Final Report Summary - SLIDELAWS (Constraining geomorphic transport laws for mass wasting processes)
SLIDELAWS aims to advance the current understanding of hillslope processes at the regional scale by constraining geomorphic transport laws for mass wasting. Study areas are two drainage basins of the Trentino Region (Central Italian Alps): Val di Sole and Val di Fiemme.
The specific objectives include the following:
1. classifying the formerly glaciated landscape into geomorphic process domains;
2. discriminating the topographic signatures of post-glacial mass-wasting processes that are superimposed to the glacial palimpsest; and
3. investigating the effects of hillslope hydro-geomorphology, as reflected by the properties of both bedrock and surficial units on mass wasting style (e.g. shallow or deep-seated) in terms of magnitude-frequency relations and sediment flux.
Research methods comprise the following:
1. interpretation of eight sequential sets of aerial photographs (from 1954 to 2006) and Light detection and ranging (LIDAR)-derived hillshade rasters;
2. fieldwork; and
3. laboratory experiments.
Preliminary analysis of LIDAR and aerial photo-based inventories identify a lithologic control. By looking at landslide density the following lithologic propensity to landsliding can be inferred: metamorphic > sedimentary > extrusive > intrusive. However, exceptionally high values in sedimentary terrain of Val di Fiemme seem to point at interactions between factors controlling landslide activity. In the next months, we will consider how the spatial distribution of surficial materials (i.e. bedrock, colluvium, and till), land use (i.e. alpine, pasture, conifer forest, and deciduous forest), and topographic attributes (i.e. elevation, slope and contributing area) may interact / confound the lithologic signal.
Inspection of LIDAR hillshades allows to identify and delineate the perimeter of large slope deformations, otherwise masked by a number of environmental conditions such as forest cover, shadow, and snow). However, in the compilation of the landslide inventories, the use of aerial photos and high-resolution LIDAR hillshades did not prove to be mutually exclusive. If on hand, the former reduces uncertainty in the identification and delineation of landslide features, on the other it is too expensive to be replicated over time to obtain multi-temporal datasets across large study areas. By contrast, aerial photographs have limitations in terms of landslide identification (Brardinoni et al., 2003), but they are relatively affordable, they have good coverage for large parts of the world since 1950s, and allow to analyse landslide sediment dynamics through time (Brardinoni et al. 2009). Ideally, the two methodologies should be employed in an integrated approach. Analysis of LIDAR-derived hillshades can be used to assess volumes of material eroded and re-deposited since deglaciation.
Multitemporal interpretation of sequential photosets can be used to assess contemporary landslide activity. Together, the two methodologies can be used contrast post-glacial versus contemporary landslide activities, as well as establish landslide visibility thresholds for aerial photos using landslide scars delineated on LIDAR hillshades as true reference. In this context, the combined use of these techniques ensures greater completeness of the inventories and holds important implications for the calculation of reliable landslide magnitude-frequency relations.
Landslide magnitude-frequency relations show robust trends. Different representations of the same data highlight distinct properties of landslide distributions and yield somewhat different results. The scaling relation expressed as the rate of landslide occurrence per unit area describes the intensity of the landsliding process. In this form, the relation can be used to test distinctiveness in the intensity of landslide occurrence as a function of selected factors, such as rock type, land use, or landslide style. The scaling relation expressed as Probability density function (PDF) reduces the distribution to probability of occurrence of events within the entire study spatial and temporal domain. It can demonstrate similarity (or dissimilarity) of probabilities of occurrence for various rock types, land use, and landslide styles. The downside is that intensity of occurrence is discounted by the normalization that defines the probability. Nevertheless, this latter representation has a number of advantages over the frequency-based one for modelling purposes as it warrants smaller scatter around the simple power-law trend in the large magnitude spectrum and a wider range of magnitudes that can be approximated by a single power-law relation (i.e. from about 1 000 m2 to over 1 000 000 m2). Relations stratified by lithology show that high magnitude-low frequency events occur exclusively in metamorphic terrain. The frequency of smaller events (i.e. < 300 000 m2) is highest on metamorphic slopes and decreases in terrain underlain by sedimentary and intrusive lithologies, which in turn exhibit comparable frequencies.
Inspection of LIDAR hillshades, in conjunction with aerial photo interpretation and available information on bedrock geology/surficial materials has been critical for identifying for the first time peculiar assemblages of geomorphic process domains along longitudinal profiles, as well as for mapping geomorphic process domains at large. Specifically, the combination of glacially-derived sediment availability (i.e. glacial till) in conjunction with the topographic characteristics of major valley walls (i.e. slope aspect, slope gradient) - variables that are heavily affected by the structure and orientation of formerly active ice flows - appears to impose the sequencing of colluvial and alluvial process domains. In our study basins, besides observing the classic hillslope-colluvial-alluvial downstream sequence (also typical of unglaciated environments (Montgomery and Foufoula-Georgiou, 1993), and the glacial counterpart with the hillslope-colluvial-alluvial-colluvial-alluvial sequence (Brardinoni and Hassan, 2006), we identify numerous instances in which headwaters have a gentle, fluvial character, then become strictly colluvial in the intermediate portion (i.e. steep slope and abundance of Quaternary surficial deposits), for then reacquiring a fluvial character further downstream towards the confluence with major valley floors. These findings have clear implications from a hydro-geomorphic point of view (for example in terms of landslide run-out distances, sediment delivery to streams, and more generally on the organization of the landslide sediment flux at the basin and regional scale), as well as for the definition and mapping of risk to infrastructures and residential areas.
It is important to mention that the original SLIDELAWS project was to cover a 24-month period. Because Dr Brardinoni secured himself a permanent position as assistant professor in the Department of Geological Sciences at the University of Milano-Bicocca, the project had to be terminated after 15 months only. For this reasons, and for reasons detailed in the mid-term review report, all objectives could not be achieved. Dr Brardinoni and Professor Crosta's research team will continue pursuing the research objectives outlined in the SLIDELAWS research proposal for the period 2010-12 in collaboration with the autonomous Province of Trento.
The specific objectives include the following:
1. classifying the formerly glaciated landscape into geomorphic process domains;
2. discriminating the topographic signatures of post-glacial mass-wasting processes that are superimposed to the glacial palimpsest; and
3. investigating the effects of hillslope hydro-geomorphology, as reflected by the properties of both bedrock and surficial units on mass wasting style (e.g. shallow or deep-seated) in terms of magnitude-frequency relations and sediment flux.
Research methods comprise the following:
1. interpretation of eight sequential sets of aerial photographs (from 1954 to 2006) and Light detection and ranging (LIDAR)-derived hillshade rasters;
2. fieldwork; and
3. laboratory experiments.
Preliminary analysis of LIDAR and aerial photo-based inventories identify a lithologic control. By looking at landslide density the following lithologic propensity to landsliding can be inferred: metamorphic > sedimentary > extrusive > intrusive. However, exceptionally high values in sedimentary terrain of Val di Fiemme seem to point at interactions between factors controlling landslide activity. In the next months, we will consider how the spatial distribution of surficial materials (i.e. bedrock, colluvium, and till), land use (i.e. alpine, pasture, conifer forest, and deciduous forest), and topographic attributes (i.e. elevation, slope and contributing area) may interact / confound the lithologic signal.
Inspection of LIDAR hillshades allows to identify and delineate the perimeter of large slope deformations, otherwise masked by a number of environmental conditions such as forest cover, shadow, and snow). However, in the compilation of the landslide inventories, the use of aerial photos and high-resolution LIDAR hillshades did not prove to be mutually exclusive. If on hand, the former reduces uncertainty in the identification and delineation of landslide features, on the other it is too expensive to be replicated over time to obtain multi-temporal datasets across large study areas. By contrast, aerial photographs have limitations in terms of landslide identification (Brardinoni et al., 2003), but they are relatively affordable, they have good coverage for large parts of the world since 1950s, and allow to analyse landslide sediment dynamics through time (Brardinoni et al. 2009). Ideally, the two methodologies should be employed in an integrated approach. Analysis of LIDAR-derived hillshades can be used to assess volumes of material eroded and re-deposited since deglaciation.
Multitemporal interpretation of sequential photosets can be used to assess contemporary landslide activity. Together, the two methodologies can be used contrast post-glacial versus contemporary landslide activities, as well as establish landslide visibility thresholds for aerial photos using landslide scars delineated on LIDAR hillshades as true reference. In this context, the combined use of these techniques ensures greater completeness of the inventories and holds important implications for the calculation of reliable landslide magnitude-frequency relations.
Landslide magnitude-frequency relations show robust trends. Different representations of the same data highlight distinct properties of landslide distributions and yield somewhat different results. The scaling relation expressed as the rate of landslide occurrence per unit area describes the intensity of the landsliding process. In this form, the relation can be used to test distinctiveness in the intensity of landslide occurrence as a function of selected factors, such as rock type, land use, or landslide style. The scaling relation expressed as Probability density function (PDF) reduces the distribution to probability of occurrence of events within the entire study spatial and temporal domain. It can demonstrate similarity (or dissimilarity) of probabilities of occurrence for various rock types, land use, and landslide styles. The downside is that intensity of occurrence is discounted by the normalization that defines the probability. Nevertheless, this latter representation has a number of advantages over the frequency-based one for modelling purposes as it warrants smaller scatter around the simple power-law trend in the large magnitude spectrum and a wider range of magnitudes that can be approximated by a single power-law relation (i.e. from about 1 000 m2 to over 1 000 000 m2). Relations stratified by lithology show that high magnitude-low frequency events occur exclusively in metamorphic terrain. The frequency of smaller events (i.e. < 300 000 m2) is highest on metamorphic slopes and decreases in terrain underlain by sedimentary and intrusive lithologies, which in turn exhibit comparable frequencies.
Inspection of LIDAR hillshades, in conjunction with aerial photo interpretation and available information on bedrock geology/surficial materials has been critical for identifying for the first time peculiar assemblages of geomorphic process domains along longitudinal profiles, as well as for mapping geomorphic process domains at large. Specifically, the combination of glacially-derived sediment availability (i.e. glacial till) in conjunction with the topographic characteristics of major valley walls (i.e. slope aspect, slope gradient) - variables that are heavily affected by the structure and orientation of formerly active ice flows - appears to impose the sequencing of colluvial and alluvial process domains. In our study basins, besides observing the classic hillslope-colluvial-alluvial downstream sequence (also typical of unglaciated environments (Montgomery and Foufoula-Georgiou, 1993), and the glacial counterpart with the hillslope-colluvial-alluvial-colluvial-alluvial sequence (Brardinoni and Hassan, 2006), we identify numerous instances in which headwaters have a gentle, fluvial character, then become strictly colluvial in the intermediate portion (i.e. steep slope and abundance of Quaternary surficial deposits), for then reacquiring a fluvial character further downstream towards the confluence with major valley floors. These findings have clear implications from a hydro-geomorphic point of view (for example in terms of landslide run-out distances, sediment delivery to streams, and more generally on the organization of the landslide sediment flux at the basin and regional scale), as well as for the definition and mapping of risk to infrastructures and residential areas.
It is important to mention that the original SLIDELAWS project was to cover a 24-month period. Because Dr Brardinoni secured himself a permanent position as assistant professor in the Department of Geological Sciences at the University of Milano-Bicocca, the project had to be terminated after 15 months only. For this reasons, and for reasons detailed in the mid-term review report, all objectives could not be achieved. Dr Brardinoni and Professor Crosta's research team will continue pursuing the research objectives outlined in the SLIDELAWS research proposal for the period 2010-12 in collaboration with the autonomous Province of Trento.