The Impact of COVID-19 on Flight Networks
Toyotaro Suzumura, Hiroki Kanezashi, Mishal Dholakia, Euma Ishii, Sergio Alvarez Napagao, Raquel Pérez-Arnal, Dario Garcia-Gasulla, Toshiaki Murofushi
TThe Impact of COVID-19 on Flight Networks *Toyotaro Suzumura ∗ , *Hiroki Kanezashi † , Mishal Dholakia ‡ , Euma Ishii § ,Sergio Alvarez Napagao ∗ , Raquel Prez-Arnal ∗ and Dario Garcia-Gasulla ∗∗ Barcelona Supercomputing [email protected], [email protected], [email protected], [email protected] † Tokyo Institute of [email protected] ‡ [email protected] § Massachusetts Institute of [email protected] ¶ Tokyo Medical and Dental [email protected]
Abstract —As COVID-19 transmissions spread worldwide, gov-ernments have announced and enforced travel restrictions toprevent further infections. Such restrictions have a direct effecton the volume of international flights among these countries,resulting in extensive social and economic costs. To better under-stand the situation in a quantitative manner, we used the Openskynetwork data to clarify flight patterns and flight densities aroundthe world and observe relationships between flight numbers withnew infections, and with the economy (unemployment rate) inBarcelona. We found that the number of daily flights graduallydecreased and suddenly dropped 64% during the second halfof March in 2020 after the US and Europe enacted travelrestrictions. We also observed a 51% decrease in the global flightnetwork density decreased during this period. Regarding newCOVID-19 cases, the world had an unexpected surge regardlessof travel restrictions. Finally, the layoffs for temporary workersin the tourism and airplane business increased by 4.3 fold in theweeks following Spain’s decision to close its borders.
I. I
NTRODUCTION
A. Background
As COVID-19 transmissions spread worldwide, govern-ments announced and enforced both domestic and internationaltravel restrictions. These restrictions affected the trend offlight networks around the world such as major airlines [1]and tourism-dependent cities [2]. All or a portion of flightsconnecting these restricted countries and areas were cancelled.The date of travel restriction enforcement varies by country,therefore the period and degree of flight reductions also varyby country and continents.
B. Overview
We quantitatively investigated how flight restrictions affectdomestic and international travel in countries and continentsthrough time-series analytics and graph algorithms. We alsoinvestigated the negative effect of these restrictions on employ-ment and the number of infected cases in Barcelona, wherethe majority of revenue depends on the travel industry. *Toyotaro Suzumura and Hiroki Kanezashi contributed equally to this work.
The structure of this paper is as follows; In Section II, weintroduce the flight dataset which we used for our analysis.We compared the dataset with other flight datasets and dis-cussed the limitations and potential biases. In Section III,we summarize the statistical analytics of airport and flightdata by country and continent. In Section IV, we analyzethe number of daily flights globally as well as by region(continent and country level), and discuss the relationshipbetween government announcements and the decline of flights.In Section V, we visualize flight networks and apply graphanalytic techniques to the network to evaluate the quantitativeimpacts on travelers. We examine economic effects on travelrestrictions and cities depending on tourism industries fromthe public data from Barcelona in Section VI, and we evaluatethe relationship between incoming flights from Europe and thenumber of new infection cases in the United States in SectionVII. Finally, we discuss our findings, limitations and relatedwork in Section VIII, and explain the summary and futureperspectives of our research in Section IX.II. M
ETHODS
A. Methods and Algorithms
In order to evaluate the effect of travel restrictions on theflights, we conducted the following analyses step by step.First, we summarized the total number of airports, flightsand potential passengers for each country and continent froman open dataset. Next, we counted up the number of dailyflights for countries and continents, and compared the periodof the flight reductions and the date of the travel restrictionenforcement. In order to evaluate the density of the flight net-work quantitatively, we also applied graph analytics methodsto the time-series flight data. Finally, we analyzed time-seriesdata sets about the number of incoming flights, number ofinfected cases and unemployment in Barcelona [3], wheretourism industries is a main revenue source, and hypothesizedthe correlation among these data. a r X i v : . [ c s . S I] J un . The OpenSky Network Dataset For time-series flight analytics, we used an open dataset [4]from The OpenSky Network [5]. The dataset contains flightrecords with departure and arrival times, airport codes (originand destination), and aircraft types. The dataset includes thefollowing flight information during 2020 January 1 to April 30.The dataset for a particular month is made available during thebeginning of the following month. Note that the year of all ofthe following dates is 2020. In total, our dataset is composedby: • • •
616 major airports covering 148 countries (out of 195world countries)III. S
TATISTICAL A NALYSIS OF F LIGHT D ATA
A. Distribution of the Number of Airports
We extracted flight data regarding 616 primary airports from148 countries. Figure 1 describes the number of major airportsper continent and country. Currently, approximately 33% ofthe world’s airports belong in North America and 170 airports(27.6%) in the United States. 27% (166) and 26% (162) belongto Asia and Europe respectively, and China (5.7%) and theUnited Kingdom (4.4%) have the highest number of airportswithin each continent.
Fig. 1. Number of Major Airports per Continent and Country
B. Number of Flights per Area and Country
Figure 2 describes the total number of flights from January1 to April 30 2020 for each continent and country. More thanhalf of flights in this dataset departed from Europe. 10% ofinternational flights departed from Germany (10%) and theUnited States (10.7%). In Asia, Hong Kong, the United ArabEmirates, and Singapore had the highest number of flights(2.9%, 2.6%, and 2.2% respectively), yet there were veryfew flights connecting to China. The number of flights inSouth America, Africa, and Oceania were too small to analyze, therefore these continents were excluded from the followingsections.
Fig. 2. Number of Departed Flights per Continent and Country
C. Number of Potential Passengers
The OpenSky Network Dataset does not contain the finalnumber of passengers for each flight. However, since the datacontained each aircraft type used, we used this to estimatethe number of passengers for each trip. Since the informationregarding the flights were extracted from the capacity data ofeach aircraft type from their respective Wikipedia pages [6],we aim to obtain more accurate information on aircraft typesand their capabilities for our future analysis.The proportion of international passengers travelling fromAsia was approximately 31.2% of all passengers, and thenumber of flights were 23.5% of overall flights. On theother hand, the total number of passengers of internationalflights within Europe is about 46.5% while the proportion ofinternational flights outside of Europe is 55.5%. The gap ofproportions between passengers and flights comes from thedifference in capacities of the aircraft.IV. D
ESCRIPTIVE A NALYSIS OF T IME - SERIES F LIGHT D ATA
A. Overview
Announcements and enforcement regarding lockdowns (bor-der closures and travel restrictions, etc.) were implemented toslow the rate of transmissions and prevent overburdening ofhealth facilities. The date of implementation, length, and extentvary by country. In the US, the Secretary of Health and HumanServices declared a public health emergency on January 31[7]. The US government suspended incoming flights fromEurope (Schengen Area) on March 14 [8], and closed theborder between the US and Canada on March 18 [9]. Germanyclosed borders with neighboring countries on March 16 [10].Countries in the European Union agreed to reinforce borderclosures and restrict non-essential travels from March 17 [11]nwards. In this section, we show the changes in the numberof daily flights in each region, and discuss the correlationsbetween travel restrictions and the number of flights in eachcountry.
B. Overall International Flights
Figure 3 describes the number of daily international flightsin the world. Before the end of February, the daily number ofinternational flights was approximately 8,000. However, thenumber of daily flights gradually decreased to 6,000 by earlyMarch and then drastically dropped to less than 1,000 (around10% of regular seasons) by the end of March.
Fig. 3. Total Number of Daily Departed International Flights
C. Inter-continental Flights
Figure 4 describes the number of intercontinental flightsthat departed from each continent. Inter-continental flightsfrom Europe, Asia, and North America drastically decreasedfrom March 13 as many flights were canceled. The decreaseprobably reflects the declaration of pandemic by WHO onMarch 11 and national emergency in the United States onMarch 13.
Fig. 4. Total Number of Daily Departed Inter-continental Flights per Conti-nent
D. Continental Flights
Figure 5 describes the number of continental flights withineach continent. In Europe, more than 3,000 flights wereoperating per day until the beginning of March, but droppedto 10% of the regular number of flights from mid Marchto the end of March. In Asia, the number of flights hasgradually decreased from February. In North America, thenumber of flights was steady until mid-March, before droppingsignificantly.
Fig. 5. Number of Daily International Flights per Continent
In Europe, the UK, Ireland, and Germany had the highestnumber of flight passengers [12]. In Germany, internationalflights decreased from the beginning of March and thendropped after the nation closed its borders on March 16 (Figure6). This reflected the EU’s agreement to ban incoming travelfrom areas outside the EU, UK, and Switzerland on March17.Most of the continental flights had been operating in NorthAmerica (the US in particular) until mid-March and halfnumber of these flights in the Beginning of January are stillin operation today (Figure 7). On the other hand, the numberof flights in Asia, Europe, and Oceania have dropped to lessthan 20% to that of January 1.
E. Domestic and International Flights in the United States
In the US, the number of domestic flights suddenly droppedfrom mid-March. However, roughly 50% of all regular flightswere still in operation on March 31 (Figure 8), although theUS government and airlines planned to ban or reduce flightsby that point [13]. In April, the number of flights graduallydeclined but 30% of regular flights were still in operation.The number of flights from European countries to theUS suddenly dropped after March 14 (Figure 9). The USsuspended incoming travel from Europe (Schengen area) fromMarch 13. In the overall number of international flights, theprimary destination from the US is Canada (green lines inFigure 10). The number of daily flights between the US andCanada gradually decreased after these governments agreed toclose their border on March 18. ig. 6. Number of Daily Flights from the United Kingdom and Germany toOther European CountriesFig. 7. Number of Daily International Flights from the United States in NorthAmerica
F. Key Takeaways
We analyzed the number of daily flights for each countryand continent, and compared the relationship between the dateof travel restrictions and the decline of flights. We foundthat many international flights suddenly declined during themiddle of March. In Europe, the number of flights dropped toabout 10% of regular seasons after the EU agreed to restrictincoming travels on March 17. In the US, approximately halfof all domestic flights were still in operation even at the endof March.
Fig. 8. Number of Domestic Flights in the United StatesFig. 9. Number of Flights Between the United States and European Countries
V. G
RAPH A NALYTICS ON THE F LIGHT N ETWORK
A. Overview
With travel restrictions, the global flight network graduallybecame more sparse, which prevented travelers from:1) Using direct flights to their destinations2) Returning to their origin once they depart from theairport3) Departing from their current location due to absence offlights in nearby airportsBy constructing daily flight networks, we quantitativelyevaluated these effects through graph analytics. Each vertexand edge represent an airport and a flight, respectively. ig. 10. Number of Flights Between the United States and top-10 Countries
1) Distribution of the number of neighboring airports andflights (degree and neighborhood distributions)2) Strongly connected component (SCC) extraction3) Isolated vertex extractionIn the construction of flight networks, we aggregated mul-tiple flights with the same origin and destination and date intoa single weighted edge in the flight network construction likeFigure 11. The edge weight is the total number of flights.
Fig. 11. Flight Edge Aggregations by Date
While multiple flights are in operation every day, not allflights are operated every day (e.g., three days a week). Todetect long-term trends more precisely, we also aggregatedflight edges by week as an optional preprocessing (Figure 12).
Fig. 12. Flight Edge Aggregations by Week
B. Connecting Airports and Flights
In regular times, many flights (edges in the flight network)connect from the airport (vertices in the flight network) tothe airport. That helps travelers to get to more destinationswith fewer transfers. However, travel restrictions prevent themfrom using direct flights and getting to their destinations. Tomeasure the effect quantitatively, we computed the followingmetrics for each airport and each date (Figure 13).1) Number of flights (edges) from the airport2) Number of 1-hop neighboring airports from the airport(the number of reachable destinations with a nonstopflight)3) Number of 1,2-hop neighboring airports from the airport(the number of reachable destinations with a singleconnecting flight)
Fig. 13. Number of Flights and Neighboring Airports from the Airport A The color and size of each vertex in Figure 14 - 19 indicatethe number of departed flights from the airport. From mid-March to the beginning of April, many flights were graduallycanceled, except for domestic flights in the United States andinternational flights between the United States and Europe.The number of flights in Europe suddenly decreased fromMarch 16 (the day before EU countries closed borders) toMarch 23. In April, only a few flights were operating andmany airports became isolated.In North America, most of the flights in North Americawere operating between the United States and Canada. In theUnited States, most of the domestic flights were operating asusual, even at the end of April.
C. Density and Clustering Coefficient
With denser flight networks, travelers can reach more des-tinations. As the barometer of the flight network density, wecomputed the following metrics.
Density:
The ratio of the actual number of edges after aggre-gation against the number of edges of the complete graph.
Clustering Coefficient:
The ratio of the total number oftriangles against the total number of triplets (connected threevertices). This is mainly used for social network analytics tomeasure the connectivity of friendships.The density (red lines in Figure 20 - 22) of the global flightnetwork rapidly decreased by half during March 8 to April 1.In Europe, drastic decreases in the flight network density wassimilar to that of the global network, and fell to approximately17% of the peak. In the US on the other hand, the densitygradually decreased, yet only by 20%. From March 13, the ig. 14. Global Flight Network on January 1Fig. 15. Global Flight Network on April 1 clustering coefficient (blue lines) of the global flight networkgradually decreased by half in the next two weeks. In Europe,it rapidly declined to about a quarter in the same period. Inthe US, the decline of the clustering coefficient is only about10% at the end of April.
D. Strongly Connected Component and Isolated Airports
In regular seasons, travelers departing from any airport cantypically go to any destination and return to their airportof origin. This is due to flight networks being a stronglyconnected component (SCC), within which travelers can gofrom/to any airports in the same SCC. With travel restrictions,however, many flights were canceled and the large SCC (bluecircle in Figure 23) has decomposed into smaller SCCs (yellowand green circles in Figure 23). In the result, travelers cannot:1) Go to other airports outside local flight networks2) Return to their origin (a red edge from E to D)3) Depart from isolated airports (a red vertex)In the global flight network, the size of the largest SCC(red line) decreased by 30% from March 13 (300) to April 6(210), and the number of isolated airports (blue line) increasedby 30% (from 310 to 400). In Europe, the largest SCC sizeon April 7 was less than half of March 10 (before border
Fig. 16. Flight Network in Europe on January 1Fig. 17. Flight Network in Europe on April 1 closures), and the number of isolated airports tripled. In the USon the other hand, the largest SCC size and isolated airports arealmost stable (80 and 100 respectively). Although the largestSCC size gradually decreased from March 13 to April 6, therate of the overall decrease is only about 10%.
E. Diameter and Radius of Flight Network
When many flights are canceled, more transfers are requiredfor travellers to reach the destinations compared to regularseasons. In order to measure the number of required transfers,we applied a graph analysis to compute eccentricity (the max-imum distance from a specified vertex to all other connectedvertices) of the flight network. An eccentricity of an airportvertex corresponds to the number of required transfers to otherdestinations in the worst case. With travel restrictions, thediameter (the maximum value of eccentricities) and the radius(the minimum value of eccentricities) of the flight networkbecome larger.The diameter (red line in Figure 28 - 30), radius (blue line),and average eccentricity (green line) of the largest SCC size(black line) in the global flight network increased from April.In the flight network in Europe, diameter and radius rapidlyincreased from March 16 in two weeks. The overall flightnetworks trend in the US did not change until the end ofApril.
F. Flight Edge Weight and Density
Although a number of flights still exist among most ofthe airports, the flight frequency has gradually decreased.To measure the decline of flight frequencies, we definedthe relative weight for each flight edge (pair of vertices asconnected airports) and network density with the weightededges. ig. 18. Flight Network in North America on January 1Fig. 19. Flight Network in North America on April 1Fig. 20. Density (red line) and Clustering Coefficient (blue line) of the GlobalFlight Network
First, we computed the moving average of the number ofdaily flights for each pair of vertices, and then computed theratio of the total number of flights in a week against thebaseline of the specified week (January 12 - January 18),before defining the ratio as the relative weight of the edge.Besides the topological density described in V-C, we computedthe density of the whole flight network with the weighted
Fig. 21. Density (red line) and Clustering Coefficient (blue line) of the FlightNetwork in EuropeFig. 22. Density (red line) and Clustering Coefficient (blue line) of the FlightNetwork in the United States edges.The blue and red lines in Figure 32 - 34 represent the densityof the daily flight network with the unweighted/weighted flightedges. The density of the whole flight network with weightededges in Europe dropped to almost zero in April.In the US, the density with weighted edges has droppedfrom 0.025 to 0.005 in March. Although the topology of theflight network remains unchanged, the frequency of domesticflights in the US significantly decreased as seen in other
Fig. 23. An Example of SCC Decompositions of Flight Networksig. 24. Largest SCC Size (red line) and the Number of Isolated Airports(blue line) of the Global Flight NetworkFig. 25. Largest SCC Size (red line) and the Number of Isolated Airports(blue line) of the Flight Network in Europe continents.
G. Key Takeaways
We applied some graph analytics to the flight data in orderto evaluate the density of flight networks. Many graph metricssuch as density, maximum SCC size, and diameter remarkably
Fig. 26. Largest SCC Size (red line) and the Number of Isolated Airports(blue line) of the Flight Network in the United States Fig. 27. An Example of the Average Eccentricity, Diameter and RadiusFig. 28. Diameter (red line) and Radius (blue line) of the Global FlightNetwork indicated that the flight networks in the overall world andEurope became more sparse in March. We also defined the”weighted” flight network density considering the frequencyof flights, and showed the flight network in the United Statesalso became more sparse from the mid of March to the midof April.VI. I
MPACT ON THE T RAVEL I NDUSTRY IN B ARCELONA
Travel restrictions have a large impact on travel industriesand tourism-dependent countries. In this section, we investi-gate the correlation and causal association between the declineof incoming flights, the number of infection cases, and theunemployment in Barcelona.
Fig. 29. Diameter (red line) and Radius (blue line) of the Europe FlightNetworkig. 30. Diameter (red line) and Radius (blue line) of the United States FlightNetwork Fig. 31. An Example of Weighted Flight Edges
A. Incoming Flight Reductions in Barcelona
According to the unemployment dataset [14], the number ofaffected temporary workers (red line in Figure 35) graduallydeclined in April after it suddenly increased on Match 24. Thenumber of flights (blue line) arriving in Barcelona graduallydecreased until the beginning of April. These numbers seemto have some correlations, but more investigation is requiredto understand the causality and relationship between thesenumbers.
B. Unemployment and New Infection Cases in Barcelona
The number of flights to Barcelona (blue line in Figure36) has been decreasing since the beginning of March, andsuddenly halved on March 16 just before Spain closed itsborders [15]. Then, the number of affected temporary workers
Fig. 32. Density of Global Flight Network with Unweighted (Blue) andWeighted (Red) Flight Edges Fig. 33. Density of Flight Network in Europe with Unweighted (Blue) andWeighted (Red) Flight EdgesFig. 34. Density of Flight Network in the United States with Unweighted(Blue) and Weighted (Red) Flight Edges (red line) suddenly increased by 4.3 times (from 15,901 onMarch 23 to 68,170 on March 25). It is notable that the datasetregarding the number of affected workers became availablefrom March 23. The number of new infection cases in Spain(orange line) gradually increased from March 12 and peakedat almost 10,000 on March 25.
C. Flights to Barcelona and New Infection Cases in the MajorOrigin Countries
Approximately half of all flights to Barcelona (BCN) comefrom France, Germany, and the UK. The government of
Fig. 35. Changes of the Number of Affected Temporary Workers with theNumber of Incoming Flights and Estimated Passengers in Barcelonaig. 36. Changes of the Number of Affected Temporary Workers (red line)in Barcelona and the Number of New Infection Cases (orange line) in Spain
Spain closed its borders on March 16, and the number offlights from these three countries decreased by half during thefollowing day. However, the number of new infection casesfor these countries has been gradually increasing even beforeborder closures. Considering the latent period of COVID-19is commonly five to six days [16], the reason the trend ofincrease of new infection cases continued for a week afterthe border closure may be due to the immigration of infectedindividuals before the border closure.
Fig. 37. Number of Incoming Flights and New Infection Cases of the Top-3Origin Countries
D. Key Takeaways
We compared the number of incoming flights, affectedtemporary workers, and the number of new infection casesin Barcelona to estimate how the travel restriction affectedthe tourism industry. We found the number of flights halvedafter Spain closed borders on March 16, and the number ofaffected workers jumped just after the government opened theapplication of unemployment on March 23. The number ofnew infection cases increased until March 25 after the borderclosures, potentially due to infections spreading during itslatent period. VII. R
ELATIONSHIP B ETWEEN I NCOMING F LIGHTS AND I NFECTION C ASES
A sequence analysis of COVID-19 virus found out that thevirus type of the most infection cases in New York City isfrom Europe [17]. The first case was confirmed in the State ofNew York on February 29, and the number of infection casesgradually increased from March 15 (Figure 38). To investigatethe effects of incoming flights from Europe, we compared thenumber of daily flights from Europe and new infection casesin the US.
Fig. 38. Number of New Infection Cases in the United States
A. Number of Infection Cases in European Countries and theUnited States
The trend of increase in newly infected cases in Europe be-gan a week before the US (Figure 39). In particular, confirmednew infections in Italy started to increase in the beginning ofMarch. Spain and Germany also confirmed cases before the US(Figure 40). The US announced to enforce travel restrictionsfrom Europe on March 13 (EU mainland) and March 14(the UK and Ireland). After these restrictions, infection casesgradually increased in the US from March 15 and exceeded30,000 on April 2.
Fig. 39. Comparison of the Total Number of Daily Infection Cases in theUnited States (bold blue line) and Europe (orange line)
B. Number of Incoming Flights from Europe and InfectionCases in United States
With the travel restrictions from European countries to theUnited States, the number of incoming flights (blue line inFigure 41) suddenly decreased from March 13 to March 24.However, the number of infection cases (red line in Figure41) began to increase just after these restrictions. However,the latent period of COVID-19 is commonly five to six daysas we mentioned in Section VI-C, therefore we cannot rule ig. 40. Comparison of the Number of Daily Infection Cases in the UnitedStates (bold blue line) and Top-10 Countries in Europe out the possibility that the infections had already spread outbefore the US enforced these restrictions.
Fig. 41. Number of Daily Incoming Flights from Europe and New InfectionCases in the United States
European countries such as Italy, Spain and Germany havealready reported a certain number of infection cases beforetravel restrictions. Figure 42 shows the number of daily inter-national flights from these countries to the US and the numberof daily new infections in the US (red lines). In Italy (blueline), more infection cases were reported than other Europeancountries from March 1 (Figure 41), but the number of flightsto the US became almost zero just after the travel restrictionwas enforced. On the other hand, some flights from Spain(orange line) and Germany (green line) to the US were stillin operation, even after the restrictions were announced andinfection cases in the US began to increase. These flights maybe one of the reasons for the gradual increase of infections.
Fig. 42. Number of Daily Incoming Flights from European Countries (Italy,Spain and Germany) and New Infection Cases in the United States
VIII. D
ISCUSSION
A. Findings
In Section 4, we found out when and how internationalflights in each continent and countries declined due to travelrestrictions. The number of flights on April 1 had dropped to10% of that of regular seasons. The number of flights fromEurope sharply decreased in the second half of March afterthe US announced the travel restriction from most of Europeancountries on March 11 and the EU agreed to close borders onMarch 17. On the other hand, half of regular domestic flightsin the US were still in operation at the end of March.Besides counting the number of daily flights, we alsoapplied several graph analytics to daily flight networks in orderto explain how these networks became sparse in Section 5.With the result of analytics, we found out the density of theglobal flight network halved in March, and it has been dividedinto small strongly connected components from the mid ofMarch to the end of April.In Section 6, we evaluated the effect of reduction ofincoming flights to Barcelona on the tourism industry (e.g.the number of affected temporary workers). The number ofincoming flights dropped just after the travel restriction bythe government of Spain on March 16, and the number ofaffected workers drastically increased after the government ofCatalonia published the number of these workers from March23. However, the number of infection cases still increased evenafter travel restrictions were announced.We also investigated the relationship between incomingflights from Europe and the number of infections in the USin Section 7. We found that the trend of daily infections inEurope began a week ahead of the US, and that some Europeancountries confirmed cases from March 1 while the number ofinfection cases in the US began to increase from March 15.We also found that flights from Europe were still accepted inthe US even after travel restrictions were announced on March14.
B. Related Work
Many researches have conducted evaluations and estima-tions how COVID-19 infections affected international flights.Stefano et al. [18] evaluated the economic impact of travelrestrictions on flights. They constructed several scenarios oftravel restrictions based on the past pandemic cases such asSARS in 2003 and MERS in 2005, and estimated the loss ofglobal GDP may amount up to 1.98% and that the number ofunemployment could reach up to 5 millions.Hien et al. [19] evaluated the correlation of domesticCOVID-19 cases and flight traffic volumes with the OpenSkyNetwork dataset, and concluded the number of internationalflight routes and passengers have a correlation with the riskof COVID-19 exposures.
C. Limitations and Possible Bias
Since the flight dataset from the OpenSky Network isaggregated voluntarily, it has some limitations. First, someflight entries do not contain origin or destination airport codes.e excluded such entries before analysis to construct validflight networks. Moreover, the dataset also has a significantregional bias. For example, many international and domesticflights connecting China, Oceania, South America, and somecountries in Africa are missing.IX. C
ONCLUSION
A. Overall Conclusion
After the declaration of a pandemic by WHO on March11 and government announcements of travel restrictions dur-ing the following week, the number of international flightsaround the world drastically declined, particularly in Europe.Although many domestic flights were still in operation in theUS, the frequency gradually declined.These travel restrictions affected tourism-dependent coun-tries. In Barcelona in particular, many temporary workersmade applications regarding their unemployment soon afterthe government opened applications on March 23. Moreover,the number of newly infected cases increased after travelrestrictions were declared on March 16, most likely due toCOVID-19s longer latency period.
B. Future Directions
Due to the limitation of publicly available data, we couldnot conduct the same analysis for many countries as Barcelonain Section 6 for this paper. Therefore, we aim to analyzerelationships among the number of flights, infection cases, andthe impact on the economy for more countries and continentsby combining various other data sets with the flight data inour next analysis. We also aim to focus on counties wherethe number of infections has soared, and investigate how theincrease of cases and reduction of flights affect the economy.We also plan to trace and predict viral mutations with theflight network data. The type of COVID-19 virus varies fromregion to region [20], [21], and this may provide insight toanalyzing the route of infections with viral types. With thedaily flight network data, we hope to estimate the date androute of infections more precisely.In order to predict the economic impacts such as propaga-tions of revenue declines in travel industries for each region,we will apply advanced machine learning techniques fortime-series graph data such as Graph Convolutional Network(GCN). We will also evaluate the importance of flight networkfeatures as well as demographic features of regions (countriesand continents) in our next analysis.R
IPSN-14 Proceedings of the 13th International Symposiumon Information Processing in Sensor Networks et al. , “Introductions and early spread of sars-cov-2 in the new yorkcity area,”
Science , 2020.[18] S. M. Iacus, F. Natale, C. Santamaria, S. Spyratos, and M. Vespe,“Estimating and projecting air passenger traffic during the covid-19coronavirus outbreak and its socio-economic impact,”
Safety Science ,vol. 129, p. 104791, 2020.[19] H. Lau, V. Khosrawipour, P. Kocbach, A. Mikolajczyk, H. Ichii,M. Zacharski, J. Bania, and T. Khosrawipour, “The association betweeninternational and domestic air traffic and the coronavirus (covid-19)outbreak,”
Journal of Microbiology, Immunology and Infection , 2020.[20] P. Sagulenko, V. Puller, and R. A. Neher, “TreeTime: Maximum-likelihood phylodynamic analysis,”
Virus Evolution , vol. 4, no. 1, 012018, vex042. [Online]. Available: https://doi.org/10.1093/ve/vex042[21] J. Hadfield, C. Megill, S. M. Bell, J. Huddleston, B. Potter,C. Callender, P. Sagulenko, T. Bedford, and R. A. Neher, “Nextstrain:real-time tracking of pathogen evolution,”