Home About HyMeX
Motivations
Science questions
Observation strategy
Modelling strategy
Target areas
Key documents
Organisation
International coordination
Working groups
Task teams
National contributions
Endorsements
Resources
Database
Data policy
Publications
Education and summer schools
Drifting balloons (BAMED)
SOP web page
Google maps data visualisation
Workshops Projects
ASICS-MED
MOBICLIMEX
MUSIC
IODA-MED
REMEMBER
FLOODSCALE
EXAEDRE
Offers Links Contacts
Science & Task teams
Science teams
Task teams
Implementation plan
Coordination
International Scientific Steering Committee (ISSC)
Executive Committee for Implementation and Science Coordination (EC-ISC)
Executive Committee - France (EC-Fr)
HyMeX France
HyMeX Italy
HyMeX Spain
Archive
by Crema, S.
Abstract:
The present work analyzes the hydrologic conditions leading to the triggering of debris flows in an alpine region. The overall analysis has been split in two parts: the first part of the analysis has been carried out at a regional and decadal scale to improve our knowledge of rainfall thresholds for debris-flow occurrences, of the uncertainty related to rainfall estimation at debris-flow initiation sites, and of the main morphometric characteristics of debris-flow triggering locations; in the second part a focus has been devoted to the analysis of the hydrologic response of some watersheds for a selection of events with the help of detailed input information, both topographic and hydrological. The study area is upper Adige River basin (Northern Italy), which occupies a distinctive hydrometeorological niche, characterized by high frequency of orographic thunderstorms. The first part of the thesis analyzes the debris-flow triggering issue at regional and decadal-scale (2000-2010). A morphometric characterization of debris-flow triggering locations has been carried out in terms of slope-area analysis. The characterization was carried out dividing considered debris flows in classes of rainfall duration, rainfall intensity and considering also the potential relation with the different geologic settings. Another issue under investigation is related to rainfall threshold estimation (from raingauges network) for debris-flow occurrence and the uncertainty related to this estimation. First, a set of rainfall thresholds has been derived adopting a frequentist approach. The rainfall estimation procedure used for threshold identification was based on two major sources: records from the nearest gauge and an Inverse Distance interpolation of all the available records at regional and decadal scale. Thresholds have been analyzed considering all the available points and also trying to group debris-flow occurrences according to seasonal occurrence, different storm characteristics and homogeneous geologic setting. The level of uncertainty related to rainfall threshold estimation has been analyzed in detail The estimation problem is especially severe for the location of the debris flows, with raingauges that are commonly located at low elevation (e.g., in the valley floors) and debris flows that originate at high elevations, in the head part of the mountain catchments. The rainfall estimation procedure used for the uncertainty assessment, is the one based on the nearest raingauge, which is used as a proxy to estimate the rainfall that has resulted in debris flows. The objective is to investigate the impact of the uncertainty inherent in the estimation of rainfall on the definition of a threshold for possible debris-flow occurrence, and on the operational use of the threshold for prediction purposes. These effects are likely to depend on two factors: the space-time rainfall variability and the distances between the debris-flow locations and the raingauges on both the horizontal and the vertical planes. Three main problems are considered: (i) the effect of the rainfall sampling problem on the estimation of the parameters of the threshold model, (ii) the effect of applying a number of procedures to filter the rainfall information on the threshold model estimation, and (iii) the examination of the impact of the precipitation sampling problem on the performance of the threshold as a predictor of debris-flow occurrence. These aspects are examined here based on a simulation experiment. The methodology examines the intensity-duration thresholds derived from a set of raingauge locations that is assumed to represent debris flow/landslide points (DFR) and an equivalent set of raingauges assumed to have the role of closest available measurement (MR). A set of reference rainfall thresholds is used to identify the rainfall events at DFR that “triggered debris flows (i.e. exceed the threshold). For these same events, the corresponding rainfall thresholds are derived from MR observations. Comparison between the rainfall thresholds derived from DFR and MR, revealed that uncertainty in rainfall estimation has a major impact on estimated intensity-duration thresholds. Specifically, results showed that thresholds estimated from MR observations are consistently underestimated. Evaluation of the estimated thresholds for warning procedures, showed that while detection is high, the main issue is the high false alarm ratio, which limits the overall accuracy of the procedure. Overall performance on debris-flow prediction was shown to be good for low rainfall thresholds and poor for high rainfall thresholds examined. Finally different interpolation techniques have been applied to a set of gauges close to the DFR to assess the eventual improvement brought by the interpolation procedure. Results on this showed that interpolation can improve estimates specifically in the case of poor DFR-MR correlation, while in the other cases, especially considering intense and/or localized storms, the benefits of considering interpolated value in respect to only one near gauge is not so significant. In the second part of the thesis the hydrometeorological and hydrological controls of these events are examined through analyses of three storm systems occurred on October 3-4, 2006, June 20-21 2007 and September 3-4, 2009. The first storm system generated a moderate flash flood along the main streams with almost no debris flows in the tributaries, the second triggered a large number of debris flows and was characterized by a minor runoff response for the major streams, and the third resulted in both a relevant flash flood response and widespread debris flows. These events have been examined by using per-event calibrated radar rainfall data, providing an important input for the evaluation of a distributed hydrological model, and by using a database reporting location and information related to the events. Raingauge and streamflow data, debris flow spatial information and observations are used along with the distributed model to analyze the hydrological and geomorphic responses to these rainstorms. The striking contrast in flood and geomorphic responses between the three floods is related to contrasts in the antecedent moisture conditions and in the space-time structure of the triggering storm. Rainfall-runoff modeling has permitted estimating discharge and runoff volumes in a number of small ungauged catchments in which debris-flow volumes had been assessed by means of post-event surveys. The computation of sediment concentration by volume using model-estimated water runoff has resulted in realistic values. The relations between water peak discharge and debris-flow volumes show higher correlation coefficients than those involving morphometric variables and rainfall amounts.
Reference:
Crema, S., 2014: Hydrologic control on the triggering and magnitude of debris flows in alpine catchmentsPhD thesis, Università di Padova.
Bibtex Entry:
@Phdthesis{Crema2014,
  Title                    = {Hydrologic control on the triggering and magnitude of debris flows in alpine catchments},
  Author                   = {Crema, S.},
  Beginningdate            = {2011.01.XX},
  Country                  = {Italy},
  Enddate                  = {2014.01.29},
  Location                 = {Padova},
  School                   = {Università di Padova},
  Supervisors              = {L. Marchi},
  Supervisorsaffiliations  = {Consiglio Nazionale delle Ricerche, Istituto di Ricerca per la Protezione Idrogeologica},
  Year                     = {2014},

  Address                  = {stefano.crema@irpi.cnr.it;},

  Abstract                 = {The present work analyzes the hydrologic conditions leading to the triggering
of debris flows in an alpine region. The overall analysis has been split in two parts: the first part of the analysis has been carried out at a regional and decadal scale to improve our knowledge of rainfall thresholds for debris-flow occurrences, of the uncertainty related to rainfall estimation at debris-flow initiation sites, and of the main morphometric characteristics of debris-flow triggering locations; in the second part a focus has been devoted to the analysis of the hydrologic response of some watersheds for a selection of events with the help of detailed input information, both topographic and hydrological. The study area is upper Adige River basin (Northern Italy), which occupies a distinctive hydrometeorological niche, characterized by high frequency of orographic thunderstorms.
The first part of the thesis analyzes the debris-flow triggering issue at regional and decadal-scale (2000-2010). A morphometric characterization of debris-flow triggering locations has been carried out in terms of slope-area analysis. The characterization was carried out dividing considered debris flows in classes of rainfall duration, rainfall intensity and considering also the potential relation with the different geologic settings. Another issue under investigation is related to rainfall threshold estimation (from raingauges network) for debris-flow occurrence and the uncertainty related to this estimation. First, a set of rainfall thresholds has been derived adopting a frequentist approach. The rainfall estimation procedure used for threshold identification was based on two major sources: records from the nearest gauge and an Inverse Distance interpolation of all the available records at regional and decadal scale. Thresholds have been analyzed considering all the available points and also trying to group debris-flow occurrences according to seasonal occurrence, different storm characteristics and homogeneous geologic setting. The level of uncertainty related to rainfall threshold estimation has been analyzed in detail The estimation problem is especially severe for the location of the debris flows, with raingauges that are commonly located at low elevation (e.g., in the valley floors) and debris flows that originate at high elevations, in the head part of the mountain catchments. The rainfall estimation procedure used for the uncertainty assessment, is the one based on the nearest raingauge, which is used as a proxy to estimate the rainfall that has resulted in debris flows. The objective is to investigate the impact of the uncertainty inherent in the estimation of rainfall on the definition of a threshold for possible debris-flow occurrence, and on the operational use of the threshold for prediction purposes. These effects are likely to depend on two factors: the space-time rainfall variability and the distances between the debris-flow locations and the raingauges on both the horizontal and the vertical planes. Three main problems are considered: (i) the effect of the rainfall sampling problem on the estimation of the parameters of the threshold model, (ii) the effect of applying a number of procedures to filter the rainfall information on the threshold model estimation, and (iii) the examination of the impact of the precipitation sampling problem on the performance of the threshold as a predictor of debris-flow occurrence. These aspects are examined here based on a simulation experiment. The methodology examines the intensity-duration thresholds derived from a set of raingauge locations that is assumed to represent debris flow/landslide points (DFR) and an equivalent set of raingauges assumed to have the role of closest available measurement (MR). A set of reference rainfall thresholds is used to identify the rainfall events at DFR that “triggered debris flows (i.e. exceed the threshold). For these same events, the corresponding rainfall thresholds are derived from MR observations. Comparison between the rainfall thresholds derived from DFR and MR, revealed that uncertainty in rainfall estimation has a major impact on estimated intensity-duration thresholds. Specifically, results showed that thresholds estimated from MR observations are consistently underestimated. Evaluation of the estimated thresholds for warning procedures, showed that while detection is high, the main issue is the high false alarm ratio, which limits the overall accuracy of the procedure. Overall performance on debris-flow prediction was shown to be good for low rainfall thresholds and poor for high rainfall thresholds examined. Finally different interpolation techniques have been applied to a set of gauges close to the DFR to assess the eventual improvement brought by the interpolation procedure. Results on this showed that interpolation can improve estimates specifically in the case of poor DFR-MR correlation, while in the other cases, especially considering intense and/or localized storms, the benefits of considering interpolated value in respect to only one near gauge is not so significant.
In the second part of the thesis the hydrometeorological and hydrological controls of these events are examined through analyses of three storm systems occurred on October 3-4, 2006, June 20-21 2007 and September 3-4, 2009. The first storm system generated a moderate flash flood along the main streams with almost no debris flows in the tributaries, the second triggered a large number of debris flows and was characterized by a minor runoff response for the major streams, and the third resulted in both a relevant flash flood response and widespread debris flows. These events have been examined by using per-event calibrated radar rainfall data, providing an important input for the evaluation of a distributed hydrological model, and by using a database reporting location and information related to the events. Raingauge and streamflow data, debris flow spatial information and observations are used along with the distributed model to analyze the hydrological and geomorphic responses to these
rainstorms. The striking contrast in flood and geomorphic responses between the three floods is related to contrasts in the antecedent moisture conditions and in the space-time structure of the triggering storm. Rainfall-runoff modeling has permitted estimating discharge and runoff volumes in a number of small ungauged catchments in which debris-flow volumes had been assessed by means of post-event surveys. The computation of sediment concentration by volume using model-estimated water runoff has resulted in realistic values. The relations between water peak discharge and debris-flow volumes show higher correlation coefficients than those involving morphometric variables and rainfall amounts.},
  Completeentry            = {No},
  Owner                    = {hymexw},
  Timestamp                = {2016.01.08},
  Url                      = {http://paduaresearch.cab.unipd.it/6602/}
}