The overall aim of this special issue is to inform the disaster risk community about the latest developments, capabilities, and limitations regarding comprehensive multi-risk assessment with specific focus on applications in Andean countries.

Risk is a function of three factors: hazard, exposure, and vulnerability (UNDRR, 2016). Modelling hazards as the potential source of harm or damage is an essential input to risk assessment. Androsov et al. (2024) contributed with a determination of which factors are crucial for a robust prediction of tsunami hazard by simulations. They performed a comparative analysis of two numerical models (Tsunami-HySEA and TsunAWI) for the calculation of tsunamis triggered by earthquakes off the coast of Peru. In addition to this comparative analysis, the role of nonlinear terms in the numerical solutions was analysed, since tsunami run-up and inundation are highly nonlinear processes. Overall, the authors conclude that nonlinearity is crucial in the estimation of inundated areas, wave heights, and current velocities, as well as in the generation of inundation maps.

When dealing with risks, the use of scenarios to describe possible events and their effects is a common practice in disaster risk management procedures. However, the selection criteria do not always consider the consequences for exposed assets. The following two papers address the analysis of risk scenarios with two different foci: Rosero-Velásquez et al. (2024) devoted their research work to a systematic selection of earthquake scenarios rather than their selection based on expert knowledge or past events. They adopted a definition of a representative scenario, associated with return periods and loss level, for selecting such scenarios among all possible earthquake scenarios. They identified scenarios for residential buildings and power supply facilities in Valparaíso and Viña del Mar, Chile. Potential extensions to this methodology may include catalogues of multiple hazards (for example, earthquake scenarios with tsunamis), loss calculations considering indirect impacts, and high-dimensional scenarios (for example, including damage states of individual components, either buildings or power network components). Feliciano et al. (2023) performed a seismic risk assessment using 18 earthquake scenarios for Sabana Centro Province in Colombia. An exposure model was created by combining information from the Global Earthquake Model (GEM) Foundation, additional surveys, and national census data. Building types in the region were assigned corresponding fragility and vulnerability curves. The OpenQuake (OQ) hazard and risk assessment tool was used to develop a hazard model for the region of interest to estimate damage and economic losses. In addition, a social vulnerability index (SVI) based on demographic information was used to assess direct economic losses in terms of replacement costs. Results show that 10 % of all buildings considered in the region would collapse and 7 % would be severely damaged. The economic losses represent 14 % of the total replacement cost of the buildings and 21 % of the annual gross domestic product (GDP) of the region.

The understanding and inclusion of different urban structures and related vulnerabilities are essential for planning activities and practical applications. The research paper from Aguirre et al. (2024) analysed seven fire case studies in central Chile in order to enhance the understanding of risk profiles in wildland–urban interface (WUI) areas. They developed a mixed-methods approach to conducting local-scale analyses that included field surveys, remote sensing using satellite and drone imagery, and GIS-based analysis of the collected data. This resulted in a geo-referenced database of damaged and un-damaged houses, including 16 variables representing their physical features, spatial location, and availability of firefighting means. A binary classification model was then used to assess the relative importance of these attributes as indicators of vulnerability. The analysis showed that the factors of spatial arrangement have a greater impact on the prediction of damage than the structural conditions and the fire preparedness of the individual units. In particular, factors such as the proximity of the dwelling to neighbours, distance to vegetation, proximity to the boundary of dwelling groups, and distance to the origin of the fire contribute significantly to the prediction of fire damage. Although more data are required to confirm this, other structural characteristics associated with less affluent dwellings may also increase the likelihood of damage. This study has implications for the design, planning, and management of WUI areas in Chile. It can provide support in the development of risk reduction strategies for both built structures and the broader territory. The study from Cienfuegos et al. (2024) carried out a detailed analysis of possible evacuation scenarios, considering the actual state of the evacuation routes of Iquique, a Chilean city at risk of tsunamis. Iquique has a significant number of urban micro-vulnerabilities, i.e. elements that impede or complicate the flow of pedestrians. Micro-scale agent-based model (ABM) simulations were used to quantify the delay in evacuation processes due to the presence of urban micro-vulnerabilities. Furthermore, these results were integrated with high-resolution tsunami inundation simulations. This allowed an estimation of the potential number of people that the tsunami could reach under different scenarios by emulating the dynamics and behaviour of the population and the decision-making regarding the start time of the evacuation.

The previous contributions mentioned so far show the importance of having information on the number of people affected. Geiß et al. (2024) went further and projected future geospatial population distributions to quantify the number of people living in earthquake- and tsunami-prone areas of Lima and Callao, Peru, over a 3-year interval until 2035. The researchers used existing gridded population time series (such as WorldPop) that are available worldwide as open data, and they implemented machine learning models tailored for time series analysis, based on long short-term memory (LSTM) networks. They concluded that areas with high seismic hazard seismicity would see population growth of nearly 30 %. It was estimated that the population in the area endangered by tsunami inundation would increase by more than 60 %. The identification and understanding of these relationships can help urban planners and policy-makers to develop effective risk reduction strategies.

The ultimate and rather complex steps towards multi-risk assessment comprise the consideration of all interactions on the level of vulnerability for multiple and/or cascading hazards. Gómez Zapata et al. (2023) assessed the cumulative damage that would occur if a series of events, such as an earthquake and/or tsunami, occurred in the vicinity of the capital city of Lima, Peru. Their method is based on proposing state-dependent fragility functions for the respective hazard to account for pre-existing damage and on harmonizing building classes and damage states through their taxonomic characterization for the different hazards. The results show that the traditional empirical tsunami fragility functions underestimate the losses for the earthquake- and tsunami-exposed residential buildings in Lima for small-M_w events and overestimate them for large-M_w events.

By integrating these scientific findings from Gómez Zapata et al. (2023) with further research on multi-risk assessment, Schoepfer et al. (2024) have analysed and described approaches to help potential users like disaster risk managers, urban planners, or critical infrastructure operators improve their ability to cope with the increasing complexity of risks. They developed a novel tool using a modular and decentralized system architecture with distributed web services. A user interface allows for the visualization and analysis of various scenarios and their related impacts. During the development process, a strong link with stakeholders and potential users was established. In various iterations, both requirements and feedback were gathered to optimize the interactive tool with regard to its practical applicability. The system components were published on an open-source platform to ensure further development towards an operational system. Exemplified for the study area of the Lima metropolitan area, Peru, the authors conclude that research can significantly contribute to meaningful and effective multi-risk analysis when good cooperation and interaction between authorities, academia, and civil society organizations are established.

Three papers in this special issue have taken the step from prevention to preparedness. They focus on research and development towards prototypical implementation of early warning systems. Fustos-Toribio et al. (2022) developed a rainfall-induced landslide early warning system (RILEWS) for the southern regions of Chile. The preoperational system is based on a logistic model and is forced by the geomorphological and atmospheric conditions in the southern Andes. The authors have shown how the Weather Research and Forecasting (WRF) model can be integrated into the RILEWS operational system without the use of ensembles and using bias correction procedures instead. Their results indicate that bias correction without costly computer iterations is a useful alternative to numerical models. This opens up the path to implementing precipitation-based prediction models without costly computational iterations through ensembles of models. As the developed methodology can be implemented for similar climate zones in different countries and latitudes, this work has high potential for practical implementation.

Werthmann et al. (2024) share insights into the development of a landslide early warning system (LEWS) prototype developed for an informal settlement on the urban–rural border of Medellín, Colombia. A preliminary LEWS has been designed, implemented, and handed over to the government. It has gone through a test and calibration phase (i.e. development of warning thresholds, procedures for the dissemination of warnings and alerts through the sensor system). The authors provide details on the scientific and technical development of this prototype and also on the living-lab experiences and challenges. The authors demonstrate that LEWS can be accurate, affordable, and socially integrated. However, there are several factors that go beyond scientific and technical capabilities – above all, social acceptance and political will. In the same context, Sapena et al. (2023) provide further information on the associated costs for the monitoring instrumentation of the LEWS. They combined data-driven landslide susceptibility mapping and population maps to identify exposed locations. In the research article, the authors estimate the cost of monitoring sensors and the installation of a LEWS with a budget ranging from EUR 5 to 41 per protected inhabitant in Medellín, Colombia, varying based on site-specific characteristics. The information can be seen as a step forward in providing guidance to decision-makers and supporting disaster risk reduction efforts.

Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors.

Parts of the research articles were funded by the German Federal Ministry of Education and Research (BMBF) as part of the German initiative “CLIENT II – International Partnerships for Sustainable Innovations” within the framework programme “Research for Sustainable Development (FONA)”, e.g. grant no. 03G0876A-J (project RIESGOS) and grant no. 03G0905A-H (project RIESGOS 2.0).