163Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.DOI: 10.15201/hungeobull.72.2.5 Hungarian Geographical Bulletin 72 2023 (2) 163–178. Introduction The city-region has been a focal point in the history of travel, from the time of Thomas Cook to present day, but particularly during the dynamic development of the tourism in- dustry from the 1960s onward (Hua, H. and Wondirad, A. 2021). Settlements located in an agglomeration are integral supporters of tourism of the nearby metropolis like sat- ellites orbiting in the Earth’s gravitational field, ensuring the efficient communication of people. These so-called satellite settle- ments provide additional labour force and purchasing power for the successful opera- tion of the metropolis, moreover, they also induce flows, especially because of their role in the tertiary or quaternary sectors (Burke, J. 1986; Merrilees, B. et al. 2013). The essence of the phenomenon described in this study as satellite tourism is the tourist mobility be- tween the large city and its surroundings, without overnight stays (i.e. day trips; Stetic, S. et al. 2011). The nature of satellite tourism may be similar in many respects to the tourist behaviour observed in transit set- tlements, namely in terms of length of stay and expenditure (Kincses, Á. et al. 2017). The 1 Geographical Institute, Research Centre for Astronomy and Earth Sciences (MTA Centre of Excellence), Budaörsi út 45. H-1112 Budapest, Hungary; University of Pannonia, Egyetem utca 10. H-8200 Veszprém, Hungary. E-mail: michalko.gabor@csfk.org 2 Geographical Institute, Research Centre for Astronomy and Earth Sciences, Budaörsi út 45. H-1112 Budapest, Hungary. E-mails: prorok.marton@csfk.org, kondor.attila@csfk.org 3 Corvinus University of Budapest, Fővám tér 8. H-1093 Budapest, Hungary. E-mail: noemi.ilyes@stud.uni-corvinus.hu 4 GEOInsight Ltd., Budapest, Brassó út 5. H-1112 Budapest, Hungary. E-mail: szabo.tunde@geoinsight.hu Mobility patterns of satellite travellers based on mobile phone cellular data Gábor MICHALKÓ1, Márton PROROK2, Attila Csaba KONDOR2, Noémi ILYÉS3 and Tünde SZABÓ4 Abstract For a long time, tourism statistics were the only reliable source of information on tourism mobility. Tourism statistics are inadequate for the analysis of tourist mobility within state borders and across Schengen Borders without using registered accommodations. Big data offers the opportunity to gain a better understanding of tourism movements, for example, same-day tourist flows in metropolitan areas. Here, we introduce the concept of the satellite travel- ler to more effectively investigate the nature of tourism between the large city and its surroundings. As tourists communicate via cellular devices, the use of mobile phones offers an opportunity for researchers to explore the mobility pattern of tourists. In this article, we discuss the specificities of mobility in Hungary by SIM card users registered in foreign countries. The analysis is based on the Telekom database. We seek to answer the question to what extent the information from the satellite tourists’ mobile phone use can help to understand their movements and to identify frequented places less commonly accounted for in tourism statistics. The most important findings of our investigation are (1) the confirmation of former knowledge about spatial characteristics of same-day tourist flows in the Budapest Metropolitan Region, (2) the insight that far away settlements are also visited by satellite travellers, and (3) the methodological limitations of mobile phone cellular data for tourism mobility analysis. Keywords: big data, metropolitan region, same-day visit, unconventional tourism, Budapest Received February 2023, accepted May 2023. mailto:prorok.marton@csfk.org Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.164 main difference is the ease of returning to the nearby city, which, unlike transit tourists, does not put satellite travellers under time pressure, allowing them to spend their free time more comfortably, get to know the place better and enjoy more colourful experiences. The need for a better understanding and more accurate observation of satellite tour- ism induces the development of the available methodological arsenal. The understanding of certain phenomena and processes of met- ropolitan tourism has been a focus of research studies for a long time (Girardin, F. et al. 2008; Yang, Y. 2012; Kádár, B. and Gede, M. 2013), but the geographical extent of the same-day tourist mobility between the metropolis and its agglomeration zone has been neglected. This topic fits well with invisible tourism under the theoretical umbrella of unconventional tour- ism mobility (Kádár, B. and Gede, M. 2022; Kovalcsik, T. et al. 2022; Timothy, D. et al. 2022; Tóth, G. and Kincses, Á. 2022) and enriches the knowledge about destinations involved in same-day visits (Wynen, J. 2013a, b) and their archetypical consumers, the day trippers (Murillo, J. et al. 2013). In our study, we seek to answer the ques- tion of the geographical range of foreigners’ journeys to and from Budapest, the capital of Hungary, that are taken within a 24-hour period. Budapest is a popular destination for international tourists (especially among young visitors) because of its favourable geographical location, wide range of tourist attractions (e.g., baths, ruin bars, UNESCO World Heritage Sites) and relatively affordable prices for goods and services (Irimiás, A. 2010; Pinke- Sziva, I. et al. 2019). In addition to Prague (16.4 M) and Vienna (15.3 M), Budapest (9.4 M) also recorded significant amounts of international tourist nights in 2019, with the Austrian capital having the strongest link in terms of tourist flows (Kádár, B. and Gede, M. 2021). Like other European capitals (Freytag, T. 2010), the surroundings of Budapest are very popular with international day trippers, mainly Szentendre (pedestrian street milieu), Gödöllő (royal palace), and Visegrád (pano- rama on the Danube river bend). This study offers an opportunity to improve the methodology for tourism-oriented cogni- tion based on the cellular information of mo- bile phones. Given the negative impact of the COVID-19 pandemic on tourism, the research was based on data from 2019, the last undis- rupted year for passenger traffic. A substantive part of the study was based on the analysis of a huge database of geo-coordinates from mo- bile phone calls placed within the municipality in 2019, using simple and complex mathemati- cal, statistical and geoinformatics methodolo- gies. The database contains information on the calls (data traffic) of foreigners who made the first and last calls from Budapest on the selected days between 0–24 hours and who also made at least one other call (data traffic) in Hungary. We seek to answer the question to what extent the information from the satel- lite travellers’ mobile phone use can help to understand their movements and to identify frequented places less commonly reflected in tourism statistics. The results contribute to broadening the theory of unconventional tourism mobility to a deeper understanding of the nature of satellite tourism and to the development of a tourism-oriented research methodology based on mobile cellular data. Big data in tourism mobility research The increasing use of info-communication technologies (ICT) and big data, including mobile phone network data for modelling location and urban mobility networks, has opened new perspectives in modelling hu- man mobility (Wang, Z. et al. 2018; Huang, B. and Wang, J. 2020). Individuals leave digital footprints in geographical space. The human digital footprint is a widely studied area that increasingly uses data integration methods to generate information on individuals’ spa- tial presence and movement. Data sources can either be terrestrial or applied to various remote sensing technologies (Sagl, G. et al. 2014; Sagl, G. and Resch, B. 2015). Terrestrial data sources are usually called geospatial big data. They are comprised of information on 165Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. daily human activities. Data are collected ei- ther by data service providers in the form of mobile cell and cellular signalling data and GPS positioning data (Steenbruggen, J. et al. 2013; Vanhoof, M. et al. 2018; Zhang, Y. et al. 2019; Mata, F.J. et al. 2020) or by business ser- vices as a by-product provided in the form of POS-terminal data or housing transactions data, or they can be crowed-sourced by social media check-ins and fitness applications, etc. (Thakuriah, P. et al. 2020). The spatial and temporal resolution of geospatial big datasets has grown over time. Telecommunication net- work data, called Call Detail Records (CDR), are widely involved in social and geographi- cal analyses, specifically in urban studies (Louail, T. et al. 2014; Pucci, P. 2015; Jiang, S. et al. 2016; Razavi, S.M. et al. 2018; Egedy, T. and Ságvári, B. 2021). Human mobility patterns show specific regularities: individually varying travel dis- tances and a defined probability of returning to frequented locations are the most striking characteristics (González, M.C. et al. 2008). A periodic recurrence in mobility patterns implies that individual spatiotemporal dis- tribution is heavy-tailed. That means most people spend most of their time in a handful of locations. Meanwhile, a smaller set of indi- viduals spends their time in a large variety of locations, that contribute less than 1 percent to an individual’s time consumption (Bayir, M.A. et al. 2009). Candia, J. et al. (2008) high- lighted that spatiotemporal dynamics follow daily and weekly commuting patterns in ag- glomeration areas. The periodicity observed in the CDR data enables us to focus on higher spatiotemporal granularity. The penetration of the social physics ap- proach and its analytical framework has been accelerated by the growing number and depth of the available datasets. Statistical data and surveys are appropriate measures for recognizing the inherent structure of tourism-related mobility networks and ex- ploring the behavioural patterns and moti- vations of the actors involved. High-scale human mobility databases, however, provide significantly more information in terms of representation and behavioural pattern dis- covery compared to classical statistical data collection and processing methods. Graph-analytical tools, on the other hand, prove to be insufficient for the inclusion of the geographical dimension as an explanatory factor in tourism research (Hannam, K. et al. 2014). Research about interconnectedness in tourism, such as destination relationships, can be conducted by analysing the internal and external links of spaceless nodes. In this case, the online connectivity (e.g., browser search- es) and offline connectivity (e.g., sequences of places which were visited on a trip) of the network nodes that constitute a local system can be compared with other local tourism networks. The network topology of tourism nodes can be approached in a similar way in global comparisons (Baggio, R. 2020). Unlike spaceless network investigations, those that apply spatial dimensions in a systematic way and as an explanatory factor are often conducted for policymakers (Chu, C.P. and Chou, Y.H. 2021), often in the fields of tourism transport infrastructure planning (Qian, C. et al. 2021) or heritage tourism (Liu, Z. et al. 2022). A detailed overview of the databases used for tourism research and visitor tracking has been given by Reif, J. and Schmücker, D. (2020). In destination management, not only the patterns and dynamics of visitation but also the order of places visited and the time spent at a given location can provide valuable information (Ahas, R. et al. 2008, 2010; Raun, J. et al. 2016; Saluveer, E. et al. 2020; Luštický, M. and Štumpf, P. 2021). Mobile cellular data for better understanding of daily tourist flow There are critiques on the efficacy of geo- spatial big data (including mobile cellular data), which have been relevant in our work as well. Among these, the positioning ac- curacy resulting from the tower coverage currently is an inescapable limitation, which significantly hinders the resolution of human mobility mapping. Another limiting factor is Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.166 the event-dependent nature of the sampling. Data is only generated when an event has oc- curred on the network. Smartphones tend to log in at least once every 30 minutes but de- vices using less complex technology do not communicate with towers outside of events (Reif, J. and Schmücker, D. 2020). Daily visi- tor traffic and commuting traffic can be simi- larly defined by specifying a sleeping point and daytime stopping point. The sleeping point can be identified by the location of the first and last events that occurred during the day and can be interpreted as the individu- al’s location at home. Our analysis targets daily visitor traffic. Therefore, we are looking for individuals who are spending their days away from their sleeping point but return for nights. That implies that in cases where the first and last event differs, even due to the inactivity of the device, the user is not includ- ed in the searched manifold, and the user’s mobility data is not incorporated into our dataset. The phenomena of data loss are a commonly cited shortfall of big data process- ing. However, there are techniques to reduce the amount of data lost. The more precisely we are able to define the target population, the less data loss we experience. In this study, our target population was satellite travellers using the Hungarian MTelekom network with a foreign SIM card. Our interpretation was that they started their first and initiated their last event of the day from Budapest but also gave a signal from a location outside Budapest during the day. The target group of tourism-purpose visitors can be easily separated from the professional traf- fic of non-Hungarian residents based on the functional distribution of places. Commuting out of the city is a non-typical movement, even among Hungarian SIM card holders, as the spatial distribution of workplaces in the Budapest Metropolitan Region is concentrated first and foremost in the central city and less significantly in the extended industrial areas located in the functional metropolitan area and the urban periphery (Dövényi, Z. and Kovács, Z. 2006; Szabó, T. et al. 2014). Workplaces are concentrated in specific delineated metro- politan areas, corresponding to industrial areas (Egedy, T. et al. 2017). The Budapest Metropolitan Region shows an extremely con- centrated distribution regarding institutional amenities (only one university is located there) and recreation and sports facilities, which im- plies that daily commuting outside of the city for study and other purposes is also infrequent among Hungarian citizens. We concluded that the spatial distribution of workplaces in the Budapest Metropolitan Region differs signifi- cantly from the topography of places of tourist destinations. Therefore, we can safely isolate the tourism traffic of non-Hungarian mobile phone users from professional traffic. In our case study, we sought to understand the processes, driving forces and dynamic changes in space when it comes to non-Hun- garian satellite tourism traffic in order to be able to correctly interpret a large amount of data and the phenomena it reveals. For this reason, we decided to adopt a classical social geography approach without using the net- work science tools applied in mainstream sci- ence. Network science tools are mainly used to analyse the interconnectedness of spatial relationships, which have little added value in studying a local system and interpreting its operational regularities. Clustering as a methodological tool for generalising human movements We consider quantitative tools such as spa- tial-temporal clustering valuable data reduc- tion and generalisation methods. We adopt the interpretation of Yuan, Y. and Raubal, M. (2012) when we use the inter-municipal clustering method, which provides a gener- alised description of individual movement processes: „Since individuals are atoms in an urban system, the spatio-temporal char- acteristics of an urban system can be viewed as a generalization of individual behaviour; therefore, mobile phone data also provide new insights into the analysis of the mobility patterns in urban systems.” (Yuan, Y. and Raubal, M. 2012, 26.) 167Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. The first step in the clustering process is selecting an appropriate set of variables con- tributing to a mobility model explaining non- Hungarian visitors’ movements. The cluster- ing procedure involves grouping records – in our study, municipalities – in such a way that any subject within the resulting group is closer to all other elements in the group than to any other element outside the group. The concept of ‘closeness’ can be understood here in relation to the variables under investiga- tion: for n observation units (that correspond to a specific settlement in the current case) if the number of variables under investigation is m, then the n observation units are placed in a standardised coordinate system of dimension m and then grouped through multiple itera- tions (Aldstadt, J. 2010). K-means clustering requires that the expected number of groups is determined in advance. In our case, we estab- lished six clusters in the first step and then re- duced the number of clusters until we obtained a stable number of clusters based on the size of the cluster populations and the location of the cluster centres. The K-means clustering proce- dure selects random cluster centres according to the predefined number of centres and then calculates the distance of all elements from the initial cluster centres. Suppose a specific record is closer to another cluster centre than its cen- tre according to the initial classification. In that case, the element is reclassified, the centres are recalculated, and the distance of the elements from the centres is measured. The iteration continues until all elements are assigned to the cluster centre closest to them. This multidimen- sional classification procedure contributes to typifying settlements according to the role they play in non-Hungarian visitors’ satellite tour- ism traffic (Lingras, P. et al. 2011; Bose, I. and Chen, X. 2015; Wang, W. et al. 2018). Data and experimental design Data source Our analysis was based on a mobile phone dataset provided by a Hungarian telecom- munications operator. Data acquisition was accomplished by CSFK (Research Centre of Astronomy and Earth Sciences) with Hun- garian Telekom Ltd (MTelekom), a member of the Deutsche Telekom Group. A research and development agreement was signed to explore the mobility performance of mo- bile phone users in Hungary in general and especially within the broader region of the Budapest Metropolitan Region. Hungarian Telekom is the long-term market leader in voice, multimedia and wireless services in mobile telecommunication, holding a stable market share of 45 percent in the Hungar- ian telecommunication market. According to the joint research agreement, Hungarian Telekom enabled CSFK to reach network sys- tem data via a SFTP-based interface. After a four-month-long period of data specification and data structuring and one extra month for the pilot run, data transmission started on 1 November 2017, and lasted until 30 Novem- ber 2019. Data was uploaded daily. Via the data transmission interface, a package of four data files arrived at CSFK every day for two years. Files contained data referring to the daily event traffic, equipment in use, equip- ment attributes and actual network coverage. Mobile phone network data analytics and visualisation reveal human mobility patterns reflecting individuals’ real-world spatiotem- poral dispersion. The geographical component of mobile telecommunication network data is registered, collected, and processed worldwide using similar technologies for data retrieval. By establishing the research database, we retained the rich-in-detail feature provided by the exceptionally high amount and scale of data (e.g., geographical and temporal res- olution, communication performance, etc.) and, at the same time, provided a compre- hensive framework for deploying mobile cell data in applied social sciences. Our research database was built on individual equip- ment records. Each object corresponded to a mobile device equipped with a mobile SIM card, carrying out the multimedia transfer on the MTelekom wireless network. Each record, therefore, pertained to an individual Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.168 user holding an MTelekom subscription or a non-Hungarian mobile device user using the MTelekom telecommunication network while staying in Hungary. The anonymity of the device holder was secured by hashing the equipment identifier by a randomly gener- ated artificial ID. The equipment identifier was recoded every 24 hours. As a result, the investigation of mobility histories was only available for 24-hour timeframes. Building the database Six designated days have been selected from the two years of data available. We involved three public holidays (15 March 2019, 20 August 2019, 23 October 2019), and three other regular work- ing days (Mondays) that followed the public holidays (18 March 2019, 26 August 2019, 28 Oc- tober 2019). By selecting the dates involved in the analysis, we intended to exclude the domes- tic satellite traffic as much as possible to avoid interference with foreign satellite movements. On the occasion of long weekends and pub- lic holidays, a higher proportion of Budapest citizens (of both foreign and national origin) leave the city for a longer period, so they do not take part in satellite traffic. We summarised the events registered by non-Hungarian satellite travellers staying at and departing for any dai- ly trip from Budapest. Satellite travellers were defined as device-holders registered abroad that started and ended their day in Budapest, meanwhile generating at least one event from a municipality other than Budapest between the first and last signal on the same day. A typical problem that arises while process- ing mobile cellular data is that the system data collected from the towers is extremely noisy and subject to errors due to the incorrect registration of towers. Data noise, among others, results in extreme speeds (> 300 km/h) when it comes to mapping individual mobility trajectories. The error rate might reach 20 percent of all detect- ed individual movements. Mobility trajectories are derived from registered telecommunica- tion events, as they store the exact time of the event and the approximate location of tower, which carried out the event. A telecommunica- tion event is defined as a user’s communication with a telecommunication tower, regardless of whether voice, text or data is distributed during the event. In the scientific literature, CDR are the data detected on telecommunication towers, which approximate the content of what we de- fine as an event. Event log data includes a time- stamp, the duration of the communication, the technology used, the nature and direction of the communication, etc. At the time of our study, 45,000 MTelekom towers served the national telecommunication network traffic. The geolocation of the cover- age area’s geometrical centre represents the towers’ positions. The exact coordinates consti- tute a business secret. We assigned each tower to the basic administrative units of Hungary (municipal level), according to the distance of the tower position to the nearest residential area. In the current study, we did not apply the filtering of extreme speed event records since the procedure used standard speed to estimate the time spent by users. As a result of the pro- cedure, extreme speed movement necessarily reduces the estimated time spent by users in each municipality to close to zero, so invalid records do not significantly change the satellite mobility model, and, at the same time, they do not cause data loss in the database. It is essential to highlight that mobile cel- lular data processing does not allow for the accurate localisation of users. The granularity of localisation refers to the spatial resolution of the telecommunication coverage areas, ranging from a hundred meters to a few ki- lometres. In densely served areas, such as the peri-urban zone, the coverage areas of the towers overlap several times. One of the consequences is the apparent hop- ping between municipalities: this occurs when a non-mobile user with a designated location close to the municipal boundary generates events on the towers assigned to one or the oth- er municipality, depending on the utilisation of the towers. Figure 1. highlights the set of set- tlements located right next to the border of the municipality of Budapest, where a high number of foreign satellite travellers seemingly spent a 169Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. Fig. 1. Number of satellite travellers and time spent in each municipality on 23 October 2019 (national holiday). Source: Authors’ own research. limited amount of time. The reason behind this is the previously defined oscillation along the border region. In this case, our data show that the user may apparently cross the Budapest boundary several times without actually mov- ing. This may result in incorrectly assigning a user to one or another municipality based on the events, and, as a result, falsely defining a user as a satellite traveller. We carefully consid- ered removing the subset of adjacent locations from the analysis. As a result of the delayed and congested stages of urban development of the post-socialist era, some of the traditional cultur- al and economic centres of the urban periphery – such as Szentendre and Gödöllő – have been incorporated into the functional metropolitan area. Because of the functional diversity and architectural heritage, some adjacent centres may be satellite tourism destinations based on their merits (Fabula, Sz. et al. 2021). Therefore, removing the affected settlements from the anal- ysis may imply a significant loss of data which would mask peculiar phenomena in an early re- search stage. We decided to keep the integrity of the initial dataset and carefully draw attention to the limitation of the methods applied. As we can see in Figure 1, a strong positive lin- ear correlation was observed, and municipalities fit the linear trend line. Vecsés, which accom- modates Budapest Ferenc Liszt International Airport (IATA: BUD), was an outlier in terms of the number of visitors and time spent. The air- port’s outstanding satellite traffic is due to its pe- ripheral location at the border between the city and the agglomeration. The same phenomenon occurred in the settlements at the city border: events were handled by towers located in both the agglomeration settlements and Budapest, making visitors’ locations seem to oscillate on the city-urban periphery border. All these uncertainties are pervasive for tele- communication big data sources, and indeed big data in general. Even so, the granularity, scope, frequency and representativity of the MTelekom dataset exceed most available data sources. This is because individual mobility histories are not even available for research purposes under the Hungarian corporate legal interpretation since the General Data Protection Regulation (2018) entered into force. Spatial features of satellite tourism in the Budapest Metropolitan Region Mobility formula for satellite tourism at the municipal level We elaborated upon a mobility model of satellite tourism at the municipal level using Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.170 minute-based location records and standard speed when calculating individual move- ment. Events were recorded at telecommu- nication towers in the form of CDR data. This dataset was complemented by the Customer Relationship Management (CRM) data gen- erated by the service provider MTelekom, which captured users’ demographics, sub- scription and device information. Only the country of subscription and device data were recorded for non-Hungarian users. If the service provider detected a non-Hungarian user participating in satellite tourism in the same municipality other than Budapest on two consecutive occasions, the time elapsed between the two events was attributed to that user (T1,2 (i) = t1 (i)), i.e. it was assumed that between the two events the user stayed at the municipality. If the user moved between two events, the airline distance between the geometric centroids (S1, S2) of the coverage areas was defined as the displacement vector (v1). The user’s individual mobility history was composed of a series of vectors (v1,2 (i)t1, v2,3 (i)t2, … vn-1,n (i)tn), summarised on a daily basis. In addition to the calculated values of the displacement vector, a record was made when the user continued to move from the previously registered settlement (L1) to the municipality registered after that (L2). In this case T1(i)L1, T2(i)L2 → v1,2(i)t1, where v1,2 corre- sponds to the displacement, and t1 is the time elapsed between events (event-minutes). The mobility model was designed to ac- count for users who spend, from a tour- ism point of view, the relevant amount of time in the location. Therefore, we needed to filter out transit users in each location. To do this, we introduced a standard dis- placement speed set at 60 km/h due to air- line movements. Knowing the speed and the displacement vector v1,2, the time spent travelling (travel-minutes, u1(i)L1,2) could be estimated. Given a location L1, the estimated time spent in the location (length of stay, t1’) is equal to the difference between the time spent between the events and the time spent travelling. The estimated time a user poten- tially spends in each settlement for tourism purposes, therefore, equals the time elapsed between the event logs in the two various places minus the time spent travelling: tn’(i) Ln = tn(i) Ln-1,n – un(i) Ln-1,n The lengths of stays calculated in this way were allocated to the municipalities con- cerned in equal proportions per user, i.e. half to L1(i), and a half to L2(i). We then summed up the estimated length of stay frequencies for each day, i.e. we determined how many users took up each length of stay value. We aggregated the number of users and lengths of stays on the selected dates separate- ly and then created three groups by simple weighted division into thirds. One-third of the users spent 1–12 minutes, another third spent 13–66 minutes and the remaining third spent more than 67 minutes (max. 1,281–1,440 min- utes) in the municipality concerned. For these three groups of lengths of stays, we mapped the daily distribution of satellite travellers by the municipality. In the first group, transit visitors, who touched the municipality dur- ing their daily movements, were accounted for. The second group included satellite trav- ellers who stopped in the municipality for a short period, for instance for resting, refuel- ling, shopping, etc. The third group included the satellite travellers we want to account for, who spend a longer period of time in a settle- ment before returning to Budapest. Interpretation of the mobility formula by introducing cluster analysis In order to understand the real-life implica- tions of the generalised mobility model, we introduced three spatial variables: distance from motorways, distance from industrial parks and distance from Budapest (Table 1). For each variable, the distance was given in linear distance calculated as the distance from the nearest motorway, the nearest in- dustrial area and the distance from the Buda- pest geometric centre to the geometrical cen- tre of the municipality concerned. The dis- 171Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. Table 1. Statistical parameters of the variables included in the explanatory model of K-means clustering (N = 3,154) Per settlement Range (= Maximum) Sum Mean Std. deviation Variance Minutes spent, total, 15 March 90,515 635,899 202 1,930 3,725,047 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 3,026 22,288 65,201 12,929 77,315 545,656 4 25 173 56 406 1,525 3,094 165,169 2,324,649 Pieces of equipment, total, 15 March 1,547 7,245 2 29 824 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 467 712 368 2,444 2,384 2,417 1 1 1 9 13 8 75 168 58 Minutes spent, total, 18 March 109,261 712,985 226 2,377 5,650,026 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 2,920 25,136 81,205 14,145 96,222 602,618 4 31 191 56 474 1,887 3,116 224,805 3,559,734 Pieces of equipment, total, 18 March 1,758 8,342 3 34 1,133 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 439 807 512 2,570 3,001 2,771 1 1 1 9 15 10 73 229 108 Minutes spent, total, 20 August 82,762 689,847 219 1,988 3,952,840 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 2,260 19,781 60,721 15,499 85,032 589,317 5 27 187 46 377 1,623 2,129 141,772 2,633,575 Pieces of equipment, total, 20 August 1,302 8,671 3 27 716 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 338 622 342 3,088 2,592 2,991 1 1 1 7 12 9 55 138 79 Minutes spent, total, 26 August 135,446 796,259 252 2,846 8,097,619 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 2,951 31,689 100,806 16,665 110,427 669,168 5 35 212 59 590 2,239 3,434 348,172 5,014,714 Pieces of equipment, total, 26 August 2,024 9,762 3 39 1,517 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 437 982 605 3,100 3,415 3,247 1 1 1 9 18 12 82 334 151 Minutes spent, total, 23 October 101,435 880,427 279 2,520 6,351,869 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 3,052 23,189 75,194 19,844 105,878 754,705 6 34 239 64 446 2,090 4,068 198,645 4,368,360 Pieces of equipment, total, 23 October 1,674 10,679 3 34 1,176 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 492 739 443 3,759 3,241 3,679 1 1 1 11 14 11 113 199 124 Minutes spent, total, 28 October 134,741 822,807 261 2,926 8,562,985 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 3,156 29,507 102,078 16,298 111,871 694,638 5 35 220 65 560 2,349 4,219 313,206 5,519,445 Pieces of equipment, total, 28 October 2,017 9,634 3 40 1,570 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) 490 920 607 2,967 3,416 3,251 1 1 1 10 17 13 109 303 162 Distance from motorways, km Distance from Budapest, km Distance from industrial areas, km 16 288 34 8,537 457,334 31,116 3 145 10 2 58 6 4,078 3,377,982 33,911 Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.172 tance from motorways was introduced into the explanatory model as a type of urbanisa- tion indicator. Our preliminary expectation was that the destinations preferred by for- eign visitors are necessarily equipped with urban services, i.e. they are also attractive to international tourists in terms of the quality of service. The distance from Budapest was assumed to be an essential variable for satel- lite tourism. Still, precisely because of the de- velopment of the motorway network, there was a narrow range of potential destinations that were generally excluded from the daily visitor flow due to distance alone. The distance from industrial parks was calculated using land use (OSM Landuse Landcover) data. 9,209 industrial land use sites were mapped in the whole country, with an average site area of 0.13 km2. Our analysis included only medium and large industrial sites greater than 0.5 km2 in size. A total of 200 industrial sites were selected across the country. The presence of indus- trial sites with a larger area was interpreted as an indicator of production activity in the municipality. Industrial production was as- sumed to be a tool for economic prosperity, especially in municipalities with less tourist attraction. Therefore, we expected an inverse correlation between the proximity of indus- trial parks and the number of visitors to mu- nicipalities involved in satellite tourism. We examined the correlation between the spatial structural indicators introduced in the model, such as the distance from motorways, industrial areas and the Budapest city centre, and the satellite traveller flow indicators, i.e. the number of visitors spending longer pe- riods and total time spent in the settlement (Table 2). Contrary to preliminary expecta- tions, Budapest’s proximity proved to have no impact on satellite tourism. Similarly, no correlation was detected between motorway access and satellite tourism. However, indus- trial areas showed a weak, significant nega- tive correlation with non-Hungarian satel- lite travellers’ mobility: the more attraction a municipality holds, the less the presence of industrial areas close to the municipality. This correlation suggests that the integrity of tourist attractions within the residential urban fabric determines the appeal for for- eign satellite travellers. It may also imply that settlements with a valuable heritage and appealing attractions have given less land area for industrialisation and logistics. If they have, these areas have been located at a suitable distance from tourism destinations of international interest. The result that the spatial dispersion of satellite traveller traffic does not overlap with the geographical dis- tribution of industrial production also serves to justify our methodology. Theoretically, we could incorrectly identify a non-Hungarian employee living in Budapest and working in the agglomeration zone as a satellite trav- eller. However, industrial areas’ proximity would positively correlate with visitor flow in this case. Thus, according to our definition, there is a high probability that the subset of non-Hungarian visitors filtered from the data excludes non-Hungarian commuting traffic for employment purposes. Finally, K-means clustering was used to identify subsets of municipalities with simi- lar characteristics based on the three correlat- ed variables (distance from industrial areas, minutes spent, and the number of visitors – Table 3), but which differ significantly from the “behaviour” of other clusters. The result was a four-tiered typology of settlements grouped according to their role in satellite traffic. The vast majority of municipalities could be grouped into two clusters, which were uni- formly unimportant in terms of visitor traffic but differed in economic nature from each oth- er. In the third cluster, municipalities such as Vecsés (BUD) and Szentendre were the most important destinations for non-Hungarian satellite travellers. Finally, the fourth cluster included a reasonably narrow range of mu- nicipalities, mainly in the Danube bend and the agglomeration, with a few rural locations. Watching the cluster-membership of mu- nicipalities (Figure 2), four major trends can be identified when looking at the spatial distribution of non-Hungarian daily satel- lite visitors: 173Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. Table 2. Co-movement of variables: correlation coefficient among the spatial-structural variables* Per settlement Distance from motorways Budapest industrial areas Pearson correlation Minutes spent, total, 15 March -.067 .014 -.092 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.045 -.037 -.073 .021 .018 .012 -.060 -.051 -.101 Pieces of equipment, total, 15 March -.050 .020 -.066 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.054 -.036 -.064 .023 .018 .017 -.067 -.051 -.086 Minutes spent, total, 18 March -.073 .016 -.102 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.054 -.047 -.079 .027 .024 .014 -.072 -.067 -.110 Pieces of equipment, total, 18 March -.056 .024 -.076 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.064 -.046 -.063 .031 .023 .018 -.079 -.064 -.088 Minutes spent, total, 20 August -.085 .019 -.100 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.074 -.051 -.091 .027 .024 .016 -.088 -.064 -.106 Pieces of equipment, total, 20 August -.075 .026 -.087 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.091 -.050 -.083 .033 .024 .019 -.100 -.063 -.095 Minutes spent, total, 26 August -.071 .017 -.094 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.069 -.045 -.076 .027 .020 .015 -.085 -.061 -.101 Pieces of equipment, total, 26 August -.061 .022 -.079 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.081 -.045 -.066 .030 .021 .017 -.096 -.067 -.090 Minutes spent, total, 23 October -.088 .024 -.112 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.075 -.057 -.092 .029 .022 .023 -.092 -.070 -.117 Pieces of equipment, total, 23 October -.076 .025 -.093 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.084 -.056 -.083 .031 .021 .021 -.100 -.069 -.103 Minutes spent, total, 28 October -.075 .019 -.100 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.066 -.051 -.079 .022 .025 .017 -.082 -.068 -.106 Pieces of equipment, total, 28 October -.063 .024 -.082 of which satellite transit (1–12 minutes) short stays (13–66 minutes) satellite tourists (67–1,281 minutes) -.074 -.050 -.067 .025 .025 .020 -.089 -.067 -.090 Distance from motorways Distance from Budapest Distance from industrial areas 1.000 -.007 .228 -.007 1.000 -.004 .228 -.004 1.000 *Distance from motorways, from Budapest and industrial areas, and satellite tourism metrics, such as the number of visitors and tourism time spent on various days. Significance level (2-tailed) = 0.0000. Variable are significant at 5% level. Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.174 The most explicit and substantial traffic is heading towards the Danube Bend, with Szentendre, Visegrád and Esztergom being the main destinations. Many non-Hungarian visitors come here for longer stays. The Budapest Metropolitan Region attracts high-scale visitor traffic, with Vecsés, as the air- port’s host municipality, standing out by far. The extreme visitor traffic flow evolves from peripheral locations, as the airport facility is supported by various towers both registered in Budapest and in the neighbouring municipali- ties around the airport. The visitors identified as taking part in satellite tourism in this local- Fig. 2. Cluster membership of municipalities regarding to international visitors on 23 October 2019. Source: Authors’ own research. Table 3. K-means clustering scoreboard, 23 October 2019, for visitors who spent more than 67 minutes in each municipality* Indicators Final cluster centres Rural areas Peri-urban areas Points of interest of satellite tourism Urban satellite destinationsnot affected by satellite tourism Distance from industrial areas, km Time spent, minutes Number of visitors 16.40 31.84 0.18 6.50 128.36 0.62 3.10 62,100.25 352.50 2.60 12,811.87 58.46 *The municipalities were clustered into four types: two rural types of municipalities not affected by visi- tor traffic (cluster 1: rural, not affected, and cluster 2: peri-urban areas, not affected), one highly visited, industrialised type (cluster 3: hot spots), and one urbanised type, attracting high-scale satellite traffic (cluster 4: urban satellite). 175Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. ity do not actually return to Budapest; their movements are apparent because they are occasionally served by Budapest towers, but their location remains unchanged in the vast majority of cases. This phenomenon of ‘appar- ent moving’ results in the peaking traffic of the agglomeration municipalities. The over-repre- sentation of agglomeration municipalities is an outcome of technological limitation (network load-optimised service delivery). Non-Hungarian visitors are heading to- wards the Western border of Hungary (Hegyeshalom, Mosonmagyaróvár) that in- corporates the daily visitor flows to Austria and Slovakia, and reflects the daily city tours to Vienna and Bratislava. The large number of long stays at the border area is also apparent, since in fact the signals disappear at the bor- der, but our method assumes a stop between the event traffic in the outbound and inbound directions. An indication of this is that we have assigned to border crossing a small number of visitors with very long stays. If data collection methods would enable tracking events beyond borders, the significance of the border-crossing settlements would immediately be reduced to a fraction, as visitors would appear elsewhere between two logins at the border area. The tourist attractions Eger and Lake Balaton are also key contributors to visitor traffic, with Balatonfüred as main destinations for non-Hungarian satellite tourism flow, with some significant seasonal variations. Discussion and conclusions Satellite tourism is a phenomenon to be dis- cussed under the umbrella of unconventional tourism mobility, which by nature is difficult to track because tourist consumption is not fulfilled, and the presence of tourists can be detected temporarily but is difficult to register (Belotti, S. 2019; Timothy, D. et al. 2022). The significant daytime tourist traffic in the ag- glomeration of large cities is rarely matched by the use of local accommodation services, so the tourism-based economy is one-sided, with moderate development in the accommo- dation sector and without the development of the night-time economy. Tourism receipts from leisure activities are mainly generated in agglomeration settlements, while the receipts from outward and return journeys to the big cities and overnight stays are mainly gener- ated in central cities. The city’s gravitational pull is due to its excellent international and domestic transport links, diversified accom- modation, hospitality and retail offerings, and colourful leisure activities. After a while, or in the case of repeat visitors (Freytag, T. 2010), travellers of big cities become saturated with local experiences, and begin looking for a new impulse, which they can experience either in an individual or organized way in the agglomeration zone. Based on our investigation of tourism mo- bility using mobile phone data, we have been able to define a model of satellite tourism (Figure 3). The novelty of the model is that it interprets the notion of satellite tourism not only in terms of flows to the settlements tra- ditionally surrounding a metropolis, but also integrates mobility outside the agglomera- tion and even beyond the national border for less than 24 hours. The model offers a new theoretical framework for understanding the tourist flows in towns involved in same- day visits because it describes a system of mobility between origin and destination set- tlements. As the studies on same-day visits have so far focused on destinations, mainly on the consumer behaviour of day trippers and its social, economic, and environmen- Fig. 3. Model of satellite tourism. Source: Authors’ own design. Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.176 tal impacts, the model can contribute to broadening the horizon of future investi- gations (Bauder, M. and Freytag, T. 2015; Bozonelos, D. 2020). Understanding the mo- tivations behind tourists returning within 24 hours will help to better outline tourism de- velopment in the areas concerned (Surinach, J. et al. 2017). Mobile phone data can be very helpful in better understanding the mobil- ity of satellite travellers, but the limitations of systematic observations using the same methodology mean that further procedural improvements are essential. Acknowledgements: The results presented in this study were supported by the OTKA project K134877 and the project NVKP 16-1-2016-0003. Supported by the ÚNKP-22-3-I New National Excellence Program of the Ministry for Culture and Innovation from the source of the National Research, Development and Innovation Fund. REFERENCES Ahas, R., Aasa, A., Roose, A., Mark, Ü. and Silm, S. 2008. Evaluating passive mobile positioning data for tourism surveys: An Estonian case study. Tourism Management 29. (3): 469–486. Available at https://doi.org/10.1016/j.tourman.2007.05.014 Ahas, R., Aasa, A., Silm, S. and Tiru, M. 2010. Daily rhythms of suburban commuters’ movements in the Tallinn metropolitan area: Case study with mobile positioning data. Transportation Research, Part C: Emerging Technologies 18. (1): 45–54. Available at https://doi.org/10.1016/j.trc.2009.04.011 Aldstadt, J. 2010. Spatial Clustering BT. In Handbook of Applied Spatial Analysis: Software Tools, Methods and Applications. Eds.: Fischer, M.M. and Getis, A. Berlin–Heidelberg, Springer, 279–300. Available at https://doi.org/10.1007/978-3-642-03647-7_15 Baggio, R. 2020. Tourism destinations: A universality conjecture based on network science. Annals of Tourism Research 82. (9): 102929. Available at https:// doi.org/10.1016/j.annals.2020.102929 Bauder, M. and Freytag, T. 2015. Visitor mobility in the city and the effects of travel preparation. Tourism Geographies 17. (5): 682–700. Bayir, M.A., Demirbas, M. and Eagle, N. 2009. Discovering spatio-temporal mobility profiles of cell phone users. IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks and Workshops 2009. MIT Open Access Articles, Institute of Electrical and Electronics Engineers. Available at https://doi.org/10.1109/ WOWMOM.2009.5282489 Belotti, S. 2019. “Sharing” tourism as an opportunity for territorial regeneration: The case of Iseo Lake, Italy. Hungarian Geographical Bulletin 68. (1): 79–91. Bose, I. and Chen, X. 2015. Detecting the migration of mobile service customers using fuzzy cluster- ing. Information and Management 52. (2): 227–238. Available at https://doi.org/10.1016/j.im.2014.11.001 Bozonelos, D. 2020. Day tripping in Jerusalem: The curious case of how Russian Orthodox Christians became the same-day religious tourists in the Holy Land. International Journal of Religious Tourism and Pilgrimage 8. (6): 12–23. Burke, J. 1986. Satellite cities of the future: Free activity zones as precursors of urban development. Habitat International 10. (1–2): 291–297. Candia, J., González, M.C., Wang, P., Schoenharl, T., Madey, G. and Barabási, A.L. 2008. Uncovering in- dividual and collective human dynamics from mo- bile phone records. Journal of Physics A: Mathematical and Theoretical 41. (22): 224015. Available at https:// doi.org/10.1088/1751-8113/41/22/224015 Chu, C.P. and Chou, Y.H. 2021. Using cellular data to analyse the tourists’ trajectories for tourism destination attributes: A case study in Hualien, Taiwan. Journal of Transport Geography 96. (10): 103178. Available at https://doi.org/10.1016/j.jtran- geo.2021.103178 Dövényi, Z. and Kovács, Z. 2006. Budapest: Post- socialist metropolitan periphery between ‘catching up’ and individual development path. European Spatial Research and Policy 13. 23–41. Egedy, T., Kovács, Z. and Kondor, A. 2017. Metropolitan region building and territorial devel- opment in Budapest: The role of national policies. International Planning Studies 22. (1): 14–29. Egedy, T. and Ságvári, B. 2021. Urban geographi- cal patterns of the relationship between mobile communication, social networks and economic development – the case of Hungary. Hungarian Geographical Bulletin 70. (2): 129–148. Fabula, Sz., Skovgaard Nielsen, R., Barberis, E., Boros, L., Hedegaard Winther, A. and Kovács, Z. 2021. Diversity and local business structure in European urban contexts. Hungarian Geographical Bulletin 70. (1): 65–80. Freytag, T. 2010. Déjà-vu: Tourist practices of repeat visitors in the city of Paris. Social Geography 5. (1): 49–58. Girardin, F., Fiore, F.D., Blat, J., Ratti, C. and Dal Fiore, F. 2008. Understanding of tourist dynamics from explicitly disclosed location information. Journal of Location Based Services 2. (1): 41–56. González, M.C., Hidalgo, C.A. and Barabási, A.L. 2008. Understanding individual human mobility patterns. Nature 453. (7196): 779–782. Available at https://doi.org/10.1038/nature06958 Hannam, K., Butler, G. and Paris, C.M. 2014. Deve- lopments and key issues in tourism mobilities. https://doi.org/10.1016/j.tourman.2007.05.014 https://doi.org/10.1016/j.trc.2009.04.011 https://doi.org/10.1007/978-3-642-03647-7_15 https://doi.org/10.1016/j.annals.2020.102929 https://doi.org/10.1016/j.annals.2020.102929 https://doi.org/10.1109/WOWMOM.2009.5282489 https://doi.org/10.1109/WOWMOM.2009.5282489 https://doi.org/10.1016/j.im.2014.11.001 https://doi.org/10.1088/1751-8113/41/22/224015 https://doi.org/10.1088/1751-8113/41/22/224015 https://doi.org/10.1016/j.jtrangeo.2021.103178 https://doi.org/10.1016/j.jtrangeo.2021.103178 https://doi.org/10.1038/nature06958 177Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178. Annals of Tourism Research 44. (1): 171–185. Available at https://doi.org/10.1016/j.annals.2013.09.010 Hua, H. and Wondirad, A. 2021. Tourism network in urban agglomerated destinations: Implications for sustainable tourism destination development through a critical literature review. Sustainability 13. (1): 285. Available at https://doi.org/10.3390/ su13010285 Huang, B. and Wang, J. 2020. Big spatial data for urban and environmental sustainability. Geo-Spatial Information Science 23. (2): 125–140. Available at https://doi.org/10.1080/10095020.2020.1754138 Irimiás, A. 2010. Budapest’s thermal spas on screen. In Health, Wellness and Tourism: Healthy Tourists, Healthy Business? Proceedings of the Travel and Tourism Research Association Europe 2010. Ed.: Puczkó, L., Budapest, Travel and Tourism Research Association, 93–101. Jiang, S., Ferreira, J. and Gonzalez, M.C. 2016. Activity-based human mobility patterns in- ferred from mobile phone data: A case study of Singapore. IEEE Transactions on Big Data 3. (2): 208–219. Available at https://doi.org/10.1109/tbda- ta.2016.2631141 Kádár, B. and Gede, M. 2013. Where do tourists go? Visualizing and analysing the spatial distribution of geotagged photography. Cartographica: The International Journal for Geographic Information and Geovisualization 48. (2): 78–88. Kádár, B. and Gede, M. 2021. Tourism flows in large-scale destination systems. Annals of Tourism Research 87. 103113. Kádár, B. and Gede, M. 2022. The measurable pre- dominance of weekend trips in established tourism regions. The case of visitors from Budapest at waterside destinations. Sustainability 14. (6): 3293. Kincses, Á., Tóth, G., Tömöri, M. and Michalkó, G. 2017. Characteristics of transit tourism in Hungary with a focus on expenditure. Regional Statistics 6. (2): 129–148. Kovalcsik, T., Elekes, Á., Boros, L., Könnyid, L. and Kovács, Z. 2022. Capturing unobserved tourists: Challenges and opportunities of processing mobile positioning data in tourism research. Sustainability 14. (21): 13826. Lingras, P., Bhalchandra, P., Khamitkar, S., Mekewad, S. and Rathod, R. 2011. Crisp and soft clustering of mobile calls. In Multi-Disciplinary Trends in Artificial Intelligence. Eds.: Sombattheera, C., Agarwal, A., Udgata, S.K. and Lavangnananda, K. Berlin–Heidelberg, Springer, 147–158. Available at https://doi.org/10.1007/978-3-642-25725-4_13 Liu, Z., Wang, A., Weber, K., Chan, E.H.W. and Shi, W. 2022. Categorisation of cultural tourism attrac- tions by tourist preference using location-based social network data: The case of Central Hong Kong. Tourism Management 90. Available at https:// doi.org/10.1016/j.tourman.2022.104488 Louail, T., Lenormand, M., Cantu Ros, O.G., Picornell, M., Herranz, R., Frias-Martinez, E., Ramasco, J.J. and Barthelemy, M. 2014. From mobile phone data to the spatial structure of cities. Scientific Reports 4. 5276. Available at https://doi. org/10.1038/srep05276 Luštický, M. and Štumpf, P. 2021. Leverage points of tourism destination competitiveness dynam- ics. Tourism Management Perspectives 38. (1): 100792. Available at https://doi.org/10.1016/j. tmp.2021.100792 Mata, F.J., Grec, F.C., Azaola, M., Blázquez, F., Fernández, A., Dominguez, E., Cueto-Felgueroso, G., Seco-Granados, G., del Peral-Rosado, J.A., Staudinger, E., Gentner, C., Kasparek, M., Backert, C., Barlett, D., Serna, E., Ries, L. and Prieto-Cerdeira, R. 2020. Preliminary field trials and simulations results on performance of hy- brid positioning based on GNSS and 5G signals. Proceedings of the 33rd International Technical Meeting of the Satellite Division of the Institute of Navigation. ION GNSS 2020, 387–401. Available at https://doi. org/10.33012/2020.17609 Merrilees, B., Miller, D. and Herington, C. 2013. City branding: A facilitating framework for stressed sat- ellite cities. Journal of Business Research 66. (1): 37–44. Murillo, J., Vayà, E., Romaní, J. and Suriñach, J. 2013. How important to a city are tourists and day-trip- pers? The economic impact of tourism on the city of Barcelona. Tourism Economics 19. (4): 897–917. Available at https://doi.org/10.5367/te.2013.0225 Pinke-Sziva, I., Smith, M., Olt, G. and Berezvai, Z. 2019. Overtourism and the night-time economy: A case study of Budapest. International Journal of Tourism Cities 5. (1): 1–16. Pucci, P. 2015. Daily mobility practices through mobile phone data: An application in Lombardy region. Springer Briefs in Applied Sciences and Technology 31. 27–70. Available at https://doi. org/10.1007/978-3-319-14833-5_3 Qian, C., Li, W., Duan, Z., Yang, D. and Ran, B. 2021. Using mobile phone data to determine spatial correlations between tourism facilities. Journal of Transport Geography 92. 103018. Available at https:// doi.org/10.1016/j.jtrangeo.2021.103018 Raun, J., Ahas, R. and Tiru, M. 2016. Measuring tourism destinations using mobile tracking data. Tourism Management 57. 202–212. Available at https://doi.org/10.1016/j.tourman.2016.06.006 Razavi, S.M., Gunnarsson, F., Rydén, H., Busin, Å., Lin, X., Zhang, X., Dwivedi, S., Siomina, I. and Shreevastav, R. 2018. Positioning in cel- lular networks: Past, present, future. In IEEE Wireless Communications and Networking Conference. Barcelona, IEEE, 1–6. Available at https://doi. org/10.1109/WCNC.2018.8377447 Reif, J. and Schmücker, D. 2020. Exploring new ways of visitor tracking using big data sources: https://doi.org/10.1016/j.annals.2013.09.010 https://doi.org/10.3390/su13010285 https://doi.org/10.3390/su13010285 https://doi.org/10.1080/10095020.2020.1754138 https://doi.org/10.1109/tbdata.2016.2631141 https://doi.org/10.1109/tbdata.2016.2631141 https://doi.org/10.1007/978-3-642-25725-4_13 https://doi.org/10.1016/j.tourman.2022.104488 https://doi.org/10.1016/j.tourman.2022.104488 https://doi.org/10.1038/srep05276 https://doi.org/10.1038/srep05276 https://doi.org/10.1016/j.tmp.2021.100792 https://doi.org/10.1016/j.tmp.2021.100792 https://doi.org/10.33012/2020.17609 https://doi.org/10.33012/2020.17609 https://doi.org/10.5367/te.2013.0225 https://doi.org/10.1007/978-3-319-14833-5_3 https://doi.org/10.1007/978-3-319-14833-5_3 https://doi.org/10.1016/j.jtrangeo.2021.103018 https://doi.org/10.1016/j.jtrangeo.2021.103018 https://doi.org/10.1016/j.tourman.2016.06.006 https://doi.org/10.1109/WCNC.2018.8377447 https://doi.org/10.1109/WCNC.2018.8377447 Michalkó, G. et al. Hungarian Geographical Bulletin 72 (2023) (2) 163–178.178 Opportunities and limits of passive mobile data for tourism. Journal of Destination Marketing and Management 18. 100481. Available at https://doi. org/10.1016/j.jdmm.2020.100481 Sagl, G., Delmelle, E. and Delmelle, E. 2014. Mapping collective human activity in an urban en- vironment based on mobile phone data. Cartography and Geographic Information Science 41. (3): 272–285. Available at https://doi.org/10.1080/15230406.201 4.888958 Sagl, G. and Resch, B. 2015. Mobile phones as ubiq- uitous social and environmental geo-sensors. In Encyclopedia of Mobile Phone Behavior. Ed.: Yan, Z., Hershey, PA, IGI Global, 1194–1213. Available at https://doi.org/10.4018/978-1-4666-8239-9.ch098 Saluveer, E., Raun, J., Tiru, M., Altin, L., Kroon, J., Snitsarenko, T., Aasa, A. and Silm, S. 2020. Methodological framework for producing national tourism statistics from mobile positioning data. Annals of Tourism Research 81. 102895. Available at https://doi.org/10.1016/J.ANNALS.2020.102895 Steenbruggen, J., Borzacchiello, M.T., Nijkamp, P. and Scholten, H. 2013. Mobile phone data from GSM networks for traffic parameter and urban spa- tial pattern assessment: A review of applications and opportunities. GeoJournal 78. (2): 223–243. Available at https://doi.org/10.1007/s10708-011-9413-y Stetic, S., Simicevic, D. and Stanic, S. 2011. Same-day trips: A chance of urban destination development. UTMS Journal of Economics 2. (2): 113–124. Surinach, J., Casanovas, J.A., Andre, M., Murillo, J. and Romaní, J. 2017. How to quantify and character- ize day trippers at the local level: An application to the comarca of the Alt Penedès. Tourism Economics 23. (2): 360–386. Szabó, T., Szabó, B. and Kovács, Z. 2014. Polycentric urban development in post-socialist context: The case of the Budapest Metropolitan Region. Hungarian Geographical Bulletin 63. (3): 287–301. Available at https://doi.org/10.15201/hungeob- ull.63.3.4 Thakuriah, P. (Vonu), Sila-Nowicka, K., Hong, J., Boididou, C., Osborne, M., Lido, C. and McHugh, A. 2020. Integrated Multimedia City Data (iMCD): A composite survey and sensing approach to understanding urban living and mo- bility. Computers, Environment and Urban Systems 80. 101427. Available at https://doi.org/10.1016/j. compenvurbsys.2019.101427 Timothy, D., Michalkó, G. and Irimiás, A. 2022. Unconventional tourist mobility: A geography-ori- ented theoretical framework. Sustainability 14. (11): 6494. Available at https://doi.org/10.3390/ su14116494 Tóth, G. and Kincses, Á. 2022. (In)visible tourism according to online cash registers in Hungary, 2018–2020. Sustainability 14. (5): 3038. Vanhoof, M., Reis, F., Ploetz, T. and Smoreda, Z. 2018. Assessing the quality of home detection from mobile phone data for official statistics. Journal of Official Statistics 34. (4): 935–960. Available at https:// doi.org/10.2478/jos-2018-0046 Wang, W., Luan, Z., He, B., Li, X., Zhang, D., Huang, Z. and Tu, W. 2018. A new hierarchical clustering approach for sparse mobile phone trajectories. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences – ISPRS Archives 42. (4): 763–770. Available at https://doi. org/10.5194/isprs-archives-XLII-4-697-2018 Wang, Z., He, S.Y. and Leung, Y. 2018. Applying mobile phone data to travel behaviour research: A literature review. Travel Behaviour and Society 11. 141–155. Available at https://doi.org/10.1016/j. tbs.2017.02.005 Wynen, J. 2013a. Explaining travel distance during same-day visits. Tourism Management 36. (June): 133–140. Available at https://doi.org/10.1016/j. tourman.2012.11.007 Wynen, J. 2013b. An estimation of the determinants of same-day visit expenditures in Belgium. Tourism Economics 19 (1): 161–172. Available at https://doi. org/10.5367/te.2013.0190 Yang, Y. 2012. Agglomeration density and tourism development in China: An empirical research based on dynamic panel data model. Tourism Management 33. 1347–1359. Yuan, Y. and Raubal, M. 2012. A framework for spa- tio-temporal clustering from mobile phone data. In Workshop on Complex Data Mining in a GeoSpatial Context at AGILE 2012. Proceedings. Eds.: de Runz, C., Devogele, T. and Perret, J., Avignon, France, AGILE Publication, 22–26. Zhang, Y., Li, Q., Wang, H., Du, X. and Huang, H. 2019. Community scale liveability evaluation inte- grating remote sensing, surface observation and geospatial big data. International Journal of Applied Earth Observation and Geoinformation 80. 173–186. Available at https://doi.org/10.1016/j.jag.2019.04.018 https://doi.org/10.1016/j.jdmm.2020.100481 https://doi.org/10.1016/j.jdmm.2020.100481 https://doi.org/10.1080/15230406.2014.888958 https://doi.org/10.1080/15230406.2014.888958 https://doi.org/10.4018/978-1-4666-8239-9.ch098 https://doi.org/10.1016/J.ANNALS.2020.102895 https://doi.org/10.1007/s10708-011-9413-y https://doi.org/10.15201/hungeobull.63.3.4 https://doi.org/10.15201/hungeobull.63.3.4 https://doi.org/10.1016/j.compenvurbsys.2019.101427 https://doi.org/10.1016/j.compenvurbsys.2019.101427 https://doi.org/10.3390/su14116494 https://doi.org/10.3390/su14116494 https://doi.org/10.2478/jos-2018-0046 https://doi.org/10.2478/jos-2018-0046 https://doi.org/10.5194/isprs-archives-XLII-4-697-2018 https://doi.org/10.5194/isprs-archives-XLII-4-697-2018 https://doi.org/10.1016/j.tbs.2017.02.005 https://doi.org/10.1016/j.tbs.2017.02.005 https://doi.org/10.1016/j.tourman.2012.11.007 https://doi.org/10.1016/j.tourman.2012.11.007 https://doi.org/10.5367/te.2013.0190 https://doi.org/10.5367/te.2013.0190 https://doi.org/10.1016/j.jag.2019.04.018