Beside the relative topographic roughness using focal statistics, what are the other best options, using ArcMap?
Depending on the size of your study area, I still believe that the human eye, paired with geomorphological experience in landslide field-mapping, is much superior to any - even highly sophisticated - algorithm. And: normally the error rate is significanty lower with visual interpretations than with machine-based interpretations. If, however, you have to cover a very large region, then you might try to apply shape parameters like vertical and horizontal slope curvatures ...
Manfred Buchroithner
Dear Edward,
Only topographic data is not enough for landslide risk assessment or occurance analyses. For example, soil stability also important. I suggest you that use the multi-criteria assessment techniques. As I know, you can find many paper in the literature.
Regards,
Experience+Field work+Good DEM+aerial photointepretation+(or) spatial modelling
actually, i want to delineate the area based on the High Resolution of LiDAR data first before any fieldwork being carried out, i know i can delineate it based on my jugement/expert jugement, but i dont want to be biased..coz i need scientific proved for that jugement although i can confirm on ground....thanks..any comments..
Depending on the size of your study area, I still believe that the human eye, paired with geomorphological experience in landslide field-mapping, is much superior to any - even highly sophisticated - algorithm. And: normally the error rate is significanty lower with visual interpretations than with machine-based interpretations. If, however, you have to cover a very large region, then you might try to apply shape parameters like vertical and horizontal slope curvatures ...
Manfred Buchroithner
The algorithm you will need to use will depend on the type of terrain, i.e. what is it you are looking for. In an area covered by forest but where the scarp will come up, you will need to look for variation of the intensity of the return plus derivatives of the local topographic changes... then can you extract the material that has moved?
... in a few words, you need to train your algorithms based on what environment you work in; there will be no miracle solution, and as some of our colleagues have pointed out you will certainly need to confirm a sample of your results against field work.
All depends the type and the size of landslide you want detected. Il the case of simple translational slide with size < 2000m², the roughness is not too bad if the rest of slope is regular. But, if you work on big palaeo-landslide, it is more complex and the automatisation is often difficult. I recommend to precise the type and size of your landslide to improve advices that we can give to you.
Dear Edward,
IMHO the scarps are the most detectable features and they remain for relatively longer time. To delineate the whole landslide it is a difficult question, especially because in some cases they can be nested and (parts of) the tongue can be destroyed or severely modified by the subsequent motion. It is even more difficult if you want to establish a chronology among them.
Assuming that you managed to filter out the vegetation points or there is no vegetation at all, for detecting the scarps you may want to find the areas with sharp changes in slope angles (they increase downslope and there is only little change in aspect) that form curvilinear features.
Kind regards,
Balázs
Perhaps it is important to consider the following: if visual interpretation is subjective or biased, why wouldn't any automated method also be biased? Unless landslides have a relatively unequivocal topographic signature, or signature topographic features, visual interpretation is fundamental. Automated methods are objective with regard to the application of the algorithms and rules and do not necessarily overcome the subjectivity associated with visual interpretation.
- Badges
- Science topic
- Similar topics
- Geoscience
- Geoinformatics