Comparison Between Two Different Methods Applied to Define Rainfall Thresholds for Landslide Forecasting in Idukki District of Kerala, India

SIGMA (Sistema Integrato Gestione Monitoraggio Allerta &#8211; integrated system for management, monitoring and alerting) is a landslide forecasting model at regional scale which is operational in Emilia Romagna (Italy) for more than 20 years. It was conceived to be operated with a sparse rain gauge network with coarse (daily) temporal resolution and to account for both shallow landslides (typically triggered by short and intense rainstorms) and deep seated landslides (typically triggered by long and less intense rainfalls). SIGMA model is based on the statistical distribution of cumulative rainfall values (calculated over varying time windows), and rainfall thresholds are defined as the multiples of standard deviation of the same, to identify anomalous rainfalls with the potential of triggering landslides.In this study, SIGMA model is applied for the first time in a geographical location outside of Italy, i.e. Kalimpong town in India. The SIGMA algorithm is customized using the historical rainfall and landslide data of Kalimpong from 2010 to 2015 and has been validated using the data from 2016 to 2017. The model was validated by building a confusion matrix and calculating statistical skill scores, which were compared with those of the state-of-the-art intensity-duration rainfall thresholds derived for the region.Results of the comparison clearly show that SIGMA performs much better than the other models in forecasting landslides: all instances of the validation confusion matrix are improved, and all skill scores are higher than I-D thresholds, with an efficiency of 92% and a likelihood ratio of 11.28. We explain this outcome mainly with technical characteristics of the site: when only daily rainfall measurements from a spare gauge network are available, SIGMA outperforms other approaches based on peak measurements, like intensity &#8211; duration thresholds, which cannot be captured adequately by daily measurements. SIGMA model thus showed a good potential to be used as a part of the local Landslide Early Warning System (LEWS).

Download Full-text

Spatio-temporal landslide forecasting based on combination of rainfall thresholds and landslide susceptibility maps: a test in the Northern Apennines (Italy).

10.5194/egusphere-egu2020-9124 ◽

2020 ◽

Author(s):

Ascanio Rosi ◽

Samuele Segoni ◽

Veronica Tofani ◽

Filippo Catani

Keyword(s):

Spatial Resolution ◽

Landslide Susceptibility ◽

End Users ◽

Rainfall Thresholds ◽

Landslide Occurrence ◽

Susceptibility Maps ◽

Coarse Spatial Resolution ◽

Landslide Forecasting ◽

Spatio Temporal ◽

Landslide Susceptibility Maps

Landslide forecasting and early warning at regional scale are difficult task and they are usually accomplished by the mean of statistical approaches aimed to define rainfall thresholds and landslide susceptibility maps. &#160;Landslide susceptibility maps are based on the analysis of predisposing factors to assess the spatial probability of landslide occurrence, while rainfall thresholds are based on the correlation, valid on a wide area, between landslide occurrence and triggering factors, which usually are a couple of rainfall parameters, such as rainfall duration and intensity. Susceptibility maps are static map that can be used for the spatial prediction of the most landslide prone areas, nut cannot be used to predict the temporal occurrence of a landslide triggering. Rainfall thresholds can be used for temporal prediction, but with a coarse spatial resolution (usually some hundreds or thousands of km2), and the reference areas could contains both plains and hillslopes, so the alerts could involves both areas, even if landslides are improbable in river plains; this means that rainfall thresholds are not very suitable to identify the most probable triggering sites. Rainfall thresholds and susceptibility maps can be therefore conveniently combined into dynamic hazard matrixes to obtain spatio-temporal forecasts of landslide hazard. To combine these inputs, they are combined in a purposely-built hazard matrix, where each parameter is classified into 3 classes: landslide susceptibility map has been classified in S1 (low susceptibility), S2 (medium susceptibility) and S3 (high susceptibility), while rainfall rate has been classified in the classes R1, R2 and R3, by the definition of 2 rainfall thresholds. The combination of the aforementioned classes allowed to define a matrix with 5 hazard classes, from H0 (null hazard) to H4 (high hazard), which was calibrated so that there was not any landslide in the H0 class and that the 90% of the landslide were in H2-H4 classes. The result of this procedure is a dynamic hazard map, where the hazard, which is calculated for each pixel, can change over the time, based on rainfall rate variations. For operational purposes, such a map cannot be used, since the pixel based resolution is too fine to be used during an emergency or to plan any activity in the planification phase, so the results have been aggregated at municipality scale, which is more easily readable for the end-users as local administrators and decision makers. In this way it is possible to overcome the issues due to the stillness of susceptibility maps and to the coarse spatial resolution of rainfall thresholds, also avoiding results which could be hardly understandable outside of the scientific community. This procedure was tested in a test site located in Northern Tuscany (Italy) and the work showed the possibility of obtaining results which are balanced between the scientific soundness and the needs of end-users like mayors, local administrators and civil protection personnel.&#160;

Download Full-text

Deterministic and Probabilistic Rainfall Thresholds for Landslide Forecasting

Advancing Culture of Living with Landslides ◽

10.1007/978-3-319-53485-5_18 ◽

2017 ◽

pp. 169-176 ◽

Cited By ~ 1

Author(s):

Pasquale Versace ◽

Davide L. De Luca

Keyword(s):

Rainfall Thresholds ◽

Landslide Forecasting

Download Full-text

A Regional-Scale Landslide Warning System Based on 20 Years of Operational Experience

Water ◽

10.3390/w10101297 ◽

2018 ◽

Vol 10 (10) ◽

pp. 1297 ◽

Cited By ~ 34

Author(s):

Samuele Segoni ◽

Ascanio Rosi ◽

Riccardo Fanti ◽

Angela Gallucci ◽

Antonio Monni ◽

...

Keyword(s):

Soil Moisture ◽

Development Process ◽

Regional Scale ◽

Warning System ◽

Operational Experience ◽

Complex Task ◽

Warning Systems ◽

Research Activity ◽

Rainfall Thresholds ◽

Landslide Forecasting

SIGMA is a regional landslide warning system based on statistical rainfall thresholds that operates in Emilia Romagna (Italy). In this work, we depict its birth and the continuous development process, still ongoing, after two decades of operational employ. Indeed, a constant work was carried out to gather and incorporate in the modeling new data (extended rainfall recordings, updated landslides inventories, temperature and soil moisture data). The use of these data allowed for regular updates of the model and some conceptual improvements, which consistently increased the forecasting effectiveness of the warning system through time. Landslide forecasting at regional scale is a very complex task, but this paper shows that, as time passes by, the systematic gathering and analysis of new data and the continuous progresses of research activity, uncertainties can be progressively reduced. Thus, by the setting up of forward-looking research programs, the performances and the reliability of regional scale warning systems can be increased with time.

Download Full-text

Quantitative comparison between two different methodologies to define rainfall thresholds for landslide forecasting

Natural Hazards and Earth System Science ◽

10.5194/nhess-15-2413-2015 ◽

2015 ◽

Vol 15 (10) ◽

pp. 2413-2423 ◽

Cited By ~ 40

Author(s):

D. Lagomarsino ◽

S. Segoni ◽

A. Rosi ◽

G. Rossi ◽

A. Battistini ◽

...

Keyword(s):

Early Warning ◽

State Of The Art ◽

Test Site ◽

Early Warning Systems ◽

The Other ◽

Threshold Models ◽

Warning Systems ◽

Rainfall Thresholds ◽

Landslide Forecasting ◽

Validation Procedure

Abstract. This work proposes a methodology to compare the forecasting effectiveness of different rainfall threshold models for landslide forecasting. We tested our methodology with two state-of-the-art models, one using intensity–duration thresholds and the other based on cumulative rainfall thresholds. The first model identifies rainfall intensity–duration thresholds by means of a software program called MaCumBA (MAssive CUMulative Brisk Analyzer) (Segoni et al., 2014a) that analyzes rain gauge records, extracts intensity (I) and duration (D) of the rainstorms associated with the initiation of landslides, plots these values on a diagram and identifies the thresholds that define the lower bounds of the I–D values. A back analysis using data from past events is used to identify the threshold conditions associated with the least number of false alarms. The second model (SIGMA) (Sistema Integrato Gestione Monitoraggio Allerta) (Martelloni et al., 2012) is based on the hypothesis that anomalous or extreme values of accumulated rainfall are responsible for landslide triggering: the statistical distribution of the rainfall series is analyzed, and multiples of the standard deviation (σ) are used as thresholds to discriminate between ordinary and extraordinary rainfall events. The name of the model, SIGMA, reflects the central role of the standard deviations. To perform a quantitative and objective comparison, these two models were applied in two different areas, each time performing a site-specific calibration against available rainfall and landslide data. For each application, a validation procedure was carried out on an independent data set and a confusion matrix was built. The results of the confusion matrixes were combined to define a series of indexes commonly used to evaluate model performances in natural hazard assessment. The comparison of these indexes allowed to identify the most effective model in each case study and, consequently, which threshold should be used in the local early warning system in order to obtain the best possible risk management. In our application, none of the two models prevailed absolutely over the other, since each model performed better in a test site and worse in the other one, depending on the characteristics of the area. We conclude that, even if state-of-the-art threshold models can be exported from a test site to another, their employment in local early warning systems should be carefully evaluated: the effectiveness of a threshold model depends on the test site characteristics (including the quality and quantity of the input data), and a validation procedure and a comparison with alternative models should be performed before its implementation in operational early warning systems.

Download Full-text

Calibration and validation of rainfall thresholds for shallow landslide forecasting in Sicily, southern Italy

Geomorphology ◽

10.1016/j.geomorph.2014.10.019 ◽

2015 ◽

Vol 228 ◽

pp. 653-665 ◽

Cited By ~ 105

Author(s):

S.L. Gariano ◽

M.T. Brunetti ◽

G. Iovine ◽

M. Melillo ◽

S. Peruccacci ◽

...

Keyword(s):

Southern Italy ◽

Shallow Landslide ◽

Rainfall Thresholds ◽

Calibration And Validation ◽

Landslide Forecasting

Download Full-text

Reanalysis of Soil Moisture Used for Rainfall Thresholds for Rainfall-Induced Landslides: The Italian Case Study

Water ◽

10.3390/w13141977 ◽

2021 ◽

Vol 13 (14) ◽

pp. 1977

Author(s):

Nejc Bezak ◽

Mateja Jemec Auflič ◽

Matjaž Mikoš

Keyword(s):

Soil Moisture ◽

Regional Scale ◽

Rainfall Threshold ◽

Economic Damage ◽

Rainfall Thresholds ◽

Forecasting Accuracy ◽

Soil Moisture Data ◽

Italian Case ◽

Landslide Forecasting

Landslides are one of the most frequent natural disasters that can endanger human lives and property. Therefore, prediction of landslides is essential to reduce economic damage and save human lives. Numerous methods have been developed for the prediction of landslides triggering, ranging from simple methods that include empirical rainfall thresholds, to more complex ones that use sophisticated physically- or conceptually-based models. Reanalysis of soil moisture data could be one option to improve landslide forecasting accuracy. This study used the publicly available FraneItalia database hat contains almost 9000 landslide events that occurred in the 2010–2017 period in Italy. The Copernicus Uncertainties in Ensembles of Regional Reanalyses (UERRA) dataset was used to obtain precipitation and volumetric soil moisture data. The results of this study indicated that precipitation information is still a much better predictor of landslides triggering compared to the reanalyzed (i.e., not very detailed) soil moisture data. This conclusion is valid both for local (i.e., grid) and regional (i.e., catchment-based) scales. Additionally, at the regional scale, soil moisture data can only predict a few landslide events (i.e., on average around one) that are not otherwise predicted by the simple empirical rainfall threshold approach; however, this approach on average, predicted around 18 events (i.e., 55% of all events). Despite this, additional investigation is needed using other (more complete) landslide databases and other (more detailed) soil moisture products.

Download Full-text

Quantitative comparison between two different methodologies to define rainfall thresholds for landslide forecasting

Natural Hazards and Earth System Sciences Discussions ◽

10.5194/nhessd-3-891-2015 ◽

2015 ◽

Vol 3 (1) ◽

pp. 891-917 ◽

Cited By ~ 2

Author(s):

D. Lagomarsino ◽

S. Segoni ◽

A. Rosi ◽

G. Rossi ◽

A. Battistini ◽

...

Keyword(s):

Early Warning ◽

Extreme Values ◽

Threshold Model ◽

Confusion Matrix ◽

Test Site ◽

Early Warning Systems ◽

The Other ◽

False Alarms ◽

Rainfall Thresholds ◽

Landslide Forecasting

Abstract. This work proposes a methodology to compare the forecasting effectiveness of different rainfall threshold models for landslide forecasting. We tested our methodology with two state-of-the-art models, one using intensity-duration thresholds and the other based on cumulative rainfall thresholds. The first model identifies rainfall intensity-duration thresholds by means of a software called MaCumBA (MAssive CUMulative Brisk Analyzer) (Segoni et al., 2014a) that analyzes rain-gauge records, extracts the intensities (I) and durations (D) of the rainstorms associated with the initiation of landslides, plots these values on a diagram, and identifies thresholds that define the lower bounds of the I−D values. A back analysis using data from past events is used to identify the threshold conditions associated with the least amount of false alarms. The second model (SIGMA) (Sistema Integrato Gestione Monitoraggio Allerta) (Martelloni et al., 2012) is based on the hypothesis that anomalous or extreme values of rainfall are responsible for landslide triggering: the statistical distribution of the rainfall series is analyzed, and multiples of the SD (σ) are used as thresholds to discriminate between ordinary and extraordinary rainfall events. The name of the model, SIGMA, reflects the central role of the SDs in the proposed methodology. To perform a quantitative and objective comparison, these two methodologies were applied in two different areas, each time performing a site-specific calibration against available rainfall and landslide data. After each application, a validation procedure was carried out on an independent dataset and a confusion matrix was build. The results of the confusion matrixes were combined to define a series of indexes commonly used to evaluate model performances in natural hazard assessment. The comparison of these indexes allowed assessing the most effective model in each case of study and, consequently, which threshold should be used in the local early warning system in order to obtain the best possible risk management. In our application, none of the two models prevailed absolutely on the other, since each model performed better in a test site and worse in the other one, depending on the physical characteristics of the area. This conclusion can be generalized and it can be assumed that the effectiveness of a threshold model depends on the test site characteristics (including the quality and quantity of the input data) and that a validation and a comparison with alternative models should be performed before the implementation in operational early warning systems.

Download Full-text

A susceptibility-based rainfall threshold approach for landslide occurrence

Natural Hazards and Earth System Science ◽

10.5194/nhess-19-775-2019 ◽

2019 ◽

Vol 19 (4) ◽

pp. 775-789 ◽

Cited By ~ 12

Author(s):

Elise Monsieurs ◽

Olivier Dewitte ◽

Alain Demoulin

Keyword(s):

Landslide Susceptibility ◽

Rainfall Threshold ◽

Exceedance Probability ◽

Equal Weight ◽

Antecedent Rainfall ◽

Rainfall Thresholds ◽

Landslide Occurrence ◽

Continental Scale ◽

Satellite Rainfall ◽

Rainfall Estimates

Abstract. Rainfall threshold determination is a pressing issue in the landslide scientific community. While major improvements have been made towards more reproducible techniques for the identification of triggering conditions for landsliding, the now well-established rainfall intensity or event-duration thresholds for landsliding suffer from several limitations. Here, we propose a new approach of the frequentist method for threshold definition based on satellite-derived antecedent rainfall estimates directly coupled with landslide susceptibility data. Adopting a bootstrap statistical technique for the identification of threshold uncertainties at different exceedance probability levels, it results in thresholds expressed as AR = (α±Δα)⋅S(β±Δβ), where AR is antecedent rainfall (mm), S is landslide susceptibility, α and β are scaling parameters, and Δα and Δβ are their uncertainties. The main improvements of this approach consist in (1) using spatially continuous satellite rainfall data, (2) giving equal weight to rainfall characteristics and ground susceptibility factors in the definition of spatially varying rainfall thresholds, (3) proposing an exponential antecedent rainfall function that involves past daily rainfall in the exponent to account for the different lasting effect of large versus small rainfall, (4) quantitatively exploiting the lower parts of the cloud of data points, most meaningful for threshold estimation, and (5) merging the uncertainty on landslide date with the fit uncertainty in a single error estimation. We apply our approach in the western branch of the East African Rift based on landslides that occurred between 2001 and 2018, satellite rainfall estimates from the Tropical Rainfall Measurement Mission Multi-satellite Precipitation Analysis (TMPA 3B42 RT), and the continental-scale map of landslide susceptibility of Broeckx et al. (2018) and provide the first regional rainfall thresholds for landsliding in tropical Africa.

Download Full-text