Assessing lahars from ice-capped volcanoes using ASTER satellite data, the SRTM DTM and two different flow models:: case study on Iztaccihuatl (Central Mexico)

Lahars frequently affect the slopes of ice-capped volcanoes. They can be triggered by volcano-ice interactions during eruptions but also by processes such as intense precipitation or by outbursts of glacial water bodies not directly related to eruptive activity. We use remote sensing, GIS and lahar models in combination with ground observations for an initial lahar hazard assessment on Iztaccihuatl volcano (5230 m a.s.l.), considering also possible future developments of the glaciers on the volcano. Observations of the glacial extent are important for estimations of future hazard scenarios, especially in a rapidly changing tropical glacial environment. In this study, analysis of the glaciers on Iztaccihuatl shows a dramatic retreat during the last 150 years: the glaciated area in 2007 corresponds to only 4% of the one in 1850 AD and the glaciers are expected to survive no later than the year 2020. Most of the glacial retreat is considered to be related to climate change but in-situ observations suggest also that geo- and hydrothermal heat flow at the summit-crater area can not be ruled out, as emphasized by fumarolic activity documented in a former study. However, development of crater lakes and englacial water reservoirs are supposed to be a more realistic scenario for lahar generation than sudden ice melting by rigorous volcano-ice interaction. Model calculations show that possible outburst floods have to be larger than similar to 5x10(5) m(3) or to achieve an H/L ratio (Height/runout Length) of 0.2 and lower in order to reach the populated lower flanks. This threshold volume equals 2.4% melted ice of Iztaccihuatl's total ice volume in 2007, assuming 40% water and 60% volumetric debris content of a potential lahar. The model sensitivity analysis reveals important effects of the generic type of the Digital Terrain Model (DTM) used on the results. As a consequence, the predicted affected areas can vary significantly. For such hazard zonation, we therefore suggest the use of different types of DTMs and flow models, followed by a careful comparison and interpretation of the results.