C.J. van Westen

Remote sensing techniques for landslide studies and hazard zonation in Europe

Abstract An inventory is presented of researches concerning the use of remote sensing for landslide studies and hazard zonation as mainly carried out in the countries belonging to the European Community. An overview is given of the applicability of remote sensing in the following phases of landslide studies: 1. (1) Detection and classification of landslides. Special emphasis is given to the types of imagery required at different scales of analysis. 2. (2) Monitoring the activity of existing landslides using G.P.S., photogrammetrical techniques and radar interferometry. 3. (3) Analysis and prediction in space and time of slope failures. The different factors required in a landslide hazard study are evaluated, and the optimum remote sensing imagery for obtaining each of these factors is indicated. Examples are given of research work carried out in these three phases from EC countries. Finally an evaluation is given of the aspects of uncertainty associated with the use of remote sensing data, and conclusions are given as to the incorporation of remote sensing techniques within the overall framework of techniques.

Analyzing the evolution of the Tessina landslide using aerial photographs and digital elevation models

Abstract Evolution of the Tessina landslide near Belluno, Italy, from 1954 to the present situation has been documented using multitemporal landslide maps. The maps were produced through the interpretation of sequential aerial photographs and direct field-mapping of the landslide in 1998 and 1999. The interpretations were converted to large-scale multitemporal topographical maps and digitized, resulting in detailed geomorphological maps of the Tessina landslide for the following periods: 1954, 1961, 1969, 1980, 1991, 1993, 1998 and 1999. A quantitative volumetric analysis was also carried out using a series of digital elevation models derived from the available 1:5000 scale digital contour maps with 5-m contour interval for 1948, 1964, 1980, 1991 and 1993. The total volume of material removed and accumulated was calculated for the entire Tessina landslide for the different time steps available. Results indicate that the Tessina landslide existed prior to its main reactivation in 1960, after which the landslide reduced in activity. From 1991, however, very large reactivations have taken place, and the landslide continues to be active. Although the landslide has reached the lateral boundaries of the old pre-1960 landslide, it is now expanding upslope where it may still mobilize large amounts of material.

Comparing Landslide Hazard Maps

The objective of the method explained in this paper isto obtain a better insight in the decision rulesapplied by geomorphologists in the direct mapping oflandslide hazard. This can be obtained by forcinggeomorphologists to specify for each unit (polygon) intheir hazard map the criteria that they used toclassify the unit as high, medium or low hazard. Whenthis is done systemically for an entire area, it ispossible to analyze those criteria statistically, andto evaluate whether they can be grouped into generaldecision rules, or whether these criteria arecompletely site specific. The same area in the Alpagoregion in Italy was mapped at 1 : 5000 scale by threeteams of experts individually. The different methodsare presented and the results are compared.

Evaluation of prediction capability, robustness, and sensitivity in non-linear landslide susceptibility models, Guantánamo, Cuba

This paper describes a procedure for landslide susceptibility assessment based on artificial neural networks, and focuses on the estimation of the prediction capability, robustness, and sensitivity of susceptibility models. The study is carried out in the Guantanamo Province of Cuba, where 186 landslides were mapped using photo-interpretation. Twelve conditioning factors were mapped including geomorphology, geology, soils, landuse, slope angle, slope direction, internal relief, drainage density, distance from roads and faults, rainfall intensity, and ground peak acceleration. A methodology was used that subdivided the database in 3 subsets. A training set was used for updating the weights. A validation set was used to stop the training procedure when the network started losing generalization capability, and a test set was used to calculate the performance of the network. A 10-fold cross-validation was performed in order to show that the results are repeatable. The prediction capability, the robustness analysis, and the sensitivity analysis were tested on 10 mutually exclusive datasets. The results show that by means of artificial neural networks it is possible to obtain models with high prediction capability and high robustness, and that an exploration of the effect of the individual variables is possible, even if they are considered as a black-box model.