Contrasted with high average returns over the last few years and despite improved technical model... more Contrasted with high average returns over the last few years and despite improved technical modeling and analytics, the insurance industry is currently experiencing lower multiples and higher volatility. Experienced, relationship-focused and market-driven leaders are working together with sophisticated theoreticians, scientists, and engineers to address this trend. In the property catastrophe reinsurance market, results predicting mean losses over hundreds to millions of years contrast with investors' primary concern about next quarter results. Unreasonable expectations with respect to scientific capabilities and accuracy, and the quest for highest resolution and impossible "all-in-one nutshell" results have caused additional confusion in this market. The highest predicted technical returns for territories/perils with actual returns among the worst results over the last several years and the perceived correlation between sub-prime losses and the number of PhDs in the financial markets have only reinforced investors' concerns. Further commoditization of products across the globe, however, stipulates the need for more detailed risk knowledge and sophisticated portfolio underwriting to ensure future success in this market. This is driving a strong quest in the industry for technical solutions without a sufficient understanding of what future work models might indeed look like. No doubt that this market is in transition while trying to leverage the vast resources of private, government science, and/or science networks for a quantum change in its success and reputation. This presentation serves as an overview of the potential collaboration across scientific academia and the insurance industry; the quest for transparent communication; and the use of cat analytics to further evaluate risk and uncertainty along with the current modeling inventory and scoring of territories, perils, and data. Additional and potentially new products are suggested, considering quantification of "random" vs. "learnable" and "talebian" risk and their implications for insurance decision-making. Measures concerning the upside, short-term hazard and risk features are discussed along with the most likely results for the next year. Managing the maximum downside or the quantification of other than exposure risk proxies including the life-time value of deals is also addressed.
—The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian ... more —The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian fault (NAF) at the northern margin of the Erzincan basin. The focal depth of about 10 km was less than given by ISC and NEIC. Erzincan and the surrounding villages were considerably damaged. In the Erzincan basin and in the neighbouring mountains a seismic network of ten stations was installed. It was operating continuously from March 21 through June 16, 1992. More than 3,000 aftershocks were recorded of which 505 could be located. The spectral parameters of 394 and the fault-plane solutions of 53 aftershocks were determined. For the given region the frequency dependent coda Q was derived as Q c = 122 f 0.68. The aftershock area increased with time, reflecting the process of stress redistribution. Some events clustered in the immediate vicinity of the town of Erzincan close to the epicentre of the main event and seem to trace the NAF. Their source mechanism is similar to that of the main event (strike slip). About 150 aftershocks clustered in the southeastern part of the Erzincan basin where a concentration of the events in a small volume of 5 × 5 × 3 km3 was observed. The majority of fault-plane solutions available for these aftershocks showed a normal faulting mechanism with an east-west directed extension. Most of the aftershocks southeast of the basin clustered between two lineaments that were mapped by satellite images. The P-wave velocity below the Erzincan basin, derived from travel-time residual analysis, is lower compared to areas NE and SW of the basin. Three-dimen sional stress modelling of the Erzincan region qualitatively explains the occurrence of the aftershocks southeast of the basin. The calculated displacement distribution which exhibits the north-westward motion of the basin and tension at its southeastern margin, caused by the Erzincan earthquake, is in agreement with derived fault-plane solutions.
—The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian ... more —The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian fault (NAF) at the northern margin of the Erzincan basin. The focal depth of about 10 km was less than given by ISC and NEIC. Erzincan and the surrounding villages were considerably damaged. In the Erzincan basin and in the neighbouring mountains a seismic network of ten stations was installed. It was operating continuously from March 21 through June 16, 1992. More than 3,000 aftershocks were recorded of which 505 could be located. The spectral parameters of 394 and the fault-plane solutions of 53 aftershocks were determined. For the given region the frequency dependent coda Q was derived as Q c = 122 f 0.68. The aftershock area increased with time, reflecting the process of stress redistribution. Some events clustered in the immediate vicinity of the town of Erzincan close to the epicentre of the main event and seem to trace the NAF. Their source mechanism is similar to that of the main event (strike slip). About 150 aftershocks clustered in the southeastern part of the Erzincan basin where a concentration of the events in a small volume of 5 × 5 × 3 km3 was observed. The majority of fault-plane solutions available for these aftershocks showed a normal faulting mechanism with an east-west directed extension. Most of the aftershocks southeast of the basin clustered between two lineaments that were mapped by satellite images. The P-wave velocity below the Erzincan basin, derived from travel-time residual analysis, is lower compared to areas NE and SW of the basin. Three-dimen sional stress modelling of the Erzincan region qualitatively explains the occurrence of the aftershocks southeast of the basin. The calculated displacement distribution which exhibits the north-westward motion of the basin and tension at its southeastern margin, caused by the Erzincan earthquake, is in agreement with derived fault-plane solutions.
—In order to study both the interplate seismic loading cycle and the distribution of intraplate d... more —In order to study both the interplate seismic loading cycle and the distribution of intraplate deformation of the Andes, a 215 site GPS network covering Chile and the western part of Argentina was selected, monumented and observed in 1993 and 1994. A dense part of the network in northern Chile and northwest Argentina, comprising some 70 sites, was re-observed after two years in October/November, 1995. The M w = 8.0 Antofagasta (North Chile) earthquake of 30th July, 1995 took place between the two observations. The city of Antofagasta shifted 80 cm westwards by this event and the displacement still reached 10 cm at locations 300 km from the trench. Three different deformation processes have been considered for modeling the measured displacements (1) interseismic accumulation of elastic strain due to subduction coupling, (2) coseismic strain release during the Antofagasta earthquake and (3) crustal shortening in the Sub-Andes.¶Eastward displacement of the sites to the north and to the south of the area affected by the earthquake is due to the interseismic accumulation of elastic deformation. Assuming a uniform slip model of interseismic coupling, the observed displacements at the coast require a fully locked subduction interface and a depth of seismic coupling of 50 km. The geodetically derived fault plane parameters of the Antofagasta earthquake are consistent with results derived from wave-form modeling of seismolog ical data. The coseismic slip predicted by the variable slip model reaches values of 3.2 m in the dip-slip and 1.4 m in the strike-slip directions. The derived rake is 66°. Our geodetic results suggest that the oblique Nazca–South American plate convergence is accommodated by oblique earthquake slip with no slip partitioning. The observed displacements in the back-arc indicate a present-day crustal shortening rate of 3–4 mm/year which is significantly slower than the average of 10 mm/year experienced during the evolution of the Andean plateau.
Using data of two GPS campaigns as well as two ERS-1/2 Single Look Complex (SLC) datasets, we stu... more Using data of two GPS campaigns as well as two ERS-1/2 Single Look Complex (SLC) datasets, we studied the distribution of co-seismic and post-earthquake surface deformation of the major (moment magnitude Mw=8.1) Antofagasta (Chile) event of 30 July 1995. Earthquake-related fault dimensions and inter-seismic surface deformation patterns were achieved by comparing results from the GPS and interferometric Synthetic Aperture Radar
Results from repeated measurements across a 90-sites global positioning system network in Central... more Results from repeated measurements across a 90-sites global positioning system network in Central Asia provide direct evidence of current high rates of tectonic deformation far north of the IndiaEurasia suture zone. The data help to quantify both the partitioning of ...
Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia he... more Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia help to constrain the motion of tectonic blocks as well as slip rates along major faults in the area. Using 3-D forward dislocation modelling, the influence of seismic elastic loading and unloading on the measured site motions are approximated. Results suggest that the northwestern Sunda arc is fully coupled seismogenically, whereas its eastern part along Java shows localized deformation. Higher horizontal velocity gradients than expected from the modelling of a fully coupled plate interface west of Manila in the Philippines suggest that deformation may be localized there. Assuming that geodetically derived convergence represents long-term rates, accumulated geodetic moments are compared to those derived using seismic data from 1977 to 2000 (Harvard CMT catalogue). If areas displaying localized deformation are dominated by creep processes, the largest difference between accumulated and seismically released deformation is located where the 2000 June 4 Mw=7.8 Sumatra earthquake occurred.
Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia he... more Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia help to constrain the motion of tectonic blocks as well as slip rates along major faults in the area. Using 3-D forward dislocation modelling, the influence of seismic elastic loading and unloading on the measured site motions are approximated. Results suggest that the northwestern Sunda arc is fully coupled seismogenically, whereas its eastern part along Java shows localized deformation. Higher horizontal velocity gradients than expected from the modelling of a fully coupled plate interface west of Manila in the Philippines suggest that deformation may be localized there. Assuming that geodetically derived convergence represents long-term rates, accumulated geodetic moments are compared to those derived using seismic data from 1977 to 2000 (Harvard CMT catalogue). If areas displaying localized deformation are dominated by creep processes, the largest difference between accumulated and seismically released deformation is located where the 2000 June 4 Mw = 7.8 Sumatra earthquake occurred.
Contrasted with high average returns over the last few years and despite improved technical model... more Contrasted with high average returns over the last few years and despite improved technical modeling and analytics, the insurance industry is currently experiencing lower multiples and higher volatility. Experienced, relationship-focused and market-driven leaders are working together with sophisticated theoreticians, scientists, and engineers to address this trend. In the property catastrophe reinsurance market, results predicting mean losses over hundreds to millions of years contrast with investors' primary concern about next quarter results. Unreasonable expectations with respect to scientific capabilities and accuracy, and the quest for highest resolution and impossible "all-in-one nutshell" results have caused additional confusion in this market. The highest predicted technical returns for territories/perils with actual returns among the worst results over the last several years and the perceived correlation between sub-prime losses and the number of PhDs in the financial markets have only reinforced investors' concerns. Further commoditization of products across the globe, however, stipulates the need for more detailed risk knowledge and sophisticated portfolio underwriting to ensure future success in this market. This is driving a strong quest in the industry for technical solutions without a sufficient understanding of what future work models might indeed look like. No doubt that this market is in transition while trying to leverage the vast resources of private, government science, and/or science networks for a quantum change in its success and reputation. This presentation serves as an overview of the potential collaboration across scientific academia and the insurance industry; the quest for transparent communication; and the use of cat analytics to further evaluate risk and uncertainty along with the current modeling inventory and scoring of territories, perils, and data. Additional and potentially new products are suggested, considering quantification of "random" vs. "learnable" and "talebian" risk and their implications for insurance decision-making. Measures concerning the upside, short-term hazard and risk features are discussed along with the most likely results for the next year. Managing the maximum downside or the quantification of other than exposure risk proxies including the life-time value of deals is also addressed.
—The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian ... more —The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian fault (NAF) at the northern margin of the Erzincan basin. The focal depth of about 10 km was less than given by ISC and NEIC. Erzincan and the surrounding villages were considerably damaged. In the Erzincan basin and in the neighbouring mountains a seismic network of ten stations was installed. It was operating continuously from March 21 through June 16, 1992. More than 3,000 aftershocks were recorded of which 505 could be located. The spectral parameters of 394 and the fault-plane solutions of 53 aftershocks were determined. For the given region the frequency dependent coda Q was derived as Q c = 122 f 0.68. The aftershock area increased with time, reflecting the process of stress redistribution. Some events clustered in the immediate vicinity of the town of Erzincan close to the epicentre of the main event and seem to trace the NAF. Their source mechanism is similar to that of the main event (strike slip). About 150 aftershocks clustered in the southeastern part of the Erzincan basin where a concentration of the events in a small volume of 5 × 5 × 3 km3 was observed. The majority of fault-plane solutions available for these aftershocks showed a normal faulting mechanism with an east-west directed extension. Most of the aftershocks southeast of the basin clustered between two lineaments that were mapped by satellite images. The P-wave velocity below the Erzincan basin, derived from travel-time residual analysis, is lower compared to areas NE and SW of the basin. Three-dimen sional stress modelling of the Erzincan region qualitatively explains the occurrence of the aftershocks southeast of the basin. The calculated displacement distribution which exhibits the north-westward motion of the basin and tension at its southeastern margin, caused by the Erzincan earthquake, is in agreement with derived fault-plane solutions.
—The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian ... more —The Erzincan strike-slip earthquake of March 13, 1992 ruptured a section of the North Anatolian fault (NAF) at the northern margin of the Erzincan basin. The focal depth of about 10 km was less than given by ISC and NEIC. Erzincan and the surrounding villages were considerably damaged. In the Erzincan basin and in the neighbouring mountains a seismic network of ten stations was installed. It was operating continuously from March 21 through June 16, 1992. More than 3,000 aftershocks were recorded of which 505 could be located. The spectral parameters of 394 and the fault-plane solutions of 53 aftershocks were determined. For the given region the frequency dependent coda Q was derived as Q c = 122 f 0.68. The aftershock area increased with time, reflecting the process of stress redistribution. Some events clustered in the immediate vicinity of the town of Erzincan close to the epicentre of the main event and seem to trace the NAF. Their source mechanism is similar to that of the main event (strike slip). About 150 aftershocks clustered in the southeastern part of the Erzincan basin where a concentration of the events in a small volume of 5 × 5 × 3 km3 was observed. The majority of fault-plane solutions available for these aftershocks showed a normal faulting mechanism with an east-west directed extension. Most of the aftershocks southeast of the basin clustered between two lineaments that were mapped by satellite images. The P-wave velocity below the Erzincan basin, derived from travel-time residual analysis, is lower compared to areas NE and SW of the basin. Three-dimen sional stress modelling of the Erzincan region qualitatively explains the occurrence of the aftershocks southeast of the basin. The calculated displacement distribution which exhibits the north-westward motion of the basin and tension at its southeastern margin, caused by the Erzincan earthquake, is in agreement with derived fault-plane solutions.
—In order to study both the interplate seismic loading cycle and the distribution of intraplate d... more —In order to study both the interplate seismic loading cycle and the distribution of intraplate deformation of the Andes, a 215 site GPS network covering Chile and the western part of Argentina was selected, monumented and observed in 1993 and 1994. A dense part of the network in northern Chile and northwest Argentina, comprising some 70 sites, was re-observed after two years in October/November, 1995. The M w = 8.0 Antofagasta (North Chile) earthquake of 30th July, 1995 took place between the two observations. The city of Antofagasta shifted 80 cm westwards by this event and the displacement still reached 10 cm at locations 300 km from the trench. Three different deformation processes have been considered for modeling the measured displacements (1) interseismic accumulation of elastic strain due to subduction coupling, (2) coseismic strain release during the Antofagasta earthquake and (3) crustal shortening in the Sub-Andes.¶Eastward displacement of the sites to the north and to the south of the area affected by the earthquake is due to the interseismic accumulation of elastic deformation. Assuming a uniform slip model of interseismic coupling, the observed displacements at the coast require a fully locked subduction interface and a depth of seismic coupling of 50 km. The geodetically derived fault plane parameters of the Antofagasta earthquake are consistent with results derived from wave-form modeling of seismolog ical data. The coseismic slip predicted by the variable slip model reaches values of 3.2 m in the dip-slip and 1.4 m in the strike-slip directions. The derived rake is 66°. Our geodetic results suggest that the oblique Nazca–South American plate convergence is accommodated by oblique earthquake slip with no slip partitioning. The observed displacements in the back-arc indicate a present-day crustal shortening rate of 3–4 mm/year which is significantly slower than the average of 10 mm/year experienced during the evolution of the Andean plateau.
Using data of two GPS campaigns as well as two ERS-1/2 Single Look Complex (SLC) datasets, we stu... more Using data of two GPS campaigns as well as two ERS-1/2 Single Look Complex (SLC) datasets, we studied the distribution of co-seismic and post-earthquake surface deformation of the major (moment magnitude Mw=8.1) Antofagasta (Chile) event of 30 July 1995. Earthquake-related fault dimensions and inter-seismic surface deformation patterns were achieved by comparing results from the GPS and interferometric Synthetic Aperture Radar
Results from repeated measurements across a 90-sites global positioning system network in Central... more Results from repeated measurements across a 90-sites global positioning system network in Central Asia provide direct evidence of current high rates of tectonic deformation far north of the IndiaEurasia suture zone. The data help to quantify both the partitioning of ...
Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia he... more Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia help to constrain the motion of tectonic blocks as well as slip rates along major faults in the area. Using 3-D forward dislocation modelling, the influence of seismic elastic loading and unloading on the measured site motions are approximated. Results suggest that the northwestern Sunda arc is fully coupled seismogenically, whereas its eastern part along Java shows localized deformation. Higher horizontal velocity gradients than expected from the modelling of a fully coupled plate interface west of Manila in the Philippines suggest that deformation may be localized there. Assuming that geodetically derived convergence represents long-term rates, accumulated geodetic moments are compared to those derived using seismic data from 1977 to 2000 (Harvard CMT catalogue). If areas displaying localized deformation are dominated by creep processes, the largest difference between accumulated and seismically released deformation is located where the 2000 June 4 Mw=7.8 Sumatra earthquake occurred.
Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia he... more Site velocities derived from repeated measurements in a regional GPS network in Southeast Asia help to constrain the motion of tectonic blocks as well as slip rates along major faults in the area. Using 3-D forward dislocation modelling, the influence of seismic elastic loading and unloading on the measured site motions are approximated. Results suggest that the northwestern Sunda arc is fully coupled seismogenically, whereas its eastern part along Java shows localized deformation. Higher horizontal velocity gradients than expected from the modelling of a fully coupled plate interface west of Manila in the Philippines suggest that deformation may be localized there. Assuming that geodetically derived convergence represents long-term rates, accumulated geodetic moments are compared to those derived using seismic data from 1977 to 2000 (Harvard CMT catalogue). If areas displaying localized deformation are dominated by creep processes, the largest difference between accumulated and seismically released deformation is located where the 2000 June 4 Mw = 7.8 Sumatra earthquake occurred.
Uploads
Papers by Gero Michel