Indonesian “tsunami-generation” in a nutshell: systematic literature review

Publications related to tsunami-generation phase hold 42.8% of all the publications over all tsunami hazard literatures in Indonesia. Corroborating the fact, tsunami-generation studies are also vital in determining the type of propagation and inundation that may appear in the surrounding coastal regions, which in practice, can also help determining disaster management specifics and risk reduction activities. A literature review regarding tsunami-generation has been done for those purpose by using our CARI! Knowledge Engine, which includes collections of both International and Indonesian journals within the scope of Indonesian locus context. This study unpacks publication trend, the most influential articles, top contributing journals, top main authors, top institutions, and dominant countries as well as the science mapping of tsunami-generation subjects. To gain deeper understanding, we also develop tsunami-generation research framework within the basis of tsunami generation force, time occurrence, monitoring, and measurement approach, as well as hazard analysis approach. From the review, an interesting peak of tsunami-generation publications has been shown in 2020 (39 publications), following the occurrence of several Indonesian major tsunamis in the previous decade and the globally impacting Aceh Tsunami (or the Indian Ocean Earthquake and Tsunami) in 2004. Among all the publications of tsunami-generation force, tsunamigenic earthquake has become the most studied topic (67.7% of all publications), while the most discussed atypical sources mainly come from volcanic eruption (46.67% of all atypical literatures) and submarine landslide (33.33% of all atypical literatures). In terms of tsunami-generation source, it is found that the Sunda Megathrust has been the most-discussed location, which is mentioned within 39 publications, followed by the Palu-Koro Fault with 18 publications and the Krakatau Volcano by 16 publications. Apart from the notice of significant difference between research of historical tsunami and paleotsunami (120:5 publications), some gaps also can be found when discussing the meteotsunami-generation mechanism in Indonesia and the Global Navigation Satellite System’s Total Electron Content (GNSS-TEC) method, which, although proven beneficial for the development of tsunami early warning system, accuracy is still the main issue in capturing tsunami signals and positioning actual source location of tsunami. These were some highlighted limitations that arise in our process to understand tsunami-generation mechanisms and the exploitations of the knowledge. Therefore, future studies specifically within these areas are further encouraged.


Introduction
Tsunami is a natural disaster that can cause significant damage to coastal communities on a wider area range across oceans.Scientifically, tsunami's life cycle can be divided into three main phases: generation, propagation, and inundation [1].During the generation phase, a tsunami is created by a sudden displacement of water, which is caused by many factors from typical (i.e., earthquakes) to atypical (i.e., volcanic eruption, landslide, atmospheric disturbance, etc) origins.According to historical records, more than 80% of world tsunamis were generated by earthquakes, and 70% of them, including large tsunamis, were observed around the Pacific Ring of Fire [2].
Situated in the Pacific Ring of Fire, Indonesia is prone to earthquakes and volcanic eruptions.As a consequence, most of the tsunamis in Indonesia are generated by earthquakes, which can occur along the subduction zones, followed by volcanic activity and landslide [3].Based on history, there were at least 246 tsunami events that occurred in Indonesia in the period between 416-2018 AD [4].Common locations where tsunamis are generated in Indonesia are those located in the Sunda Trench, Java Trench, and eastern part of Indonesia as shown in figure 1.The Sunda Trench or Sunda Megathrust, which runs along the west coast of Sumatra and Java, is the source of several recent devastating tsunamis, including the 2004 Aceh tsunami and the 2010 Mentawai tsunami.The eastern part of Indonesia also has as many historical records of tsunami events as the western part.However, the main source of the tsunamis in the eastern part of Indonesia is rarely known and there are no detailed sources of its information [3].Other sources of tsunamis in Indonesia include volcanic eruptions such as the recent 2018 Sunda Strait tsunami caused by the Anak Krakatau Eruption [5].Indonesia is particularly vulnerable to tsunamis due to the interconnected nature of its waters, which means that the effects of a tsunami in one area can impact surrounding regions, increasing the likelihood of damage.Coupled with the high population density in coastal areas, the risk of a tsunami disaster in Indonesia considered as high [6].The generation of tsunami has been one of the important subjects within tsunami research and understanding the concept of tsunami-generation can give insights about This study utilized CARI! Knowledge Engine database for its collection of literature pertaining to disasters.This database sources literature from Scopus, Directory of Open Access Journal (DOAJ), and Portal Garuda and has the added feature of automatically tagging literature by study location and hazard type which simplifies the process of collection during SLRs.The research keywords used were designed to be broad enough to capture relevant literature while also excluding irrelevant results.The keywords used for each stage during the selection process of tsunami-generation topics are listed in Table 1.The literature was obtained by multistage filtering by keywords from stage 1 to stage 4 with a separate intention rationale of each stage.The search was conducted between November 2022 and January 2023.
To identify literature related to tsunamigeneration

Manually by researcher
The aim of this review method is to summarize studies that specifically investigate the assessment of "tsunami-generation" literatures in more detail.Literature that does not pertain to the preferred rationales in table 1 were excluded from the study.The study also uses content analysis by developing framework on tsunami-generation research, thus we deliberately tagging each literature based on our proposed conceptual analysis on tsunami analysis approach (e.g., numerical, empirical, or other approaches), historical tsunami occurrences (e.g., historical or paleo tsunami), tsunami-generation sources (e.g., earthquake or volcano locations), and tsunami-generation types (e.g., tectonic, volcanogenic, or others).
To examine the productivity and impact of a specific research area, a performance analysis is used.This approach is similar to profiling participants in research studies but uses more advanced bibliometric metrics for greater accuracy [8].This study aims to discover published publication trends in number, impactful articles, impactful journals, productivity from main authors, leading institutions, dominant countries, and the methods and context used in tsunami-generation research.
Furthermore, this study leverages the science mapping and network analysis method to gain the understanding of co-analysis covering in literature corpus.Network analysis uses keywords extracted from abstract literature, then network analysis helps classify the studies clusters to show information diversity in the field by identifying differences among keywords.This analysis was assisted by VOS viewer software to create co-occurrence networks of terms extracted from English-language textual data such as titles and abstracts of publications.In this study, abstract of literature written in Bahasa Indonesia were manually translated to English in order to complied with software analysis.Moreover, VOS viewer enables researchers to omit noun phrases that have a low relevance score, allowing them to get rid of general noun phrases.The remaining terms typically represent topics related to the domain of interest, which can be visualized in a co-occurrence network.
Ultimately, this study developed an organizing framework of tsunami-generation research by adopting the previous work of tsunami hazard research.The framework is used to quantitatively and qualitatively analyze tsunami-generation contents based on their interconnection between topics identified from science mapping.Quoting from the November 2022's Indonesian Disaster Knowledge Update (IDKU) of http://caribencana.id,tsunami frameworks can be divided on several concepts, such as tsunami phase, time occurrence, monitoring and measurement approach, and its hazard analysis approach (Figure 3).As the generation of tsunami is the first part of tsunami phase, we prefer to exclude tsunami propagation and inundation phase from the discussion and instead, employ the generation force of tsunami as one of the proposed frameworks.The other three concepts are then analyzed within the scope of tsunami-generation which especially within the region of Indonesia.Eventually, wordings from this process are constantly checked with the Tsunami Glossary to ensure universal understanding of tsunami-related terms.It is important to understand that terms generated from the science mapping process might appear to be different than those in the glossary.For the frameworks purpose, similar terms presented in the glossary will be used.
Within the whole processes, Sankey diagram may be used to analyze relationship between subtopics of tsunami-generation research theme (portrayed by boxes).This may be visualized proportionally based on the number of literatures.The larger size of the box and the wider lines connecting two boxes indicate a greater number of publications accounted for them.In this paper, word cloud is also used to understand topics related with tsunami-generation.

Result findings and discussion
After conducting the PRISMA protocol, this study has analysed a review of 186 literatures comprises of 166 journal articles and 20 proceedings which are published between 1995 and 2022.The dataset literature consists of 123 pieces of literatures sourced from Scopus, 36 pieces sourced from Portal Garuda, and 27 pieces from DOAJ directories.According to the written language used in the literatures, it is found that 157 literatures has been written in English, and 29 literatures written in Bahasa Indonesia.These Indonesian publications are then translated to English to ensure the inclusion of the publications within the corpus.The review analysis in this study has started from bibliometric analysis of performance and science mapping, followed by a deep dive into content analysis using the developed tsunami-generation framework.

Performance analysis
The performance analysis study has successfully discovered the quantitative metrics of the publication trends, the most impactful articles, top contributed journals, top contributing main authors, top contributing institutions, contributed countries, and the methods and context used in tsunami-generation research.

Publication trend.
The publication trend of research on tsunami-generation in Indonesia over the years is shown in Figure 4.The data shows that the earliest article on this topic in Indonesia was published in 1995 [10], and that the number of publications has steadily increased in the last 30 years.There were only a few publications in the early years, but the field has grown significantly, especially after 2009s, with the highest record of 39 tsunami-generation publications published in 2020.Notably, from the figure we can see that the research activity in this field has increased significantly following the Indian Ocean Tsunami (IOT) of 2004, as well as the following Mentawai tsunami of 2010, and even more so after the Palu tsunami and Sunda Strait tsunami of 2018.Badan Meteorologi Klimatologi dan Geofisika (BMKG) has recorded 189 tsunami events from year 416 until 2018 that took place in the Indonesian region [4].Following the major tsunami caused by Krakatoa Eruption in August 1883, there were numerous tsunamis eventuate after this event but have less impacts until the Indian Ocean Tsunami (IOT) of 26 th December 2004 took place.The IOT became the deadliest tsunami event that occur in modern era, due to its casualties of 227.898 persons in countries adjacent to Indian Ocean [11] [12].Approximately, there are 169.000casualties counted that are coming from Indonesia.This event triggered many researchers around the world to study this tsunami event from many disciplines and topics including tsunami-generation even after a decade.

Top influential articles and top contributing journals.
The top ten influential articles indicated by articles citation number about tsunami-generation in Indonesia has been presented in below.The table shows M. Chlieh's article [13] as the most cited article in the field with the total of 344 citation since published in 2007 or average of 22,9 citations per year.Interestingly, the top 5 cited articles [13][14] [15][16] [17] are all investigated about tsunami-generation caused by Great Sumatran megathrust in Indian Ocean, also all of authors managing a research collaboration among the expert from different affiliation.The next most cited articles studied probabilistic tsunami hazard assessment authored by Horspool [18], numerical simulation of flores tsunami by Imamura [10] and Sulawesi tsunami by Heidarzadeh [19], and source mechanism of tsunami triggered from volcano eruption by Paris [20].The literature corpus of 186 articles on tsunami-generation were published across 98 journals, Figure 5 below only shows the journal with a minimum of three articles published within (contains 16,3% or 92 articles).The size of the text indicating the number of articles published within and their colour indicating the total citation number of each journal.The most contributing journal is Pure and Applied Geophysics (PAG) with 18 articles, followed by Natural Hazards and Earth System Sciences (NHESS), Natural Hazards, and Science of Tsunami Hazards with each 8 articles published.However, in terms of influence, Natural Hazards and Earth System Sciences (NHESS) leads the position with 365 citations each, followed by Journal of Geophysical Research: Solid Earth (JGR) with 338 citations, while Pure and Applied Geophysics (PAG) have 174 citations.Even though NHESS and JGR have much less published articles related to tsunami-generation in Indonesia compared to PAG, the articles become most impactful, this is consistent from their h-index which considerably higher than the other journals.The top ten contributing authors for tsunami-generation in Indonesia is presented with Table 3.The table shows that Chlieh M. from California Institute of Technology, United States and Subarya C. from National Coordinating Agency for Surveys and Mapping, Indonesia are the most influential is evidenced by the number of citations (TC) in their publications.While the most productive authors with many citations are Natawidjaja D.H. from Lembaga Ilmu Pengetahuan Indonesia, Indonesia, Løvholt F. from Norwegian Geotechnical Insitute, Norway, and Mustafa B. from Universitas Andalas, Indonesia with 3 total publications (TP) with overall 174 total citations, 2 total publications and 83 total citations, 2 total publications and 72 total citations respectively.
In general, most of Indonesian researchers tend to have studied tsunami-generation across regions when compared to researchers from abroad who have studied more in the Sumatra and Sulawesi regions (Figure 8).As shown in Figure 8 Most of Indonesian researcher affiliations based in Java region, followed by the national institution affiliations such as central government agencies (BMKG, LIPI, Badan Geologi, etc).There are very few research affiliations come from Sumatra and Sulawesi, despite these two areas are prone to tsunamis and devastating tsunamis have occurred in their respective regions.No affiliations of Indonesian researchers were found from the Bali & Nusa Tenggara regions and the Maluku & Papua regions.This condition is due to the difference in the number of universities and research institutions, especially those concerned with disaster management issues, which are more common in the Java region than in the Sumatra and Sulawesi regions.Similar to the previous explanations, the Sumatra region is still the main focus for researchers from local univeristies in Indonesia, followed by national-wide and conceptual study category and then research about tsunami hazard in Java, Sulawesi, Bali & Nusa Tenggara, and Maluku & Papua.

Science mapping
The science mapping process has successfully generated network maps of literature in tsunamigeneration, in addition we also split the analysis within the network and temporal basis to better gained understanding of the evolution of produced knowledge in the topic within the years.

Network analysis.
The overview of the network produced by VOS Viewer has generated the nodes of keywords and their related links within tsunami-generation field, which can be seen in Figure 9.The circle nodes' sizes symbolize the frequency of keywords occurrences, while the nodes that are closer together indicate a higher occurrence of the keywords appearing together, and the thickness of the link connecting them shows the significance of the co-occurrence.Four clusters of networks are found which consisted of central node which is tsunami (blue; 389 occurrences), earthquake (green; 424 occurrences), model (red; 116 occurrences), and parameter (yellow; 59 occurrences).Network analysis reveals the co-occurrence of keywords in a literature and effectively maps the strengths and traces the relationship structure in the theme.In general view, study about tsunami is very linked to earthquake hazard.

Temporal analysis.
The average number of literatures published per year and its earliest published literature shown in Figure 11 tends to focus on tsunami that generated by earthquake and located in Sumatra and Indian Ocean.Moving to the mid-2016s, research on tsunami-generation tend to focus on the parameter, model, and developing model.On the latest published literature, research tend to focus on the recent big tsunami event such as The Sunda Strait Tsunami and Palu Bay Tsunami in 2018.Other important topics that can be captured within this time are around the submarine landslides, volcanicgenerated tsunami, and events around Anak Krakatau.In this section, topic dynamics within several time periods will be discussed to see the changes that occurs overtime within the tsunami-generation field.
Assessing popular topics within time periods requires several time markers to highlight the most used topic words between each period.For this, the majority of research on tsunami-generation has been examined on the past tsunami event that happen across Indonesian water.The three recent big tsunamis become the most studied topic with just below 30 literatures, which are: tsunami event of Palu 2018, Aceh 2004, and Krakatau (Sunda strait) 2018, respectively.Some literatures included more than one tsunami event as their studies, in order to conduct comparison of tsunami-generation mechanism or different tsunami event that occur on the same location.
For the temporal analysis, the most relevant markers are set for the top-3 major tsunami events in Indonesia shown in Figure 12.Thus, the first marker is set on the time of Indian Ocean Tsunami, which happened on the 26 th of December 2004; the second is the Mentawai Earthquake and Tsunami on the 225 th of October 2010; and the last were the Sunda Strait Tsunami and Palu Tsunami in 2018, which were happened on the 22 nd of December and the 28 th of September, respectively.Because all those events were occurred during the second half of their respective year, the year 2004, 2010, and 2018 are marked as the end of a period and their next years are counted as the start of the next period.A total of 66 papers were counted within a period of 2011 until 2018 where the number of literatures varies annually, between 4 to 12 papers per year.From 2019 onwards, the number of papers were skyrocketed to a maximum of 39 in 2020 (Figure 4), with a total of 90 literatures counted to date.
For each of the period, word clouds are used to determine major topics discussed within each time.The series of figures below (Figure 13, Figure 14, Figure 15, Figure 16) shows the variety of word clouds over the course of four periods of time.Each of them has undergone several word filters, which ranging from the stop words until the common words in tsunami research.However, these common words filter are not generalized.For example, the word 'Tsunami' in the first period has been removed due to the very high frequency of the word compared to other words (Figure 13a).Thus, to help maximizing the visual of the other words, 'Tsunami' has been removed (Figure 13b).Similar case happened with the fourth period where 'Tsunami' in Figure 16a was removed, resulting in the word cloud in Figure 16b.
Before 2004 (Figure 13), tsunami-generation topic was not focused on any specific source mechanism.The most talked topic was about the main generations of tsunami in Indonesia [10][22][23] [24], which are circling around earthquake and landslides.Publications of tsunamigeneration during this time were inclined towards the kind of propagated waveforms generated from the source, and historical tsunami catalog.The inundation risk was also being discussed along with the generation topic, also the risk components in "Riangkroko" and other Sulawesi coast areas were analyzed.Until 2004, the highest frequency of historical tsunami events in Indonesia came from the Makassar Region [22], hence the large "Makassar" and "Strait" words in the picture.
Unlike the years before the Indian Ocean Tsunami of 2004, tsunami caused by "earthquake" was the most popular topics in the following years until 2010 (Figure 14).This topic was on par with the "tsunami" topic itself, since the 2004 tsunami was generated from fault ruptures over the Andaman Sea, near Sumatra Island [25][17] [11].From the figure, the tsunamigenic discussion was also extended to the Thai coastal area, covering topics such as the megathrust study, earthquake magnitude, the seafloor uplift, tsunami elevation, and effects to corals due to sedimentation.Discussions circling around tsunami risk from its propagation and inundation like the period before 2004 were less present, and so the comparative study to tsunamis around the world like those in Pacific or the Indian tsunamis itself.
After the 2010 Mentawai Earthquake and Tsunami, it is evident that both the two key terms in the previous period were still the focus in this period (Figure 15).Similar with the times before the Indian Ocean Tsunami takes place, the propagation waves and their parameters are one of the topics that were popular, while the discussions about earthquake's geological parameters were more becoming after the last period.During this period, there was also an introduction of the Indonesian Tsunami Early Warning System (TEWS) that was established in 2008 [26], and this time, the introduction of tsunami observation analysis using satellite imageries appeared and were becoming one of the popular topic discussions in Indonesia.
After being hit by two major tsunamis in Indonesia [27][28], earthquake-generated tsunamis are still the main topics to be discussed till date (Figure 16).The introduction of tsunami in Sulawesi [28] shows the major increase in the discussions about tsunami in Palu, one of the province capitals in Sulawesi.The liquefied gravity-flow induced tsunami gives "liquefaction" the same amount of attention as tectonic-generated tsunami.Due to the occurrences of both tsunami that are located in straits (Sunda Strait and Makassar Strait), the coastal areas are almost equally highlighted the "wave" discussion.The high number of literatures published might also be one of the tsunami awareness campaigns coming from of the establishment of Sendai Framework for Disaster Risk Reduction on the 14-18 March 2015 [29] and the UN GA 70/203 World Tsunami Awareness Day on the 5 th of November 2015 [30].Nevertheless, the graph in Figure 4 shows a spiking interest from people to learn more about "tsunamigeneration" as an important part of tsunami hazard assessment in Indonesia.

Tsunami-generation research framework
Since the tsunami phase context has been excluded from the discussion, the remaining contexts of time occurrence, monitoring and measurement approach, and hazard analysis approach are analysed along with the tsunami-generation force.These frameworks below are applicable for Indonesia, and the terms identification for each category will be discussed thoroughly.

Tsunami-generation force.
As portrayed in the network analysis section, several tsunamigeneration forces have been identified, which are: earthquakes, volcano eruptions, submarine landslides, and flank collapse or the aerial landslides.In some literatures [2][31], there are two major classifications of tsunami-generation, which are typical and atypical tsunami.The typical source of tsunami is the type of tsunami that is produced by tectonic earthquakes.In some literatures, tsunamis this type is mentioned as the tsunamigenic earthquakes [32][33] [34].Looking at Figure 17a, it is evident that this type of tsunami has been leading by 126 literatures, or 67.7% of all tsunami-generation literatures in this study.The atypical sources of tsunami, on the other hand, is classifies as landslides (submarine or aerial), atmospheric disturbance, and volcanic eruptions.Tsunamis originated from landslides are called tsunamigenic landslides [35], while tsunamis generated by atmospheric disturbance are called meteorological tsunami or meteotsunami [2][36], and those coming from volcanic eruptions are called the volcanogenic tsunami [37] [38].In total, these tsunamis constitute 60 literatures (32.26% of all literatures in this study), where volcano eruptions been studied within 28 literatures (46.67% of atypical literatures) or the most popular atypical tsunami studies within the field (Figure 17a).This is followed by submarine landslides by 20 literatures (33.33% of atypical literatures).The 'combination' findings in Figure 17a denotes the study of tsunami-generation that investigates one until three combinations of force, while the 'all' findings shows the literatures that study all the tsunamigeneration forces.A thorough study for specific tsunami source-generator identifies 89 literatures with a specific tsunami-generating source (Figure 17b), while the other 97 literatures are mostly discussing either several tsunamis at once, focus on lessons learned for Indonesians from tsunamis outside of Indonesia [39] [17], or the study of tsunami-generation in general.Eventually, the figure shows some source generators that are circling around the tectonic fracture zones (Figure 17b).Among those, Sunda Megathrust has been the most frequent source that are discussed among scholars by 39 literatures, which is followed by Palu-koro fault by 18 literatures.The third popular source studied among scholars is the Krakatau Volcano by 16 literatures.Apparently, Krakatau is the only volcano in Indonesia that is found to be studied for its tsunami-generating mechanism.Although meteotsunami is not frequent in Indonesia, some studies abroad may mentioned the meteotsunami as the result of typhoons [40] or tropical cyclones, which may possibly be happen over the northern and southern Indonesia due to the frequent cyclones.
It is also noteworthy to mention that the sources in Figure 17b may or may not be the factor that generates tsunami.Some of them, for example, the Palu-Koro fault in Sulawesi, has generated tsunamigenic landslides that creates wave-related hazard along the Palu Bay, Indonesia [41][28] [19].Note that the Palu-Koro fault is the location and eventually, source generator for the following tsunamigenic landslides, not the one that causes tsunami.Despite that, the findings in Figure 17b corroborates the dominance of tectonic fracture zones in generating tsunamigenic earthquakes in Indonesia (Figure 17a), followed by volcanic eruptions.

Time occurrence.
Studying tsunami based on their time occurrence is important for scholars to understand which methodology to use to analyse the specific tsunami.The data for recent tsunami can (b) (a) easily be found from several data sharing platform (e.g., NGDC/WDS Global Historical Tsunami Database [42]), while for ancient tsunamis where the documentation of any sort is unable to be found, many methods such as radiocarbon dating are being used [34][43] [44].
In general, there are two major classes of tsunami type based on the time occurrence, which are historical tsunami and paleotsunami.The historical tsunami denotes the tsunami that has been documented through eyewitness or instrumental observation within the historical record.In practice, historical tsunami ranges from several approaches: the manual documentation [45], which dated back from the first tsunami's written documentation from observation; and instrumental observation [46], which dated back from the 19th century onwards, from the time when the first automatic gauge recording was available.The paleotsunamis, on the other hand, are tsunamis occurred prior to the historical record or for which there are no written observations.To successfully identify this type of tsunami, scientists need to use specific measurement methods, one of which is the tsunami deposit dating [47][48] [49] to obtain the paleotsunami characteristics.Our study highlights the significant differences between research discussing historical tsunamis and paleotsunamis (Figure 18).As shown in the figure, there are 120 research publications that discuss historical tsunamis, with 5 research publications related to paleotsunamis, and 3 publications discussing both types.This finding indicates that there is a large knowledge gap between historical tsunamis and paleotsunamis.One reason might come from the fact that more recent tsunami events have advantages in the use of manual or instrumental observation [46], thus are easier to study due to the availability of sequential dataset.Meanwhile, the data needed to study paleotsunamis are more difficult to obtain due to the required field observation [34].In addition, the recent major tsunami events that have occurred in Indonesia over the past two decades might have encouraged researchers to study historical tsunamis.

Monitoring and measurement approach.
Identifying tsunami can be done using several methods or physical tools that function manually or automatically.These tools are used as monitoring purpose by measuring data and store them in the cloud, which could also be done in both manual and automatic modes.These tools are classified as mareograph, remote sensing, tsunami deposit mapping and tsunami deposit dating.From all the literatures, there are 68 that are specifically focus on monitoring and measurement approach (Figure 19).Among these, remote sensing has been used by 41,18% of monitoring literatures, followed by tsunami deposit mapping by 23.53%, mareogram by 22.06%, and tsunami deposit dating by 11.75% of monitoring literatures.
As a tool to measure and monitor time series of sea level, mareogram has been around since 1830 [45] in the form of mechanical float and stilling gauges until this day with the advancement of seafloor bottom pressures.In Indonesia, studies using mareogram data has occurred since 2011, where all of them implementing gauges data in their study.An advanced example of a mareograph that is built specifically for rapid tsunamigenic earthquake detection is the Deep-ocean Assessment and Reporting of Tsunamis (DART) that records tsunami signal from bottom pressure recorder.This device, which is widely known as the tsunameter, is located in the rim close to subduction zones, and data transfer from the bottom to ground controller is done through the connected surface buoy and satellite receiver [50][51].This real-time tsunami monitoring system has been positioned in strategic locations throughout the ocean, which includes the surrounding of Indonesian waters [52].As early as the founding of mareograph, a remote observation in the form of aerial photography was first demonstrated in 1850 [46].Ever since, remote sensing area have been developed in the forms of coastal high frequency radar for wave monitoring and ocean monitoring system using satellites.In Indonesia, remote sensing studies related to tsunami-generation has been around since 2007.The recent fruitful implementation of tsunami detection using remote sensing methodology is the use of Ionospheric Total Electron Content (TEC) disturbances data derived from Global Navigation Satellite System's (GNSS) ground-based stations.Using signal analysis, tsunami wave can be extracted to find the time and location of tsunami occurrence [53].This GNSS-TEC method has also been tried out in Indonesia [54] [52] to develop the Ionospheric Tsunami Power Index (ITPI) using data from nine tsunami events.
Unlike both mareogram and remote sensing techniques that can only be useful for modern tsunamis, the tsunami sediment deposit identification and mapping can be done to anticipate future tsunamigenerator forces that may occur around a certain area.This method could work for both historical and paleotsunami data and usually done to anticipate earthquakes.Tsunami deposit mapping has been used several times since 2005 in Indonesia, as a general assessment of tsunamigenic earthquake using field surveys of surface ruptures, bathymetry mapping, and seismic-reflection survey [55].In Indonesia, it has been demonstrated as a post-disaster survey of Palu Tsunami 2018.
As previously mentioned, paleotsunami assessment is usually conducted using tsunami deposit dating.This includes radiocarbon dating of deep-sea core sampling of sediment deposit, coral sample, and seashell sample analysis [43]

Hazard analysis approach.
In assessing tsunami-generation research, some approaches to analyse tsunami is used to understand source mechanism underlying the propagation of the tsunami itself.The analysis ranges from the general assessment and hazard mapping to a series of modelling approaches.
The general assessment counts any information related to tsunami-generation assessment in a general setting, which includes tsunami-generation characteristic analysis [48][24] using a set of known parameters from previous assessments, such as paleo tsunami study or a case-area study.The papers with a focus on the making of tsunami hazard map is considered as the hazard mapping category [57].Between the modelling methods, empirical model comprises of specific estimation parameters using a set of statistical analysis.This includes the inverse estimation of source's parameters and estimation of parameters' change, such as earthquake's rupture length, speed reduction and travel time delay [58] [59].
The numerical model, on the other hand, is using advanced mathematical model to simulate tsunami's parameter on a wider setting (two or three-dimensional, wider basin, detailed resolution, etc.).On the implementation, numerical model can be as simple as the Deterministic Tsunami Hazard Assessment (DTHA) [60] that only model one scenario using a fixed set of input parameters or can be as complicated and advanced as the Probabilistic Tsunami Hazard Assessment (PTHA) [18][10] that can model several scenarios at once.A travel time maps, coastal amplitude maps, velocity, and energy directivity map is one of the results to be expected from using PTHA.
A real-time model and forecast relate to the rapid but reliable model to provide real-time sea-level changes using data gathered from deep sea platform located around the subduction zones on the rim of the ocean basin [50].Further, this connects to the discussion about tsunami early warning and early detection analysis.Another form of real-time forecast related to tsunamigenic landslide is the Dinsar application that can detect flank collapse before the occurrence of tsunami [61].On Figure 20, sankey diagram illustrates the distribution and relationship of research literature between studied location, tsunami modelling, and tsunami-generation force.Research that examined in Sumatra region tends to use general assessment of tsunami-generation studies, accompanied by numerical modelling approach.While research on national or conceptual level generally tends to use both approaches equally.In contrast, research on Sulawesi region tends to use more numerical modelling approach to study tsunami-generation compared to general assessment approach.In Java region, studies are more drawn to the approach of numerical modelling accompanied by empirical modelling, while research in Bali & Nusa Tenggara, Maluku & Papua is comparatively equal among tsunami modelling approach.Emphasizing on tsunami hazard analysis, all approaches are used to highly investigate tsunami-generation force by tectonic mechanism.

Conclusion
The generation of tsunami has been one of the important subjects within tsunami research as it holds the key to map and analyze the propagation and inundation area of a tsunami, which can later be the base of disaster management system and disaster risk reduction activities.From times, there has been some improvements in literatures of tsunami-generation, where 2020 marked the peak of publications in the field, by 39 publications.As a typical source of tsunami, tsunamigenic earthquakes have been the most studied articles where 67.7% of the literatures covering the subject, while the highly discussed atypical sources, volcano eruptions, has a lot of mentioning that covers 46.67% of all the atypical tsunami literatures in Indonesia.The next popular atypical studies are within the topic of submarine landslides which covers 33.33% of all atypical studies.Eventually, the focus study of the typical tsunami is mainly circling around the Sunda Megathrust (39 publications), while the three most popular generation sources for atypical tsunami were around Sumatra and Sulawesi regions (18 in Palu Koro Fault, 16 in Krakatau, and 7 in Mentawai Fault).
While these locations are known mostly for their historical tsunamis (shown by 120 publications within historical tsunamis against only 5 publications within paleotsunamis), more paleotsunamis in the regions also store some important tsunami-generation characteristics that can be traced around the megathrust area.The research review regarding paleotsunami was evident to be low in Indonesia and thus this gap can be reduced by increasing the interest and opportunities related to paleo research.
Over the eastern Indonesia, historical tsunami database from year 416 until 2018 shows as frequent tsunami appearances as those in the western areas, but our study shows that the tsunami-generation literatures linked to the locus were very low compared to those on the western side.As the information about tsunami generations and their cycles can raise some ideas regarding disaster management, conducting future studies over tsunami-generation in the eastern side of Indonesia is strongly encouraged to promote disaster risk reduction among the fellow citizens living around the tsunami-prone areas.
In Indonesia, the development of monitoring tools is always ever-progressing, especially in terms of the use of remote sensing and mareogram.One thing that is mostly escaped the researcher's sight are the monitoring and analysis of meteotsunami, which is due to their generation mechanisms that are yet to be fully understood.Considering the vast areas of atmospheric disturbance surrounding the northern and southern Indonesia that is constantly hit by cyclones, this field of research will be very interesting to be studied in the future as it can lessen the probability of misdiagnose the disaster, which can better the mitigation activities.
Further improvements have also been seen regarding the analyzing technique used by researchers.One of which to be highlighted is the use of GNSS TEC where this method has been shown to be very promising to support real-time forecasting of tsunami-generation in the future.However, due to the limitations of the number of GNSS ground-based stations, a less accurate tsunami source location and the probable tsunami propagation to the surroundings can only be achieved.This could be a challenge to Indonesian scholars to build more ground stations for tsunami early warning system purpose.
Over the decades, monitoring and analyzing tsunamis has been inseparable.It has been shown from our study that the development of both processes can still be developed further.The examples from the literature reviews have shown the valuable efforts scholars did to understand the tsunami generation and from those, increase the effectiveness of tsunami early warning system.Limitations may arise, but those are the more reason why these monitoring and analyzing technique needs to be further studied and developed.

Figure 2 .
Figure 2. PRISMA 2020 flow diagram of systematic review used in this study.

Figure 4 .
Figure 4. Number of published literatures per year since 1994 to 2022.

Figure 5 .
Figure 5. Top contributing published journal
, the proportion of research region in the Sumatra, Sulawesi, and nationwide/conceptual is studied more widely when compared to the regions of Java, Maluku & Papua, and Bali & Nusa Tenggara.The regions of Sumatra and Sulawesi have become the focus of research due to the devastating tsunamis that occurred in Aceh in 2004, Mentawai in 2010, and Palu in 2018, highlighting the inseparable relationship between the region and the need for further study.The research locus in the Java region was driven by the 2006 Pangandaran and 2018 Krakatau tsunamis.There is only a small proportion of the total studies that examine tsunami-generation with a research locus in central and eastern Indonesia, especially the Bali & Nusa Tenggara and Maluku & Papua regions.
TP = Total publication b TCP = Total cited publication c TC = Total citations d TC/TP = Average citations per publication e TC/TCP = Average citations per cited publication

Figure 8 .
Figure 8. Sankey diagram showing the origin of researcher affiliation by country (left part), locus study (middle part), and origin of Indonesian researcher affiliation by Indonesia subregions (right part)

Figure 9 .
Figure 9. Keyword network colour representing the generated cluster from selected literature.

Figure 11 .
Figure 11.Keyword network color representing the average number of literatures per year.

16 Figure 13 .
Figure 13.Word clouds that show the time period before 2004 with 'Tsunami' topic included (a) and without 'Tsunami' topic (b).

Figure 14 .
Figure 14.Word clouds that show the time period between 2005 and 2010.Figure 15.Word clouds that show the time period between 2011 and 2018.

Figure 15 .
Figure 14.Word clouds that show the time period between 2005 and 2010.Figure 15.Word clouds that show the time period between 2011 and 2018.

Figure 16 .
Figure 16.Word clouds that show the time period within 2019 and beyond with 'Tsunami' topic included (a) and without 'Tsunami' topic (b).

Figure 19 .
Figure 19.Number of literatures according to observation and measurement methods (n=68).

Table 2 .
Top ten most influential articles about tsunami hazard assessment in Indonesia.

Table 3 .
Top ten most influential main author about tsunami hazard assessment in Indonesia.