent with each other. This paper confirms the statement that the CORINE database is useful in landscape analyses, both in relation to administrative units (municipalities, metropolitan areas) and geometric basic fields. It is a particularly reliable source in studies of urban areas undergoing constant change [31]. It must be emphasized that the value of ULII could be influenced by many different factors e.g., different types of a dataset, map scale, type of basic units (their size and shape). While using ULII these limitations must be taken into account.
