TY - JOUR
T1 - Modeling the October 2005 lahars at Panabaj (Guatemala)
AU - Charbonnier, Sylvain J.
AU - Connor, Charles B.
AU - Connor, L. J.
AU - Sheridan, M. F.
AU - Oliva Hernández, J. P.
AU - Richardson, J. A.
PY - 2018/1/1
Y1 - 2018/1/1
N2 - An extreme rainfall event in October of 2005 triggered two deadly lahars on the flanks of Tolimán volcano (Guatemala) that caused many fatalities in the village of Panabaj. We mapped the deposits of these lahars, then developed computer simulations of the lahars using the geologic data and compared simulated area inundated by the flows to mapped area inundated. Computer simulation of the two lahars was dramatically improved after calibration with geological data. Specifically, detailed field measurements of flow inundation area, flow thickness, flow direction, and velocity estimates, collected after lahar emplacement, were used to calibrate the rheological input parameters for the models, including deposit volume, yield strength, sediment and water concentrations, and Manning roughness coefficients. Simulations of the two lahars, with volumes of 240,200 ± 55,400 and 126,000 ± 29,000 m3, using the FLO-2D computer program produced models of lahar runout within 3% of measured runouts and produced reasonable estimates of flow thickness and velocity along the lengths of the simulated flows. We compare areas inundated using the Jaccard fit, model sensitivity, and model precision metrics, all related to Bayes’ theorem. These metrics show that false negatives (areas inundated by the observed lahar where not simulated) and false positives (areas not inundated by the observed lahar where inundation was simulated) are reduced using a model calibrated by rheology. The metrics offer a procedure for tuning model performance that will enhance model accuracy and make numerical models a more robust tool for natural hazard reduction.
AB - An extreme rainfall event in October of 2005 triggered two deadly lahars on the flanks of Tolimán volcano (Guatemala) that caused many fatalities in the village of Panabaj. We mapped the deposits of these lahars, then developed computer simulations of the lahars using the geologic data and compared simulated area inundated by the flows to mapped area inundated. Computer simulation of the two lahars was dramatically improved after calibration with geological data. Specifically, detailed field measurements of flow inundation area, flow thickness, flow direction, and velocity estimates, collected after lahar emplacement, were used to calibrate the rheological input parameters for the models, including deposit volume, yield strength, sediment and water concentrations, and Manning roughness coefficients. Simulations of the two lahars, with volumes of 240,200 ± 55,400 and 126,000 ± 29,000 m3, using the FLO-2D computer program produced models of lahar runout within 3% of measured runouts and produced reasonable estimates of flow thickness and velocity along the lengths of the simulated flows. We compare areas inundated using the Jaccard fit, model sensitivity, and model precision metrics, all related to Bayes’ theorem. These metrics show that false negatives (areas inundated by the observed lahar where not simulated) and false positives (areas not inundated by the observed lahar where inundation was simulated) are reduced using a model calibrated by rheology. The metrics offer a procedure for tuning model performance that will enhance model accuracy and make numerical models a more robust tool for natural hazard reduction.
KW - Debris flow
KW - Panabaj
KW - Extreme rainfall
KW - FLO-2D
KW - Validation metrics
KW - Model precision
KW - Model sensitivity
KW - Noneruptive event
UR - https://digitalcommons.usf.edu/geo_facpub/1556
UR - https://doi.org/10.1007/s00445-017-1169-x
U2 - 10.1007/s00445-017-1169-x
DO - 10.1007/s00445-017-1169-x
M3 - Article
VL - 80
JO - Bulletin of Volcanology
JF - Bulletin of Volcanology
ER -