The link between sea level fluctuations and
volcanic activity on volcanic islands over the
last glacial-interglacial cycle
Msc Earth Science Thesis
Environmental Management track (30 EC)
Computational Geo-Ecology Research Group
Institute for Biodiversity and Ecosystem Dynamics (IBED-CGE)
March 2020
Author:
Supervisors:
Floris Veloo
Dr. K. F. Rijsdijk
Abstract
As a result of a warming climate, the sea level is expected to rise rapidly during the coming centuries. Sea level fluctuations could, in theory, have an influence on volcanic activity. Increased volcanic activity results in higher greenhouse gas emissions, which would further warm up the climate and pose a threat to local communities. However, the link between sea level fluctuations and volcanic activity has never been well demonstrated. If it could be demonstrated that a link exists, methods can be developed to better forecast volcanic hazards. This would also improve our knowledge and understanding of the earth-climate system, which could result in the reduction of uncertainty in various global climate models. Therefore, the aim of this study is to examine the relation between sea level fluctuations and volcanic activity on volcanic islands. Two time series are used over which the total change in sea level is large. The first time series covers a complete glacial-interglacial cycle (0-120ka), while the latter covers the past glacial (0-32.5ka). In this research, a new method is proposed to examine the linkage based on the extent of radiometrically dated volcanogenic flows and the change in sea level. Both volcanogenic time series are digitized in ArcGIS Pro 2.2 and categorized into 12 timesteps of 10.000 years and 13 timesteps of 2.500 years. The method was tested and analysed on four study areas; Pico island in the Azores, Tenerife in the Canaries and the Aeolian islands of Lipari and Vulcano. A Pearson correlation test resulted in no statistically significant association between the time series of the absolute change in sea level and the volcanogenic flow reconstructions, with the exception of Vulcano. The statistics show a R 2of 0.73 with a p-value of 0.0071 between the amount of absolute sea level change and the volcanogenic flow reconstructions over a time series of 120.000 with 10.000 years time intervals. Therefore, these results suggest that a link could exist and may be related to the amount of fluctuation in sea level rather than the direction of change.
Table of Contents
Abstract
2
1. Introduction
………... 5
1.2 Societal relevance………... 6
1.3 Research aim………... 6
1.4 Research questions………. 7
1.5 Theoretical framework……… 7
1.5.1 Area-based vs volume-based method……….. 8
1.5.2 Study area (geological background)………. 9
2. Methods
……… 11
2.1 Preprocessing raw data………... 12
2.2 Development of the volcanogenic flow time series……….. 12
2.3 Spatial interpolation………....……….. 13
2.4 Temporal interpolation……….. 14
2.5 Calculation of the sea level change………...……….... 14
2.6 Accuracy assessment……….. 15
2.6.1 Spatial accuracy……….………... 15
2.6.2 Temporal accuracy……….………... 16
3. Results
………...……….. 17
3.1 Main result……….…..17
3.1.1 Volcanogenic area distribution…….……….17
3.1.2 Correlation between the absolute change in sea level and the
volcanogenic flow reconstructions……….….19
3.1.3 Correlation between the change in sea level and the
volcanogenic flow reconstructions…………...………... 20
3.2 Accuracy assessment………... 20
3.2.1 Area distribution of the volcanogenic flow reconstructions…... 20
3.2.2 Spatial interpolation………... 22
3.2.3 Temporal interpolation……….. 23
4. Discussion
……….. 24
4.1 Interpretation of the results……….. 24
4.2 Discussion on the methodology………... 25
4.3 Future research………. 26
5. Conclusion
……….. 27
6. References
……….. 28
Appendix 1
Data tables
31
Appendix 2
Accuracy assessment & correlation statistics
38
1. Introduction
During the lifetime of the Earth, volcanism has always had a dominant role in the creation and formation of the Earth's surface. Volcanism can be defined as a process through which the Earth transfers heat from its interior to the surface (Wilson, 2014). This heat is mainly transferred through the crust in the form of molten rock, called magma. When the magma flows through the crust to the surface, the lava cools and solidifies. Over time, the accumulation of volcanic material on the seafloor can even contribute to the the formation of volcanic islands.
The influence of volcanism is measured all over Earth. Wilson (2014) estimated that at least three quarters of the surface rocks on Earth are directly formed through volcanic activity. But that is not all. Volcanic activity influences the climate system (Hammer et al., 1980). During the eruption of a volcano, greenhouse gases and aerosols are released into the atmosphere and change the atmospheric concentrations. Therefore, volcanic activity directly affects the solar radiation balance of the Earth (Myhre et al. 2013). The influence of an eruption on the climate system depends on the type and intensity of that eruption (Wilson, 2014). The effect of a sudden increase of the aerosol concentrations in the atmosphere short after an eruption is observed through a decrease of global mean temperatures. On longer timescales, the warming effect of volcanic greenhouse gases is paramount over the effect of aerosols on the climate (Myhre et al. 2013).
However, there are signs that the reverse could also be true meaning that the climate could influence volcanic activity as well. In the past million years, the glacial cycles have characterized the climate (Sigman et al., 2010). According to the Milankovitch theory, the astronomical cycles have driven temperature changes resulting in the alternation of glacials with interglacials. Over the duration of a complete glacial cycle, sea levels change by more than a hundred meter (Figure 1). Therefore the
Figure 1. Smoothed time series of the relative sea level (m) over the past 120.000 years (Grant et al., 2012). The
brown line represent the extracted time series from Lambeck et al. (2014), while the red line represent the time series that was extracted from Grant et al. (2012).
hydrostatic pressure of the water column on the ocean plate changes. The rate of pressure change could have an effect on the pressure within the magma chamber of volcanic islands, which may trigger volcanic activity. Some studies have shown indications of a relation on relative short timescales (Dzurisin, 1980; Stothers, 1989; Girona et al. 2018). For instance, Girona et al. (2018) showed a statistical relation between fortnightly tides and a eruption of the Ruapehu volcano. However a conclusive answer on how and to what extent sea level fluctuations have an effect on volcanic activity can not be given yet.
1.2 Societal relevance
According to Gehrels & Woodword (2013), the rate of sea level rise has increased since the start of the 20th century. This trend is likely to continue over the coming centuries (Church et al., 2013). As a result, there is a potential for more volcanic activity. Consequently, the risk for humans and infrastructure on volcanic islands as well as in the surrounding communities would increase (Zuccaro et al., 2015). If it could be demonstrated that a link exists between volcanic activity and sea level fluctuations, methods can be developed to better forecast volcanic hazards (Girona et al., 2018). Besides, more volcanic activity means more greenhouse gases emitted into the atmosphere, which would contribute to the greenhouse effect on the long term (Hammer et al., 1980; Jihong, 2010). Furthermore, by analysing the influence of sea level fluctuations on volcanic activity, our knowledge and understanding of the earth-climate system will improve. This could reduce uncertainty in various global climate models and improve the projections of the Coupled Model Intercomparison Project (Jihong, 2010; Flato et al., 2013). As a result, this allows us to better be able to implement climate change solutions.
1.3 Research aim
Earlier studies suggested a potential link between sea level fluctuations and volcanic activity (Dzurisin, 1980; Stothers, 1989; Girona et al., 2018). However, previous studies remain inconclusive about how this link precisely works. Moreover, the few available studies mostly report about the relation on short timescales based on observational data and overlook the longer timescales over which sea level change is largest. Dzurisin (1980) reported that the likelihood of an influence of fortnightly tides on the eruption patterns of the Kilauea volcano on Hawaii is about 90%, while the eruptions of the Mauna Loa are randomly distributed. Stothers (1989) carried out a time series analysis for about the past 500 years, but only found 2 statistically weak links between the frequency of volcanic eruptions and cycles of solar activity. A more recent study (Girona et al., 2018) showed that the Ruapehu volcano in New Zealand was statistically sensitive to fortnightly tides prior to the 2007 eruption. In recent decades, the amount of long-term data on volcanic activity for some volcanic islands has increased incredibly. Currently, the possibility arises to gather enough data points of volcanic activity in order to create a timeseries over longer timescales.
Therefore, the aim of this study is to examine the potential relation between sea level fluctuations and volcanic activity for two volcanic islands over the last 120.000 year and for two
1.4 Research questions
In order to examine the potential relation between sea level fluctuations and volcanic activity the volcanic activity has to be correlated to the sea level change over the time series used. Therefore, the following main question needs answering:
To what extent is the volcanogenic area related to the change in sea level?
In order to provide an answer to the main question, a time series of the volcanogenic area has to be established for each island. Due to the distinction in the time series between the islands, the next sub questions have been formulated.
a. How much volcanogenic area was produced every 10.000 years on Lipari and Vulcano?
b. How much volcanogenic area was produced every 2.500 years on Tenerife and Pico?
In order to correlate the volcanogenic area to the changes in sea level, a distinction between the islands is also made in the time series of the change in sea level.
c. What is the change in sea level per timestep of 10.000 years? d. What is the change in sea level per timestep of 2.500 years?
1.5 Theoretical framework
Volcanic activity is defined as all volcanic material that has been erupted from the vents of a volcanic island. In this research, volcanic activity is measured through the area of volcanogenic flows. In this context, volcanogenic flows can also be described in a similar matter as volcanic activity. Therefore, volcanogenic flows not only include lava flows, but also pyroclastics and pumiceous successions. In general, volcanic material that is erupted through the submarine vents are excluded. It would be preferable to include the volcanogenic flows from submarine vents, because this would make the areal reconstruction of the volcanogenic flows through time more complete. However, due to the inconsistent availability of usable data and the time limit of the research, submarine flows were not covered for Pico and Tenerife. However, submarine flows were included in the determination of the volcanogenic flow area of Lipari and Vulcano as these flows were already part of the
1.5.1 Area-based vs volume-based method
In order to research the link between sea level fluctuations and volcanic activity, the volcanic activity is best quantified in terms of volume erupted material. However, volume estimates of volcanogenic flows remain scarce or are absent for the areas that are studied in this research. In order to be able to use a sufficient amount of data points, the area extent is used. The area extent of volcanogenic flows is better researched and more data is available for the study areas that are studied in this research. Nevertheless, new methods for lava flow volume estimations have recently been developed and tested on different study areas. Figure 2 shows the degree of linear dependency between volume and area of several lava flows from the Tolbachik volcanic complex in Kamchatka and El Reventador in Ecuador (Kubanek et al., 2017; Arnold et al., 2019). The positive correlations between area and volume of these lava flows is quite strong and significant at the 95% significance level. Even though there is still a certain amount of inaccuracy accompanied with the TanDEM-X DEM method for lava flow volume estimation, it is currently the best method available (Kubanek et al., 2017).
With this in mind, the area extent of volcanogenic flows is assumed to also have a linear correlation with the volume in the study areas researched in this work. In other words, the area is proportionally equal to the volume with a constant depth. In theory, such a correlation could also be explained by the rheology of volcanogenic flows and the shape of a volcano. Rheology explains the way that matter flows with both fluid and solid characteristics. The rheology of a flow as well as the topography of the volcano determines the distribution of a flow over the volcanoes slopes. Because a volcano could, in theory, be seen as a perfect cone, the surface of a slope could be imagined as a smoothened surface with a surface roughness of zero. In this case, the volcanogenic flow would be equally distributed over the volcanoes slope. As a result, the area of a volcanogenic flow would have a perfect correlations with the volume.
Figure 2. Graphs of the correlations between the area and volume of several lava flows. A: Correlation between
lava flow data of the Tolbachik volcanic complex in Kamchatka, Russia (Kubanek et al., 2017). B: Correlation between lava flow data of the El Reventador, Ecuador (Arnold et al., 2019). The black line shows the trendline., while the gray shade shows the 95% confidence interval.
1.5.2 Study Areas (geological background)
The linkage between sea level fluctuations and volcanic activity has been examined for four volcanic islands (Figure 3). These islands include Tenerife in the Canarian archipelagos, Pico in the Azores and the volcanic island of Lipari and Vulcano in the Aeolian archipelagos. The areas were selected based on the geological settings of the volcanic islands and the amount of available
radiometrically dated data that corresponds to the time series intervals. The study areas are similar in the sense that it are volcanic islands which have been surrounded by the sea during the past 120.000 years. Therefore, the magma chamber is thought to be continuously influenced by the hydrostatic pressure resulting from the oceanic water column. Theoretically, changes in the hydrostatic pressure regime surrounding volcanic island systems could result in changes in volcanic activity for several reasons. Firstly, the omnipresent pressure regime around the cone is in equilibrium with the pressure exerted from the magma chamber. Changes in the omnipresent pressure regime changes the
equilibrium. Because the omnipresent pressure regime consists partly of the hydrostatic pressure derived from the oceanic water column, pressure changes of the water column changes the
omnipresent pressure regime around the cone. Secondly, the oceanic crust is relatively thin. Therefore, the capacity of the Earth’s crust to reduce the influence of sea level changes on the magma fluxes underneath the crust could be less for volcanic islands compared to volcanoes near or on the
continents. Lastly, the basalts have a low viscosity state, which could be easier influenced by pressure changes than high viscosity lavas.
Furthermore, the difference of the geological settings between the study areas could help to explain the results of this research. The island of Tenerife originates from a hotspot, while the islands of Lipari and Vulcano have their genesis from the plate tectonic boundaries. Lipari and Vulcano are the result of the collision between the African and Eurasian Plate. Moreover, Pico island has been formed through the complex interaction between the proximity of an inferred mantle plume (hotspot) and the triple junction of the North American plate, Eurasian Plate and African plate. The divergence of the Mid-Atlantic Ridge in combination with the presence of an hotspot shaped the Azores Plateau as well as the Canarian archipelagos (Ablay & Marti, 2000; Forni et al., 2013; Nunes, 1999).
Most of the volcanogenic flows, that were erupted during the last glacial cycle, were created by stratovolcanoes (Ablay & Marti, 2000; Massimiliano et al., 2005; Nunes, 1999). Moreover, all volcanic islands have had several periods of volcanic activity during the last 120.000 years, which makes them especially suitable in this research. The main differences as well as similarities are used as an opportunity to compare the linkage between sea level fluctuations in different volcanic island settings. By taking geological differences into account, data can be analysed more realistically. As a result, this research helps to improve our current understanding between sea level fluctuations and volcanic activity.
Figure 3. Locations of the three study areas on the world map (Source: Imagery basemap in ArcGIS Pro 2.2,
2. Methods
In this research, an indirect method for the assessment and mapping of volcanogenic flows is used. The data is obtained from previous studies, which used direct (e.g. field measurements and observational data) as well as indirect methods (satellite records) for their data quantification. A table with all the input data is added to Appendix 1 in table 1. The reconstructions of the volcanogenic flow datasets, interim calculations and visualizations are done in ArcGis Pro 2.2 as well as in ArcGIS desktop 10.6.1. The data analysis is done in ArcGis Pro 2.2 as well as in RStudio. The datatables can be found in Appendix 1. All scripts are given in Appendix 3. Figure 4 shows the workflow of this research. The workflow will be elaborated on in the next sections.
Figure 4. Generalised workflow of the research including four main components of (1) preparation, (2)
development of time series, (3) data analysis and (4) deliverables. Arrows connect the sub-components and give additional information for certain steps.
2.1 Preprocessing raw data
The workflow of the research consists of four main components (Figure 4). The preparation component includes a literature review as well as the preprocessing of the sea level datasets and the geological maps.
Sea level data of the past 32.500 year and 120.000 year were extracted from respectively Lambeck et al. (2014) & Grant et al. (2012). The data was transferred and organised in 2 new .txt-files, which can be found in the data folder. The data was then imported into Rstudio.
Areal data of volcanogenic flows were obtained with the use of the geological maps of Tenerife (Carracedo et al., 2007) and Pico (Nunes, 1999). For Lipari (Forni et al., 2013) and Vulcano (de Astis et al., 2013), the reconstructions of the eruptive epochs were used. The data was then transformed into a .tiff-file and georeferenced in ArcGIS Pro 2.2. A 1st order polynomial
transformation was selected in the georeferencing process of the original rasters. This is done because the original rasters only needed to be scaled and preservation of the shape of the raster deemed to be more important than acquiring the best fit. In the geodatabase, a new feature class was created for the areal reconstruction data for every timestep and named accordingly. Feature classes were also created for the volcanogenic flow area covered by younger material and for individual flows, which are used in the data analysis.
2.2 Development of the volcanogenic flow time series
With the newly created feature classes added to the ArcGIS data frame, new templates were created for every timestep of the volcanogenic flow reconstructions. The volcanogenic flow
reconstructions of Tenerife and Pico were drawn from the geological maps provided by respectively Carracedo et al. (2007) and Nunes (1999). Moreover, the volcanogenic flow reconstructions of Lipari and Vulcano were drawn from the eruptive epochs provided by Forni et al. (2013) and De Astis et al. (2013). During an edit session, the georeferenced geological maps were used to copy the outline of geological formations to the feature classes, resulting in the creation of the reconstructions of the volcanogenic flow area per timestep (Figure 5). The reconstructions were drawn with the use of the ‘freehand’- option in the ‘create feature’-function. This was merely a sketch of the reconstructions. In order to get a more precise digital copy of the outline of the geological formations, the outline had to be adjusted by zooming in on the vertices and manually move the vertices into place. As a result of the size of certain volcanogenic flows, it was not always possible to draw the outline at once. Therefore, several polygons were created, which all cover a part of the outline (Figure 5: step 3). These polygons were later merged into one polygon in order to get the outline of a volcanogenic flow reconstruction. The leftover vertices, which were remnants of the merged polygons, were removed (step 4). The resulting polygon still contained area of older strata. These areas were then drawn as a separate polygon and removed from the main polygon with the ‘cut’-polygon tool in the edit bar in ArcGIS desktop 10.6.1. (step 5). Another ArcGIS version was used, because the ‘split’-tool (similar to the ‘cut’-tool) had some software issues in the latest version of ArcGIS Pro. The ‘trace’-tool was used to easily cut the subset of polygons from the main polygon.
Figure 5. Example of the process to create a polygon of a younger volcanogenic flow. (1) Younger
volcanogenic flow. (2) Older volcanogenic flow. (3) Two polygons of the younger volcanogenic flow that need to be merged. (4) Overlap between the two “younger”- polygons that needs to be removed. (5) Two polygons of the older volcanogenic flow that need to be cut.
The selection procedure of the volcanogenic flow reconstructions was based on several criteria. Firstly, only volcanogenic flows were incorporated. Surface area classified as alluvium, fluvial and colluvium were excluded from the selection. Secondly, only volcanogenic flows
originating from the same volcanic island were included. Therefore, fallout layers from eruptions on other islands are not taken into account. Lastly, Carracedo et al. (2007) documented the age of volcanogenic flows with radiocarbon ages and calibrated radiocarbon ages. This could result, in certain cases, in a situation wherein a volcanogenic flow could be assigned to two timesteps. This has been solved by assigning the calibrated radiocarbon age from Carracedo et al. (2007) to a
volcanogenic flow. Whenever the calibrated radiocarbon ages were not available, the other
radiocarbon ages were used. Estimated ages from K/Ar-dating have only been used in the case of the Volcán del Portillo, because there were no radiocarbon age datings available.
2.3 Spatial interpolation
Due to the burial of older volcanogenic flows by younger volcanogenic flows, the area extent of older flows is often unknown. Some of the older flows have been well preserved at the surface. An explanation could be on the one hand, because these older flows have not been overflown by younger flows. On the other hand, older flows can reappear at the surface as a result of the erosion of younger material. In order to reconstruct the outline of buried volcanogenic flows per timestep, an
interpolation method was developed.
For Tenerife and Pico, this has been done by using the known surface area of the volcanogenic flows, that are dated within a similar time period, as an outline. With the use of literature, geological cross sections and estimations of the flow direction, the maximum known area extent could be determined. Wherever, the known surface area was fragmented due to burial, the outline of the known surface areas were connected to those nearest. Exceptions were made in cases where this outline would overlap even older volcanogenic flows or would not match the flow direction or coastline. In a single case, the same outline was used for the pahoehoe lavas (25-27.5ka) of the
Pico Viejo on Tenerife as was used for the intermediate and felsic (phonolotic) eruptions
(17.5-22.5ka). This decision was made, because even though the known surface area is limited, the volume and area extent of the pahoehoe lavas is estimated to widely exceed those of the intermediate and felsic eruptions.
The volcanogenic flow reconstructions of Lipari and Vulcano did not need interpolation, because the input datasets were already retrieved as reconstructions. Therefore, the outline of the volcanogenic flow reconstructions correspond to the outline of the volcanogenic flows from the Eruptive Epochs of de Astis et al. (2013) and Forni et al. (2013).
2.4 Temporal interpolation
The volcanogenic flows were categorized in periods of volcanic activity. Therefore, the age range of these periods often overlapped the timesteps used in this research. The choice has been made to assume an equal eruptive rate for the period over which a flow was erupted. After the periods of volcanic activity had been digitized and reconstructed, the proportion of overlap between the periods of volcanic activity and the timesteps determined the area assigned to a timestep. For example, when a period of volcanic activity spanned 14-24 k years and the timesteps used were 10k years, then the area from the 10k year period of volcanic activity was proportionally divided. In this case, 60% of the area was assigned to the 10-20 ky timestep, while 40% of the area was assigned to the 20-30 ky timestep. Even though, temporal interpolation is required which could make the data less realistic, there were several reasons to do it. Firstly, some volcanogenic flows were erupted over a longer period than the extent of the timesteps. Secondly, it is a matter of including all the available information or excluding parts. In this research, all the available information is included, because it gives a more complete perspective on the total erupted material and also shows the current state of this research field. Lastly, the temporal interpolation was done in order to provide a framework for future research. Whenever the radiometric dating of volcanogenic flows is improved, the results can be added to this research.
2.5 Calculation of the change in sea level
The smoothed relative sea level data with a 1ky moving Gaussian filter from the red sea (Grant et al., 2012) was used to calculate the change in sea level per 10k years. This was done for Lipari and Vulcano over the period of 120ka years. A more detailed version of the sea level change of the past 32.5k years was used for Tenerife and Pico (Lambeck et al., 2014). Due to a lack of usable data for the reconstruction of volcanogenic flow area from before 32.5ka years, the resolution was increased from 10k to 2.5k years. Therefore the amount of timesteps changed to 13. The data was copied to a .txt-file and imported into Rstudio. The change in sea level time series was created by calculating the difference in sea level between 2 timesteps in meters. For instance, the sea level at 20ka was subtracted from the sea level at 10ka for the time series of Lipari and Vulcano. This was done for every timestep resulting in 12 data points. The same steps were followed in the calculation of the sea level change for Tenerife and Pico. However, the timesteps only lasted 2.5k years resulting in 13 data points.
2.6 Accuracy assessment
In order to give an indication about the accuracy of the interpolation methods, an accuracy assessment has been carried out. The accuracy assessment includes the assessment of the spatial interpolation method as well as the assessment of the temporal interpolation method. Similar to the development of the volcanogenic flow time series, several feature classes were created to draw and store the volcanogenic flow area that is covered by younger material. The accuracy assessment had to be done for Tenerife and Pico due to the large extent of interpolated area both in the temporal as in the spatial dimension. However, a spatial accuracy assessment was not deemed necessary to do for Lipari and Vulcano, because the original data were already reconstructions. The volcanogenic flow
reconstructions correspond to the outline of the volcanogenic flows from the Eruptive Epochs of de Astis et al. (2013) and Forni et al. (2013). However, a temporal accuracy assessment has been carried out for Lipari and Vulcano, because the area calculated for some timesteps had been the result of interpolation.
Furthermore, the accuracy of the input data can be found in table 7 in Appendix 1. In other words, the resolution and the GIS-projections of the geological maps are summarized here.
2.6.1. Spatial accuracy
In order to validate to what extent the volcanogenic flow reconstructions represent the volcanogenic flows at the surface, a spatial accuracy assessment had been carried out. Firstly, the accuracy assessment shows the area extent of a reconstructed volcanogenic flow that is covered by younger material per timestep. In this case, the definition of younger material includes younger volcanic material, colluvium/alluvium material and undifferentiated material. Therefore, this accuracy assessment shows all the spatially interpolated area used in the reconstructions (Figure 9). Secondly, the accuracy assessment shows a classification of the percentages of the volcanogenic flow
reconstructions which are covered by younger material. Therefore, the accuracy of the volcanogenic flow reconstructions is easily represented. Table 1 gives an overview of the classification that is used to assess the accuracy of the spatial interpolation method. The accuracy classes range from “Very high” to “Very low” degree of accuracy. For instance, a “Very low” degree of accuracy is assigned to a volcanogenic flow reconstruction if 80% to 100% of the area needed to be interpolated and thus is covered by younger material.
Table 1. Accuracy classification of the spatial interpolation method.
Spatial interpolation
Class Degree of accuracy Percentage of coverage
1 Very high 0-5
2 High 5-25
3 Moderate 25-50
4 Low 50-75
2.6.2. Temporal accuracy
In the accuracy assessment of the temporal interpolation method, the accuracy of the timestep areal data is based on the amount of volcanogenic flow reconstructions that were included in the calculation of the area of that timestep. In other words, the accuracy assessment on the temporal interpolation method shows the amount of volcanogenic flow reconstructions that were used to calculate the area of a timestep. However, the assessment does not differentiate between the includement of 1 full volcanogenic flow reconstruction per timestep and a proportion of a
volcanogenic flow reconstruction per timestep. For example, when a timestep includes the area of only 1 volcanogenic flow of 9000 km2, the degree of accuracy is equal to the accuracy of a timestep that includes ⅓ of a volcanogenic flow. Table 2 gives an overview of the classification that is used to assess the accuracy of the temporal interpolation method. The accuracy classes range from “Very high” to “Very low” degree of accuracy. For instance, a ‘Very high’ degree of accuracy is assigned to a timestep if the area of that timestep is calculated with only 1 volcanogenic flow reconstruction. Therefore, no temporal interpolation has been carried out for this specific timestep and the final area corresponds to a single radiometric age dating. A “Very high” degree of accuracy is also assigned to a timestep in case no temporal interpolation has been carried out.
Table 2. Accuracy classification of the temporal interpolation method
Temporal interpolation
Class Degree of accuracy Amount of volcanogenic flow reconstructions included in timestep 1 Very high 1 2 High 2 3 Moderate 3 4 Low 4 5 Very low 5
3. Results
3.1 Main results
3.1.1. Volcanogenic area distribution
The data analysis consists of several components. Firstly, the area changes resulting from the volcanogenic flow additions over the timesteps is shown and discussed. Secondly, the relation is analysed between the change in sea level and the area of volcanogenic flow reconstructions. Furthermore, an accuracy assessment is presented. A general overview is given by showing the area distribution of the volcanogenic flow reconstructions and spatially interpolated area for Tenerife and Pico. Moreover, the amount of spatially interpolated area is categorized and put in a table in order to show the accuracy of the reconstruction method as well as to show current data gaps for the
categorization of volcanogenic flows over the used timesteps. Lastly, the amount of volcanogenic flow reconstructions used to calculate the final area of a timestep, is categorized in order to show the accuracy of the temporal interpolation method.
Furthermore, additional data is organised as follows. Polygons of the volcanogenic area reconstructions, individual volcanogenic flows as well as of the interpolated volcanogenic area can be found in the geodatabase of the to their related data folder. Data tables of the volcanogenic area reconstructions, individual volcanogenic flows as well as of the interpolated volcanogenic area can be found in Appendix 1. The R-script used in the data analysis is added to Appendix 3. Figure 6 displays the volcanogenic area distribution over the timesteps per island.
Figure 6. Volcanogenic area distribution over the timesteps per island. The timesteps of Lipari and Vulcano are per 10.000 years, while the timesteps of Tenerife and Pico are per 2.500 years. Y-axis values differ.
3.1.2 Correlations between the absolute change in sea level and the volcanogenic flow reconstructions.
Figure 7 shows scatterplots of the correlation between the absolute change in sea level and the reconstructed volcanogenic area based on the Pearson correlation test. These results are highlighted in this section.
The 95% confidence interval is shown in light gray. The correlation-coefficient and p-values are added to the graphs. The Pearson correlation test gives the following results: Lipari: -0.15,
Vulcano: 0.73, Tenerife: -0.53, Pico: -0.43. For Vulcano, these results would imply that with a normal distribution there is a positive relation of about 73% between the absolute change in sea level and the area of volcanogenic reconstructions. In other words, about 73% of the reconstructed volcanogenic area time series follows the same trend as the absolute change in sea level. It is noticeable that the correlation is statistically significant at the 95% significance level for Vulcano. However, the p-values of Lipari, Tenerife and Pico are all above 0.05. These results are not statistically significant at the 95% significance level. This was already expected, because there are too few data points to perform a statistically significant correlation test. Nevertheless, it is noticeable that Tenerife shows a statistically significant correlation at the 90% significance level.
Figure 7. Scatterplots of the correlation between the absolute change in sea level and volcanogenic area. The 95% confidence interval is shown in light grew. The Pearson correlation-coefficient and p-values are added to the graphs. Notice the difference in axis-values between the graphs.
3.1.3 Correlation between the change in sea level and the volcanogenic flow reconstructions.
Figure 11-14 in Appendix 2 show the correlations between the change in sea level and the volcanogenic area reconstructions. The correlation coefficients are all weak and there seems to be no correlation. The p-values are all far from statistically significant at the 95% significance level.
3.2 Accuracy assessment
3.2.1 Area distribution of the volcanogenic flow reconstructions
Figure 8 shows the volcanogenic area per polygon that is covered by younger material. The x-axis represent the time period in which the volcanogenic flows have been deposited. The most recent time period is not covered by younger material as it is the youngest time period. Figure 8 shows partly the accuracy assessment of Tenerife. Most of the volcanogenic area is interpolated in time period 25-27.5ka. The time period includes only the Pahoehoe lavas of Pico Viejo. The area extent of the Pahoehoe lavas were made through a specific interpolation method, namely to follow the outline of the overlying time period (17.5-22.5ka). Therefore, the similarity in covered material between both time periods was expected.
Figure 8. Area distribution of the volcanogenic flow reconstructions of Tenerife and Pico. The total area of the
reconstructed volcanogenic polygons is shown in brown. The area that is covered by ‘younger’ material is given in green. The x-axis represents the time period that a volcanogenic flow reconstruction was dated at.
3.2.2. Spatial interpolation
Tables 3 and 4 show the results of the accuracy assessment for the spatial interpolation method of Pico and Tenerife. Both tables were derived from figures 9 & 10 in Appendix 2. Noticeable is that older volcanogenic flow reconstructions are generally less accurate and more spatial
interpolation needed to be done.
Table 3. Spatial accuracy assessment of the volcanogenic flow reconstruction for Pico.
Volcanogenic flow reconstruction (ka)
0 - 2 1,5 - 5 2 - 10 5 - 10 5 - 40 10 - 30 10 - 50 30 - 250
Class 1 3 2 1 4 2 4 2
Accuracy Very high Moderate High Very high Low High Low High
Table 4. Spatial accuracy assessment of the volcanogenic flow reconstruction for Tenerife.
Volcanogenic flow reconstructio n (ka) 0 - 2,5 2,5 - 5 5 - 7,5 5 - 10 7,5-10 0-10 12,5-15 15-17,5 17,5-22,5 25-27,5 30-37,5 Class 1 3 1 1 2 2 3 2 4 5 2 Accuracy Very high Moderate Very high Very high
3.2.3. Temporal interpolation
Tables 5 & 6 show the accuracy assessment of the temporal interpolation method. The results show a relatively high accuracy for Lipari and Vulcano, while the accuracy for Pico is relatively low. The temporal interpolation for Tenerife is less accurate in the four recent timesteps, while older timesteps required no temporal interpolation.
Table 5. Temporal accuracy assessment of the timesteps used in the time series for Tenerife and Pico. Mod = Moderate. Timestep (ka) 0 - 2,5 2,5 - 5 5 - 7,5 7,5 - 10 10-1 2,5 12,5-15 15-17, 5 17,5-20 20-22,5 22,5-25 25-27,5 27,5-30 30-32,5 Tenerife Class 2 2 3 3 1 1 1 1 1 1 1 1 1
Accuracy High High Mod Mod Very High Very High Very High Very High Very High Very High Very High Very High Very High Pico Class 3 2 3 3 3 3 3 3 3 3 3 3 3
Accuracy Mod High Mod Mod Mod Mod Mod Mod Mod Mod Mod Mod Mod
Table 6. Temporal accuracy assessment of the timesteps used in the time series for Lipari and Vulcano. Timestep (ka) 0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 Lipari Class 1 1 1 1 1 1 1 1 1 1 1 1 Accuracy Very high Very high Very high Very high Very high Very high Very high Very high Very high Very high
Very high Very high
Vulcano Class 1 1 1 1 3 2 1 1 1 1 1 1 Accuracy Very high Very high Very high Very high
Moderate High Very high Very high Very high Very high
4. Discussion
4.1 Interpretation of the results
Results should always be carefully interpreted. Often it is the case that assumptions are being made, resulting in the establishment of causal relations, where no causal relation exists. Therefore, it is important to analyse the results from the perspective of the research. The volcanogenic area that is calculated in this research, has been reconstructed and is therefore an estimation of the area of volcanogenic flows. The volcanogenic area used in this analysis is by definition not the real
volcanogenic area that had been erupted during a timestep. However, the reconstructed volcanogenic area should provide insights into the distribution of erupted volcanic material over the past glacial cycle for 4 different volcanic islands. Besides, the research a new method for the volcanogenic area reconstructions of these islands and improvements to be made.
In theory, changes in hydrostatic pressure of the sea level could result in the omnipresent pressure changes on the magma chamber of volcanic islands. However, it is unclear to what extent and how the change in hydrostatic pressure exactly affect the pressure balance in the magma chamber of a volcanic island. Therefore, it is more likely that a range of stress-related factors influence the activity of volcanoes (McGuire et al., 1997).
In this research, two hypotheses were tested to understand to what extent the change in sea level is related to volcanic activity. Firstly, the volcanic activity is related to absolute values for the change in sea level. This hypothesis assumes that it is irrelevant if the hydrostatic pressure, that is added to the omnipresent pressure regime of a volcanic island, increases or decreases and that only the amount of change influences volcanic activity. The second hypothesis states that volcanic activity is related to the direction of hydrostatic pressure changes. Therefore, it is assumed that it does matter if the hydrostatic pressure surrounding a volcanic island increases or decreases. This would mean that an increase of sea level would have the opposite effect on volcanic activity in comparison to a decrease in sea level.
The results of this research suggest the following. Firstly, the null hypothesis that volcanic activity is not related to the direction of hydrostatic pressure changes (sea level rise as well as decline) is accepted for the volcanic islands covered in this research. There is no clear association found between the change in sea level and the volcanogenic area time series. Besides, the p-values are all not statistically significant. Furthermore, the null hypothesis that volcanic activity is not related to the amount of change in hydrostatic pressure is accepted for Lipari, Tenerife and Pico at the 95% significance level. Therefore, there is no clear association found between the absolute change in sea level and the volcanogenic area time series. Interestingly, this null hypothesis is rejected for Vulcano.
In general, when the fluctuations in sea level were large, more volcanogenic area had been erupted at Vulcano. In the context of the geological setting of Vulcano, a stronger significant
correlation would be expected for Lipari. This is due to the influence of the shared tectonic regime on the magma chambers between the volcanic islands. The different outcomes may be related to the independent magma sources of Vulcano and Lipari (De Astis et al., 2013; Lucchi et al., 2010; Forni et
Another point to be made, are the reverse trends seen in the absolute sea level change
correlation with the volcanogenic reconstructions for Tenerife and Pico. Even though these trends are not statistically significant and have weak correlations, the trends do show a reverse trend. This may be explained through the accuracy assessment. The accuracy of the spatial interpolation method as well as the temporal interpolation method may have resulted in the distortion of the time series.
Nevertheless, the results for Vulcano give an indication that there could be a link between the absolute change in sea level change and volcanic activity. However, an conclusive answer can only be given through further research.
4.2 Discussion on the methodology
There were several reasons to increase the resolution of the timesteps to 2500 years instead of 10.000 years for Tenerife and Pico. Firstly, the availability of areal extent data of volcanogenic flows from before 32.500 ka BP is limited. This is because most of the older flows are often overflown by younger flows. As a result, the most information is available for volcanogenic flows that lay at the surface. This is probably because these flows are better accessible. Secondly, even though radiometric age data of older flows are available through boreholes and galleries, the reconstruction of the areal extent of older flows has not yet been done. For these reasons, the resolution of Tenerife and Pico has been increased to timesteps of 2.500 years over a 32.500k time serie in order to still obtain sufficient data points (increase from 3 to 13).
There were several factors that limited the accuracy of the reconstruction data. The following paragraphs highlight these factors.
Even though the geochronology of volcanogenic flows is often well documented, radiometric age data remains scarce. This complicates the grouping of volcanogenic flows in timesteps. Especially when the resolution of the timesteps is increased, because more areal data is needed. Another factor was the division of volcanogenic flows in periods. Often periods overlap the timesteps, which
therefore required a consistent method to coup with. Favourable would be to use radiometric data that fall within each timestep. Also, the available radiometric age data is an approximation and therefore leaves room for uncertainty. However, only in some individual cases this formed a problem. In most cases the age data had a low enough uncertainty to fall within the timesteps.
The resolution of the original geological maps could be improved. Often the vertices on the outline of the reconstruction data had a range of approximately 10m to which it could be assigned to. The vertices were usually drawn at about the middle of this range. This was lesser of a problem for Tenerife as there were two types of geological maps (Carracedo et al., 2007). These maps could be compared in order to draw a more exact outline. In general, the resolution could make the
reconstructions more accurate, but was sufficient enough to draw the main shapes of the geological formations/reconstructions. However, the exact outline of the geological formations remains a matter of interpretation and can only be as accurate as the resolution of the input data allows.
The use of the eruptive epochs provided by Forni et al. (2013) for the volcanogenic flow reconstructions of Lipari and Vulcano adds an additional uncertainty factor to the reconstructed area extent. The eruptive epochs were made in the form of a sketch and therefore are biased to the
interpretation of its authors. However, the eruptive epoch sketches were created by an geology expert with local knowledge and based on the geochronology of the volcanic strata, boreholes and
accurate (and most likely more) as the reconstructions created in this research. Nevertheless, with limited data (re)sources, volcanogenic reconstructions will remain to be open for interpretation. Still, the eruptive epochs are very useful in this type of research.
On the contrary, there were several factors that also reduced the limited accuracy of the reconstruction data. The following section highlight these factors.
Firstly, the methodological approach to use timesteps of 2.500 and 10.000 years reduces a large amount of uncertainty. On the one hand, because the uncertainty that is accompanied with the radiometric dating of volcanogenic flows is reduced. The period in which a volcanogenic flow could be dated becomes larger and therefore the approximated age dating is more likely to fall within a timestep. For example, a volcanogenic flow is dated at an average value of 4200 (+-150) years. The uncertainty of 150 years has, in this case, become irrelevant as a result of the categorization of the flow in timesteps of 10.000 years.
Secondly, spatial undersampling is tackled. Most volcanogenic flows are only dated with about 1 sample at most. The age of a considerable amount of flows has only been estimated through the geochronology of the surrounding strata (Carracedo et al., 2007). By categorizing the
volcanogenic flows in longer lasting timesteps (e.g. 10.000 years), an approximation of the age is often sufficient enough to be useful in the time series analysis.
4.3 Further research
There are several factors to consider for future research. First of all, research could focus on the reconstructions of volcanogenic flows from volcanic islands. That is, because for this type of research, the best datasets are obtained by using areal reconstructions of eruptive phases of volcanic islands. These datasets are valuable, because they also include all the area of volcanogenic flows before weathering processes had taken place as well as reconstructions of submarine flows. With the use of borehole data, chemical analyses etc. more realistic reconstructions can be made. Therefore, future research should focus on detailed reconstructions of the area of volcanogenic flows in the period in which they were erupted.
Furthermore, the reconstruction dataset can be widened as well as deepened. For instance, the best volcanic island would be one for which several glacial cycles of volcanic activity could be reconstructed. Unfortunately, it is not easy to do precise reconstructions of older volcanic activity due to erosion, burying and deformation. The further back in time the harder it gets to get reliable data. By gathering volcanic activity data from different volcanic islands over the past glacial cycle, a larger dataset can be established in order to research the link between the rate of sea level change and volcanic activity. This will improve the datasets required for research on the link between sea level fluctuations and volcanic activity.
Moreover, the sea level data used in this research was generated in order to show a global average (Lambeck et al., 2014; Grant et al., 2012). However, the exact change in sea level is also dependent on the location on Earth. In other words, transformation of the sea level data to local values would give different results.
Conclusion
According to the strong relation between the area and volume of volcanogenic flows, it seems reasonable to research the erupted volcanic material through the volcanogenic area extent. However, considering the geological differences between the study areas, the few data points and the uncertainty of the reconstructions, it is hard to assign conclusive causal relations to the research. Nevertheless, the results give us some new insights.
For Lipari, Vulcano, Tenerife and Pico, the null hypothesis that volcanic activity is not related to the direction of hydrostatic pressure changes is accepted. There is no statistical significant
association found between the time series of the change in sea level and the volcanogenic flow reconstructions.
For Lipari, Tenerife and Pico, the null hypothesis that volcanic activity is not related to the amount of change in hydrostatic pressure is accepted. There is no statistical significant association found between the time series of the absolute change in sea level and the volcanogenic flow reconstructions. Interestingly, this null hypothesis is rejected for Vulcano with a correlation coefficient of 0.73 and a p-value of 0.0071.
Despite the uncertainties currently arising from the lack of geological data, the volcanogenic reconstructions give an estimation of the erupted volcanic material through time. Therefore, the volcanogenic time series made in this research are representations of our current geological knowledge of the four volcanic islands combined with the applied reconstruction method. Even though no definitive conclusions can be ascribed to the research with the amount of uncertainty involved, this research does explore the current limitations and possible solutions to research the link between sea level fluctuations and volcanic activity. Besides, there is a clear potential for
improvement, as the accuracy assessment of the spatial and temporal interpolation methods show. Therefore, this research shows the current state of the research field and opens up potential future research directions. In order to prove to what extent and how the sea level fluctuations and volcanic activity are related, more geological records are needed as well as more reconstruction attempts. This could be done by extending the time series per island further into the past or by including more volcanic islands. Nevertheless, this research adds indications to existing research that there could be a relation.
Literature
Ablay GJ, Martı́ J, 2000. Stratigraphy, structure, and volcanic evolution of the Pico Teide–Pico Viejo formation, Tenerife, Canary Islands. Journal of Volcanology and Geothermal Research, 103(1–4), pp. 175-208.
Arnold DWD, Biggs J, Dietterich HR, Vallejo Vargas S, Wadge G, Mothes P, 2019. Lava flow morphology at an erupting andesitic stratovolcano: A satellite perspective on El Reventador, Ecuador, Journal of Volcanology and Geothermal Research, 372, pp. 34-47.
Carracedo J, Badiola E, Guillou H, Paterne M, Scaillet S, Torrado F, Paris R, Fra Paleo U, Machín A, 2007. Eruptive and structural history of Teide Volcano and rift zones of Tenerife, Canary Islands. Geological Society of America Bulletin. 119(9-10). pp. 1027-1051.
Church JA, Clark PU, Cazenave A, Gregory JM, Jevrejeva S, Levermann A, Merrifield MA, Milne GA, Nerem RS, Nunn PD, Payne AJ, Pfeffer WT, Stammer D, Unnikrishnan AS, 2013. Sea Level Change. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Stocker TF, Qin D, Plattner GK, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds.), Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA. pp. 1191.
De Astis G, Lucchi F, Dellino P, Volpe L, Tranne C, Frezzotti ML, Peccerillo A, 2013. Geology, volcanic history and petrology of Vulcano (central Aeolian archipelago). In: Lucchi F, Peccerillo A, Keller J, Tranne CA, Rossi PL (eds.), The Aeolian Islands volcanoes. Geological Society, London,37(1), pp. 281-349.
Dzurisin D, 1980. Influence of fortnightly earth tides at Kilauea volcano, Hawaii. Geophysical
Research Letters, 7, pp. 925‑928.
Flato G, Marotzke J, Abiodun B, Braconnot P, Chou SC, Collins W, Cox P, Driouech F, Emori S, Eyring V, Forest C, Gleckler P, Guilyardi E, Jakob C, Kattsov V, Reason C,
Rummukainen M, 2013. Evaluation of Climate Models. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Stocker TF, Qin D, Plattner GK, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds.), Cambridge University
Press, Cambridge, United Kingdom and New York, NY, USA. pp. 746-749.
Forni F, Lucchi F, Peccerillo A, Tranne C, Rossi P, Frezzotti ML, 2013. Stratigraphy and geological evolution of the Lipari volcanic complex (central Aeolian archipelago). In: Lucchi F, Peccerillo A, Keller J, Tranne CA, Rossi PL, (eds.), The Aeolian Islands volcanoes. Geological Society London Memoirs, 37(1), pp. 213-279.
Gehrels R, Woodworth PL, 2013. When did modern rates of sea-level rise start? Global and Planetary Change, 100, pp. 263-277.
Girona T, Huber C, Caudron C, 2018. Sensitivity to lunar cycles prior to the 2007 eruption of Ruapehu volcano. Scientific Reports, 8, pp. 1-9.
Grant KM, Rohling EJ, Bar-Matthews M, Ayalon A, Medina-Elizalde M, Ramsey CB, Satow C, Roberts AP, 2012. Rapid coupling between ice volume and polar temperature over the past 150,000 years. Nature, 491, pp. 744-747.
Hammer CU, Clausen HB, Dansgaard W, 1980. Greenland ice sheet evidence of post-glacial volcanism and its climatic impact. Nature, 288, pp. 230-235.
Jihong CD, 2010. Volcanoes and climate. Wiley Interdisciplinary Reviews: Climate Change, 1, pp. 824 - 839.
Kubanek J, Westerhaus M, Heck B, 2017. TanDEM-X time series analysis reveals lava flow volume and effusion rates of the 2012–2013 Tolbachik, Kamchatka fissure eruption. Journal
of Geophysical Research: Solid Earth, 122, pp. 7754– 7774.
Lambeck K, Rouby H, Purcell A, Sun Y, Sambridge M, 2014. Sea level and global ice volumes from the Last Glacial Maximum to the Holocene. Proceedings of the
National Academy of Sciences of the United States of America,111(43), pp. 15296–15303.
Lucchi F, Tranne CA, Rossi PL, 2010. Stratigraphic approach to geological mapping of the late Quaternary volcanic island of Lipari (Aeolian Archipelago, southern Italy). In: Groppelli G, Viereck-Goette L, (eds.), Stratigraphy and Geology of Volcanic Areas.
Geological Society of America, Boulder, Special papers, 464, pp. 1–32.
Massimiliano F, Karátson D, Mazzuoli R, Pareschi MT, Ventura G, 2005. Volcanic
geomorphology and tectonics of the Aeolian Archipelago (Southern Italy) based on integrated DEM data. Bulletin of Volcanology, 68, pp. 57-170.
Mcguire B, Howarth RJ, Firth CR, Solow A, Pullen A, Saunders S, Stewart I, Vita-Finzi C, 1997. Correlation between rate of sea level and frequency of explosive volcanism in the Mediterranean. Nature, 389, pp. 473-476.
Myhre G, Shindell D, Bréon FM, Collins W, Fuglestvedt J, Huang J, Koch D, Lamarque JF, Lee D, Mendoza B, Nakajima T, Robock A, Stephens G, Takemura T, Zhang H, 2013. Anthropogenic and Natural Radiative Forcing. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Stocker TF, Qin D, Plattner GK, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds.), Cambridge University
Nunes JC, 1999. A actividade vulcânica na Ilha do Pico do Plistocénico Superior ao Holocénico: mecanismo eruptivo e hazard vulcânico. PhD thesis, University of
Azores, Ponta Delgada. pp. 357.
Picturetomorrow, 2019. World Physical Map Mercator Large With Of Mountains. Retrieved on 02-10-2019 from http://picturetomorrow.org/map-of-world-with-mountains/world-physical-m ap-mercator-large-with-of-mountains/
Sigman DM, Hain MP, Haug GH, 2010. The polar ocean and glacial cycles in atmospheric CO2 concentration. Nature,466, pp. 47-55.
Stothers RB, 1989. Volcanic eruptions and solar activity. 94(B12), pp. 17371-17381.
Wilson L, 2014. Planetary Volcanism. In: Spohn T, Breuer D, Johnson TV. Encyclopedia of the Solar System (3rd Edition), pp. 101-119.
Zuccaro G, Gregorio D, Baxter P, 2015. Human and Structural Vulnerability to Volcanic Processes, pp. 261-288. In: Papale P, (eds.). Volcanic Hazards, Risks and Disasters. Elsevier, pp. 531.
Appendix 1
Data Tables
Table 7. Metadata of the datasets.
Dataset Source Content Projected
coordinate system
Resolution Type
Sea level time
series Grant et al., 2012 Eustatic sea level - global average relative to 1950
Global 0 - 150k yr. BP. Time series
Sea level time
series Lambeck et al., 2014 Eustatic sea level - global average relative to 1950
Global 0 - 40k yr. BP. Time series
Geology Forni et al., 2013 Geological units
of Lipari, Italy Roma 1940 Gauss Boaga Est 1:10.000 Map
Geology De Astis et al.,
2013 Geological units of Vulcano Roma 1940 Gauss Boaga Est 1:10.000 Map
Geology Carracedo et al.,
2007 Geological units of Tenerife REGCAN95 UTM Zone 28N 1:100.000 Map
Geology Nunes JC, 1999 Geological units
Table 8. Areal data of Lipari per timestep. Geological formation abbreviations correspond to those used by
Forni et al. (2013). Timestep Total area
(km2) Area used in timestep (km2) Area of total (%) EE Geological formation(s)
0-10 13,989 13,989 15,89 9 frr, frr1, lm, sa, sa1, fv, po, vg
10-20
20-30 26,016 26,016 29,54 8 cr, cd1, cd2, cas, sg, gi, gi1, gu, gu1 30-40
40-50 12,361 12,361 14,04 7 fa1, fa2, fa3, pe, pe1,
50-60 60-70 70-80 80-90 1,539 1,539 1,75 6 cc, cc1, cc2 90-100 1,950 1,950 2,21 6 ch1, ch2, ch3 100-110 14,822 14,822 16,83 5 pu, sp, tp, tpa
110-120 17,381 17,381 19,74 4 tc, tr, tra, ma1, ma2, pf, pf1, pf2, sc1, sc2
Table 9. Areal data of Vulcano per timestep. Geological formation abbreviations correspond to those used by De Astis et al. (2013). Timestep (ka) Total area (km2) Area used in timestep (km2) Area of total (%) EE Geological formation(s)
0-10 11,043 11,043 14,49 8 gc1, gc2, pc1, pc2, vu1, vu2, vu3, fv, ca1, ca2, gp1, gp2, gp3, fa, pn1, pn2, pn3, al, gr, ms1, ms2, cr
10-20 11,848 11,848 15,55 7 ro, cl, ml, gr, cm
20-30 4,755 4,755 6,24 6 lm, pb, sl, cf1, cf2, pm1, pm2, mm1, mm2
30-40
60-70 3,353 4,40 5 md, ml, mo
70-80 7,098 7,098 9,31 4 tc, cp, pl, ma1, ma2, ma3, ma4
80-90
90-100 4,650 4,650 6,10 3 sc
100-110 12,302 12,302 16,14 2 cg,pm
110-120 13,113 13,113 17,21 1 & 2 pr, cg,pm
Table 10. Areal data of Tenerife per timestep. Geological formation names correspond to those used by
Carracedo et al. (2007). Timestep (ka) Total area (km2) Area used in timestep (km2) Area covered by younger material (km2) Fraction of covered area (%) Fraction of covered area used in timestep (%) Geological formations (1,2 = additional area from table 5)
0 - 2,5 111,295 141,372 0 0 4,60 2 Historical volcanism (All)
Montaña Reventada Volcán los Hornitos Lavas Negras Roques Blancos Montaña Blanca (All)
2,5 - 5 35,445 65,522 12,825 36,18 29,50 2 Volcán El Ciego
Montaña de Chio El Boquerón
5 - 7,5 31,918 70,868 185 0,58 9,88 12 Montaña Cuevas del Ratón
La Abejera Baja La Abejera Alta
7,5 - 10 30,767 69,717 5,172 16,81 17,20 12 Montaña Liferfe
Montaña Negra - Los Tomillos Volcán del Portillo (vp2) Bocas de Doña Maria
10 - 12,5
12,5 - 15 6,152 6,152 2,379 38,67 38,67 Volcán El Portillo (vp1)
15 - 17,5 12,591 12,591 885 7,03 7,03 Montaña Mostaza
Montaña Arenas Negras Nw-rift - Basic and intermediate eruptions (Pleistocene)
17,5 - 20
132,009
66,005
81,639 61,84
61,84 Pico Viejo - Terminal magmatic and phreatomagmatic eruptions Pico Viejo - intermediate and felsic (phonolotic) eruptions
20 - 22,5 66,005 61,84
25 - 27,5 150,347 150,347 139,302 92,65 92,65 Pico Viejo - Pahoehoe lavas 27,5 - 30 30 - 32,5 112,223 37,408 12,232 10,90
10,90 Ne-rift - Basic and intermediate eruptions (Pleistocene)
32,5 - 35 37,408 10,90
35 - 37,5 37,408 10,90
37,5 - 40
Table 11. Areal data of additional timestep polygons for Tenerife. Geological formation names correspond to those used by Carracedo et al. (2007).
Temporal range of the polygons (ka) Total area (km2) Area covered by younger material (km2) Fraction of covered area (%) Geological formations 5 - 10 17,746 652 3,67 1 Pico Cabras 0 - 10 120,307 25,968 21,58
2 Montaña de Los Corrales - Grupo Montaña Abeque - La Corredera
Montaña Los Conejos - Montaña Negras Montaña de Las Lajas - Montaña de Juan Évora Montaña de La Cruz - Grupo Montaña Cruz Montaña Majúa - Grupo Montaña del Estrecho Mancha Ruana - Las Montañetas Negras Los Gemelos - Montaña Samara Hoya del Abrunco - Montaña Cascajo Montaña La Botija - Volcán de Cuevas Negras Montaña Cruz de Tea - Volcán Negro
Table 12. Areal data of Pico per timestep. Eruptive periods of the volcanic complexes correspond to those used
by Nunes (1999). Colors represent the period that correspond to the area used in the timestep.
Volcanic complex Timestep (ka) Total area (km2) Area used in
timestep (km2)
de Montanha Sao Roque - Piedade Topo - Lajes 0 - 2,5 151,909 196,682 All Superior Intermédia Superior All Superior Intermédia Superior 42,311 Intermédia Intermédia 2,462 Intermédia Intermédia 2,5 - 5 105,778 118,086 12,308 5 - 7,5 1,701 28,860 Intermédia Inferior Superior 14,851 12,308 7,5 - 10 1,701 28,860 14,851 12,308 10 - 12,5 14,851 18,687 0,682 3,154 12,5 - 15 14,851 18,687 0,682 3,154 15 - 17,5 14,851 18,687 0,682
3,154 Intermédia Inferior Intermédia 17,5 - 20 14,851 18,687 0,682 3,154 20 - 22,5 14,851 18,687 0,682 3,154 22,5 - 25 14,851 18,687 0,682 3,154 25 - 27,5 14,851 18,687 0,682 3,154 27,5 - 30 14,851 18,687 0,682 3,154 30 - 32,5 14,851 18,058 Inferior 3,154 0,053 32,5 - 35 14,851 18,058 3,154 0,053 35 - 37,5 14,851 18,058 3,154 0,053 37,5 - 40 14,851 18,058
Table 13. Original areal data of Pico per polygon.
Temporal range of the polygons (ka)
Total area (km2)
Area covered by younger material, undifferentiated & colluvium (km2)
Covered area relative to total area (%) 0 - 2 151,908 0 0 1,5 - 5 148,089 50,552 34,14 2 - 10 39,387 3,361 8,53 5 - 10 3,401 0 0 5 - 40 207,911 150,443 72,36 10 - 30 5,456 699 12,81 10-50 50,470 29,203 57,86 30-250 4,674 588 12,58
Appendix 2
Accuracy assessment & correlation
statistics
Figure 10. Percentage of spatially interpolated area per volcanogenic flow reconstruction for Pico.
Figure 12. Scatterplot of the correlation between the change in sea level and volcanogenic area of Vulcano.
