Advanced search
1 file | 94.67 MB Add to list

Tsunamis in south central Chile : evidence from coastal lakes

Philipp Kempf (UGent)
(2016)
Author
Promoter
(UGent) and (UGent)
Organization
Abstract
The south central Chilean coast lies on the Peru-Chile subduction zone. As a result, the area is prone to megathrust earthquakes. The convergent motion of the oceanic Nazca Plate and the South American Plate during these earthquakes causes co-seismic vertical deformation on the seafloor, which in turn generates tsunamis. The tsunamis propagate across the Pacific Ocean and eventually inundate the South American and other coasts, where they sometimes cause catastrophic damage. Alongside the destruction of infrastructure and the loss of life, tsunamis deposit distinct sediments. This thesis focuses on tsunami deposits in two coastal lakes, Lake Cucao and Lake Huelde, on the west coast of Isla de Chiloé, south central Chile (42.6° S), and their immediate surroundings. The thesis establishes the sedimentary characteristics of the most recent tsunami deposit in the lakes' sedimentary records. The tsunami in question was caused by the AD 1960 Great Chilean Earthquake (M\textsubscript{W} 9.5). The AD 1960 tsunami inundated both lakes, which are located \textasciitilde{}2 km apart and \textasciitilde{}1.2 km behind the present coastline. The stratigraphy of both lakes features gyttja, interrupted by the abrupt emplacement of a sandy layer with mud rip-up clasts and a mud cap. This sandy layer reflects a sudden change in the sedimentary environment. Using grain size analysis and comparisons with samples from modern environments, it is demonstrated that the proximal (seaward) part of the tsunami deposit consists of a mixture of sand derived from subaerial sources and reworked lacustrine gyttja. In the distal (landward) part of Lake Cucao, the tsunami deposit lacks the sand component and consists entirely of remobilised lake sediments. The repetition of tsunami deposit sequences (or subsets thereof) in Lake Huelde suggests a minimum of three inundating waves. Sub-bottom profiles and side scan sonar mosaics reveal tsunami inundation over the barrier and more prominently through the outlet river channel. The dominant role of the river channel as a pathway for sediment transport is also described in core samples by tsunami deposits that are fining away from the channel mouth. The identification and description of the AD 1960 deposit provide important insights into tsunami sedimentation in coastal lakes and set a baseline as to what tsunami deposits should look like in the deeper sedimentary record of the two coastal lakes. In Lake Huelde, the established expectation of the sedimentary characteristics of tsunami deposits is used to investigate the exceptionally long and continuous sedimentary record. Lake Huelde's sedimentary record contains 17 distinct detrital layers from the last 5500 yrs, which are interpreted as tsunami deposits. With the unusually high number of tsunami deposits in a single sedimentary record it is possible to analyse the recurrence pattern. Recurrence rates and pattern are an integral part to tsunami hazard assessment. In probabilistic tsunami hazard assessment tsunami recurrences are typically modelled with a Poisson distribution, which, as I demonstrate, does not necessarily represent the hazard correctly. Results of \textsuperscript{137}Cs- and \textsuperscript{210}Pb-dating, infrared stimulated luminescence (IRSL) dating and radiocarbon dating are used as input for Bayesian age-depth modelling. The resulting age-depth relationship is used to estimate the age of tsunami deposits. The inferred mean recurrence time of \textasciitilde{}325 yrs broadly agrees with the existing regional paleoseismic and paleotsunami records. However, our record reveals a large temporal variability between tsunami events, which forms a bimodal probability density function for the recurrence pattern with modes at \textasciitilde{}115 and \textasciitilde{}490 yrs. As the mean recurrence time coincides with the lowest probability between the two modes, I conclude that the sense of safety in coastal communities in south central Chile, 55 yrs after the last catastrophe, may be deceiving. A tsunamigenic earthquake in the near future would not redefine the extremes of the presented record. Lake Cucao's sedimentary history is more complex than Lake Huelde's, however, its sedimentary record holds valuable paleotsunami information, too. The area of the lake basin has been submerged since the postglacial sea-level rise and may have recorded tsunami inundations in its sedimentary record since then. A radiocarbon date from a sediment core and internal acoustic reflections, which converge towards a tidal channel, indicate that tidal currents were active at least sporadically over the last 3700 yrs. Little vertical displacement over the last 3700 yrs maintained the ability of Lake Cucao to record tsunamis in its sedimentary record. In total 15 detrital layers are interpreted as tsunami deposits with a varying level of confidence. The level of confidence depends on five criteria; there are site-specific criteria, i.e. i) high magnetic susceptibility of the sediment indicating high detrital content in contrast to the organic-rich gyttja, ii) core-to-core correlation, and iii) acoustic reflection correlation to the sedimentary record (the latter two indicate traceability of the detrital layer in the sedimentary record), and general criteria, i.e. iv) presence of mud clasts indicating an episode of extreme flow conditions in an otherwise calm environment and v) age correlation to known paleotsunamis in the area. The neighbouring Lake Huelde contains 14 or 15 tsunami deposits in the same time interval as Lake Cucao, suggesting that both lakes may have been equally reliable in recording tsunami inundation. This study adds a long paleotsunami record on a coastline, where extreme tsunamis occur relatively frequently and where long paleotsunami records are still sparse. Many important sedimentary features were revealed by side scan sonar data and sub-bottom profiles, which demonstrates how indispensible acoustic imagery can be in understanding the depositional environment, especially in lakes as dynamic as the coastal Lake Cucao. Imaging tsunami deposits in coastal lakes with sidescan sonars and sub-bottom profiles can now be added to the growing list of tools in the toolkit to observe, describe and recognise tsunami deposits. The amount of data collected about the numerous tsunami deposits in both lakes allows a ranking of tsunami deposits according to multiple criteria, which may represent the size of the associated tsunami. These criteria are i) maximum magnetic susceptibility, ii) average magnetic susceptibility, iii) traceability (lateral extent), iv) average tsunami deposit thickness and v) percentage of mud rip-up clast intervals. Integrating the two coastal lake records, it is evident that the AD 1960 tsunami was one of the major events in both lakes in the last \textasciitilde{}4000 yrs. Two older tsunami deposits of similar characteristics as the AD 1960 tsunami deposit are tentatively correlated across both lakes. Their ages are \textasciitilde{}3740 and \textasciitilde{}3890 cal. yrs BP. However, the relative ranking of tsunamis only compares the sedimentary effect in both lakes and cannot be translated directly to relative tsunami size, because environmental changes are not accounted for. For example, landscape evolution, horizontal and vertical shoreline displacement all affect how and where an environment records a tsunami inundation. Models that could quantify the tsunami size in absolute terms need this information. These models can be physical or numerical, however, physical models of tsunamis focus on subaerial landslide-induced tsunamis. Of the numerical models, there are forward models and inverse models. Forward models are extremely complex dynamic models and can include all aspects of tsunamis from fault slip simulation during the causative earthquake to vertical seafloor deformation, tsunami generation, propagation and inundation, and eventually erosion and deposition of coastal sediments. Inverse models try to infer tsunami inundation parameters, e.g. flow depth, from characteristics of tsunami deposits. However, none of the existing inverse tsunami models are applicable to tsunami sediments in coastal lakes. Furthermore, only little is known of how the environment changed around Lake Cucao and Lake Huelde, e.g. relative sea-level change, vegetation changes, horizontal shoreline displacement and more. Further investigation towards the environmental history and application of numerical forward models could facilitate quantitative paleotsunami research.
Keywords
Chile, Tsunami, tsunami deposits, coastal lake, sub-bottom profiles, side scan sonar, age-depth modelling, sediment cores, earthquakes, subduction zone

Downloads

  • Philipp Kempf - Doctoral Thesis - 2016 - low res.pdf
    • full text
    • |
    • open access
    • |
    • PDF
    • |
    • 94.67 MB

Citation

Please use this url to cite or link to this publication:

MLA
Kempf, Philipp. “Tsunamis in South Central Chile : Evidence from Coastal Lakes.” 2016 : n. pag. Print.
APA
Kempf, P. (2016). Tsunamis in south central Chile : evidence from coastal lakes. Ghent University. Faculty of Sciences, Ghent, Belgium.
Chicago author-date
Kempf, Philipp. 2016. “Tsunamis in South Central Chile : Evidence from Coastal Lakes”. Ghent, Belgium: Ghent University. Faculty of Sciences.
Chicago author-date (all authors)
Kempf, Philipp. 2016. “Tsunamis in South Central Chile : Evidence from Coastal Lakes”. Ghent, Belgium: Ghent University. Faculty of Sciences.
Vancouver
1.
Kempf P. Tsunamis in south central Chile : evidence from coastal lakes. [Ghent, Belgium]: Ghent University. Faculty of Sciences; 2016.
IEEE
[1]
P. Kempf, “Tsunamis in south central Chile : evidence from coastal lakes,” Ghent University. Faculty of Sciences, Ghent, Belgium, 2016.
@phdthesis{7248513,
  abstract     = {The south central Chilean coast lies on the Peru-Chile subduction zone. As a result, the area is prone to megathrust earthquakes. The convergent motion of the oceanic Nazca Plate and the South American Plate during these earthquakes causes co-seismic vertical deformation on the seafloor, which in turn generates tsunamis. The tsunamis propagate across the Pacific Ocean and eventually inundate the South American and other coasts, where they sometimes cause catastrophic damage. Alongside the destruction of infrastructure and the loss of life, tsunamis deposit distinct sediments. This thesis focuses on tsunami deposits in two coastal lakes, Lake Cucao and Lake Huelde, on the west coast of Isla de Chiloé, south central Chile (42.6° S), and their immediate surroundings. 
The thesis establishes the sedimentary characteristics of the most recent tsunami deposit in the lakes' sedimentary records. The tsunami in question was caused by the AD 1960 Great Chilean Earthquake (M\textsubscript{W} 9.5). The AD 1960 tsunami inundated both lakes, which are located \textasciitilde{}2 km apart and \textasciitilde{}1.2 km behind the present coastline. The stratigraphy of both lakes features gyttja, interrupted by the abrupt emplacement of a sandy layer with mud rip-up clasts and a mud cap. This sandy layer reflects a sudden change in the sedimentary environment. Using grain size analysis and comparisons with samples from modern environments, it is demonstrated that the proximal (seaward) part of the tsunami deposit consists of a mixture of sand derived from subaerial sources and reworked lacustrine gyttja. In the distal (landward) part of Lake Cucao, the tsunami deposit lacks the sand component and consists entirely of remobilised lake sediments. The repetition of tsunami deposit sequences (or subsets thereof) in Lake Huelde suggests a minimum of three inundating waves. Sub-bottom profiles and side scan sonar mosaics reveal tsunami inundation over the barrier and more prominently through the outlet river channel. The dominant role of the river channel as a pathway for sediment transport is also described in core samples by tsunami deposits that are fining away from the channel mouth. The identification and description of the AD 1960 deposit provide important insights into tsunami sedimentation in coastal lakes and set a baseline as to what tsunami deposits should look like in the deeper sedimentary record of the two coastal lakes.
In Lake Huelde, the established expectation of the sedimentary characteristics of tsunami deposits is used to investigate the exceptionally long and continuous sedimentary record. Lake Huelde's sedimentary record contains 17 distinct detrital layers from the last 5500 yrs, which are interpreted as tsunami deposits. With the unusually high number of tsunami deposits in a single sedimentary record it is possible to analyse the recurrence pattern. Recurrence rates and pattern are an integral part to tsunami hazard assessment. In probabilistic tsunami hazard assessment tsunami recurrences are typically modelled with a Poisson distribution, which, as I demonstrate, does not necessarily represent the hazard correctly. Results of \textsuperscript{137}Cs- and \textsuperscript{210}Pb-dating, infrared stimulated luminescence (IRSL) dating and radiocarbon dating are used as input for Bayesian age-depth modelling. The resulting age-depth relationship is used to estimate the age of tsunami deposits. The inferred mean recurrence time of \textasciitilde{}325 yrs broadly agrees with the existing regional paleoseismic and paleotsunami records. However, our record reveals a large temporal variability between tsunami events, which forms a bimodal probability density function for the recurrence pattern with modes at \textasciitilde{}115 and \textasciitilde{}490 yrs. As the mean recurrence time coincides with the lowest probability between the two modes, I conclude that the sense of safety in coastal communities in south central Chile, 55 yrs after the last catastrophe, may be deceiving. A tsunamigenic earthquake in the near future would not redefine the extremes of the presented record.
Lake Cucao's sedimentary history is more complex than Lake Huelde's, however, its sedimentary record holds valuable paleotsunami information, too. The area of the lake basin has been submerged since the postglacial sea-level rise and may have recorded tsunami inundations in its sedimentary record since then. A radiocarbon date from a sediment core and internal acoustic reflections, which converge towards a tidal channel, indicate that tidal currents were active at least sporadically over the last 3700 yrs. Little vertical displacement over the last 3700 yrs maintained the ability of Lake Cucao to record tsunamis in its sedimentary record. In total 15 detrital layers are interpreted as tsunami deposits with a varying level of confidence. The level of confidence depends on five criteria; there are site-specific criteria, i.e. i) high magnetic susceptibility of the sediment indicating high detrital content in contrast to the organic-rich gyttja, ii) core-to-core correlation, and iii) acoustic reflection correlation to the sedimentary record (the latter two indicate traceability of the detrital layer in the sedimentary record), and general criteria, i.e. iv) presence of mud clasts indicating an episode of extreme flow conditions in an otherwise calm environment and v) age correlation to known paleotsunamis in the area. The neighbouring Lake Huelde contains 14 or 15 tsunami deposits in the same time interval as Lake Cucao, suggesting that both lakes may have been equally reliable in recording tsunami inundation. This study adds a long paleotsunami record on a coastline, where extreme tsunamis occur relatively frequently and where long paleotsunami records are still sparse. Many important sedimentary features were revealed by side scan sonar data and sub-bottom profiles, which demonstrates how indispensible acoustic imagery can be in understanding the depositional environment, especially in lakes as dynamic as the coastal Lake Cucao. Imaging tsunami deposits in coastal lakes with sidescan sonars and sub-bottom profiles can now be added to the growing list of tools in the toolkit to observe, describe and recognise tsunami deposits. 
The amount of data collected about the numerous tsunami deposits in both lakes allows a ranking of tsunami deposits according to multiple criteria, which may represent the size of the associated tsunami. These criteria are i) maximum magnetic susceptibility, ii) average magnetic susceptibility, iii) traceability (lateral extent), iv) average tsunami deposit thickness and v) percentage of mud rip-up clast intervals. Integrating the two coastal lake records, it is evident that the AD 1960 tsunami was one of the major events in both lakes in the last \textasciitilde{}4000 yrs. Two older tsunami deposits of similar characteristics as the AD 1960 tsunami deposit are tentatively correlated across both lakes. Their ages are \textasciitilde{}3740 and \textasciitilde{}3890 cal. yrs BP. However, the relative ranking of tsunamis only compares the sedimentary effect in both lakes and cannot be translated directly to relative tsunami size, because environmental changes are not accounted for. For example, landscape evolution, horizontal and vertical shoreline displacement all affect how and where an environment records a tsunami inundation. Models that could quantify the tsunami size in absolute terms need this information. These models can be physical or numerical, however, physical models of tsunamis focus on subaerial landslide-induced tsunamis. Of the numerical models, there are forward models and inverse models. Forward models are extremely complex dynamic models and can include all aspects of tsunamis from fault slip simulation during the causative earthquake to vertical seafloor deformation, tsunami generation, propagation and inundation, and eventually erosion and deposition of coastal sediments. Inverse models try to infer tsunami inundation parameters, e.g. flow depth, from characteristics of tsunami deposits. However, none of the existing inverse tsunami models are applicable to tsunami sediments in coastal lakes. Furthermore, only little is known of how the environment changed around Lake Cucao and Lake Huelde, e.g. relative sea-level change, vegetation changes, horizontal shoreline displacement and more. Further investigation towards the environmental history and application of numerical forward models could facilitate quantitative paleotsunami research.},
  author       = {Kempf, Philipp},
  keywords     = {Chile,Tsunami,tsunami deposits,coastal lake,sub-bottom profiles,side scan sonar,age-depth modelling,sediment cores,earthquakes,subduction zone},
  language     = {eng},
  pages        = {XXI, 172},
  publisher    = {Ghent University. Faculty of Sciences},
  school       = {Ghent University},
  title        = {Tsunamis in south central Chile : evidence from coastal lakes},
  year         = {2016},
}