Introduction
Avalanche focusing is one of the most complex processes involving data assimilation to predict temporal solutions and varying spatial. Forecasting involves the assimilation and prediction of information and data that help describe snowpack, weather, and stability within a given period. Conventional avalanche is carried out without direct data or numeric models by stakeholders of avalanche forecasters. The experts usually easily apply redundant and diverse data sources to solve forecasting problems. Therefore, the forecast can range from the next few hours to several days in an area. Spatial prediction varies from specific levels to local forecasts in a particular area covered by different mountain ranges.
Over the last few decades, a wide range of tools and models have been developed to help forecasters make the right decisions. The model ranges from various statistically based methods to physical models of developing a snowpack. Predictors using statistical techniques apply regression trees, discrimination analysis, and nearest neighbors (NN) (Nowak et al., 2020). Most of these stakeholders prefer using NN techniques as they assume that similar situations and events are likely to happen under the same conditions. In avalanche forecasting, the data used in offering the prediction is divided into three classes that include three classes that define the metrological factors they consider. A lot of ambiguity is realized as experts try to use different techniques to achieve their targets of making knowledge-based predictions and decisions. However, with effective visualization, such ambiguities are solved with ease. Visual analytics plays a significant role in predicting uncertainties that help avoid disasters that come with an avalanche.
Problem Statement
Some uncertainties relate to avalanche forecasting. One of the standard issues that cause these uncertainties to rise includes collaborative analysis and ambiguity. It is critical to note that ambiguity, as per forecasters, is defined as a state where several predictions and interpretations become equally plausible. The aspect means several ways of interpreting data and not data inaccuracy. Uncertainties in data are known to cause some ambiguity in data interpretation. Therefore, the project will forecast on identifying the role of visual analytics in solving the uncertainty and ambiguity issues of inaccurate avalanche forecasting.
Research Question
How does visual analytics enhance avalanche forecasting?
Research Method
The project embraced secondary data analysis, which involved getting information from other researchers to answer the research question of the role played by visual analysis in avalanche forecasting.
State-of-The-Art
Different researchers have been using diverse approaches to address the issue of ambiguity. Most of these forecasters use a glyph-based approach to offer their predictions. Glyphs usually operate at multiple scales that provide an overview of data without delay (Nowak et al., 2020). The approaches allow visual aggregation operations that are critical in summarizing data and detecting all the outliners and the trends that can be used to offer a perfect prediction (Nowak et al., 2020). The approach has been embraced by other researchers who claim that they help showcase granular data that helps reveal different particulars that are critical in avalanche forecasting. Experts prefer using bubble chart graphs where each circle represents an individual report. In this case, the stakeholders of avalanche forecasting code each prediction with different circle sizes. The methodology embraces color luminance or saturation based on the number of viewed avalanches. The circles used in this care are arranged in a packed layout to predict upcoming avalanches. These glyphs support various forms of visualization analysis hence creating channels offering some of the most accurate avalanche predictions.
Some experts have used computer assimilation to visualize data and make professional and experienced avalanche forecasts by manually testing snowpacks. The simulated snow covers tend to detect and track weak layers and identify the avalanche using a different approach (Giabbanelli & Baniukiewicz, 2019). The merit of using this approach is that it can rely on other tools, especially in cases where the local snowpack data is unavailable. The simulation is critical as they help determine the risk associated with each avalanche using an artificial release that focuses on identifying the problems associated with new snow, persistent weak layers, and wind slabs.
Comparison
The use of the glyph-based approach is different as compared to the use of computer simulation. The aspect relates to the fact that glyph-based approaches depend on various scales that help in providing an overview of the data in question without any form of delay. The approach summarizes data and helps determine the trends and outliers that might affect the art of having accurate predictions. On the other hand, computer simulations depend on manual tests of snowpacks to detect weak layers that form the basis under which the predictions are made (Giabbanelli & Baniukiewicz, 2019). The glyph method embraces different color saturation and luminance to help in forecasting (Nowak et al., 2020). At the same time, computer simulation depends on artificial data release to make predictions based on the problems associated with weak layers and the occurrence of an avalanche in a region.
Literature and Current Contributions
Avalanches Disasters
Avalanche forms part of the natural disasters that occur when there are some forms of instabilities in the snowpack. According to Horton et al. (2020), the aspect is detrimental as it causes a release of a mass of snow that slides downhill with a destructive force. These avalanches, therefore, pose significant risks to people recreating and working around mountain terrains (Nowak et al., 2020). For these reasons, avalanche forecast is mainly concerned with predicting current and future snow stabilities that might lead to the sliding of the avalanches and identifying some of the human activities that trigger the phenomenon. According to Schweizer et al. (2020), researchers and forecasters view avalanche forecasting as inductive. The new information received from each prediction helps update the models and the conditions of the entire season. Pourraz et al. (2017) pointed out that the predictors assess and characterize all the hazards associated with avalanches by answering questions such as the types of avalanches that exist in an area and their location, their likelihood of occurrence, and how large they might be. The forecasters and the researchers utilize a variety of observations and data to access the occurrence and the hazards associated with these avalanches.
Ambiguities and Uncertainties
A lot of data and observations are required for compelling predictions. The reported data include file observation of the weather conditions associated with the avalanches and the snow and avalanche activities associated with various hazards. According to Conger (2014), analyzing this data requires a lot of visual analytics that help identify the repetitive features in the data that help make the necessary predictions (Nowak et al., 2020). However, the variability of operational and the interpretation of this data largely depend on subjective judgment and a lot of discernment that helps fill the gaps in understanding. According to Helbig et al. (2015), the aspect creates room for multiple interpretations that create uncertainties and ambiguities associated with these predictions. According to Nowak et al. (2020), the challenges posed in this situation relate to the similar uncertainty issues that predictors of another phenomenon face. Therefore, the aspect necessitates visual analytics that helps analyze the data in question, let alone the observations associated with such data, to make accurate and relevant predictions.
Current Contributions
Visual analytics is outlined as the use of sophisticated methods and tools to analyze data sets using various visual representations of data. Visualizing data in charts, maps, and graphs help users identify the patterns followed by a particular phenomenon which is critical in making predictions and taking the right actionable insights Varga et al. (2020). These visualizations are critical as they help make appropriate and accurate predictions in avalanche forecasting.
Ambiguity in avalanche forecasting is addressed using visual analytics using a glyph-based approach and maps. In these maps, the elevations and different types of avalanches are reported using intervals that are not uniform (Nowak et al., 2020). Several charts and maps are developed to encode all the reports and prevent issues that relate to over-plotting. The visual analysis required arcs and critical segments to represent all the non-uniform intervals (Schindler et al., 2020). Data encoding is critical in making relatively accurate reports and is well-represented in graphs. The reports, in most cases, are presented across multiple displays that are coordinated hence supporting the standard highlighting and brushing of interactions. The art of selecting reports and highlighting them in the corresponding visualization provides a multidimensional perspective on data critical in accessing all the unstructured data that might be required to make the necessary predictions.
The visual overviews provided in these graphical maps deliberately use perceptually weak visual encodings to help make multidimensional and holistic predictions based on the points where predictions started. Visual analytics, therefore, allows a forecaster to review data quickly and make the necessary predictions based on the data provided (Nowak et al., 2020). Experts in interpreting these data help answer questions about avalanches’ hazardous nature and the likelihood of occurring their sizes, and their location in different terrains. The visual tools allow the forecasters to detect all the trends made by these avalanches. The ability to encode the information presented by these visual tools help experts in identifying and outlining the characteristic of expected avalanches. The aspect is critical as it helps evade the ambiguity and uncertainties associated with avalanche forecasting.
Future Research Challenges
A key aspect worth noting from the research study is that identifying the features associated with the sources of ambiguity is critical in designing a tool that ensures accurate data visualization. From the research study, it is clear that addressing ambiguity more explicitly is still a challenge. When an analysis is shared and the same data is revised or revisited by other experts, ambiguities are identified in how data was captured, let alone its subsequent analysis using various visualization tools. This aspect means that even after a comprehensive analysis, the data can still be re-analyzed and offer different predictions and avalanche forecasts based on the tools and techniques deployed (Nowak et al., 2020). The other challenge that future researchers will face is the problem of externalizing information. Researchers have identified that externalizing sources requires a lot of effort as it can be disruptive or part of transferring errors from one research. The art of predicting the appearance of future avalanches will be based on errors carried forward by externalizing references with bias. Therefore, unless the errors are detected at an early stage of future research, ambiguities, and uncertainties will still be viable, and they may affect the predictions given.
Climate changes have affected how weather forecasters predict future climatic conditions. The aspect means that even with proper utilization of visualization tools and analytics, climate change may affect the standard patterns assumed by most of these avalanches. Therefore, future researchers should be proactive and have rich sources that will help solve the uncertainties and ambiguity issues. Short-term analysis should be embraced to help extrapolate the errors associated with climate change.
Conclusion
Avalanches are common natural disasters that affect people who leaves near mountainous terrains. Therefore, Avalanche forecasters play a critical role in predicting their features and the effects or hazardous nature associated with each slide. The predictions are based on their forecasts of the data and observations made. The aspect creates room for multiple interpretations of this information hence bringing up a lot of ambiguities and uncertainties. However, with proper visualization techniques, the margin of error associated with ambiguity and uncertainty reduces, making accurate predictions and decisions. Therefore, future studies should analyze short-term data using critical visualization techniques. The analysis will help reduce the margin of error in making avalanche predictions and help avoid the disasters associated with their occurrence in any mountainous region.
References
Conger, S. (2014). Uncertainty and risk, merging theory with practical adaptation in avalanche hazard management. International Snow Science Workshop 2014 Proceedings, Banff, Canada. Web.
Giabbanelli, P. J., & Baniukiewicz, M. (2019). Visual analytics to identify temporal patterns and variability in simulations from cellular automata. ACM Transactions on Modeling and Computer Simulation, 29(1), 1–26. Web.
Helbig, N., van Herwijnen, A., & Jonas, T. (2015). Forecasting wet-snow avalanche probability in mountainous terrain.Cold Regions Science and Technology, 120, 219–226. Web.
Horton, S., Nowak, S., & Haegeli, P. (2020). Enhancing the operational value of snowpack models with visualization design principles. Natural Hazards and Earth System Sciences, 20(6), 1557–1572.
Nowak, S., Bartram, L., & Haegeli, P. (2020). Designing for ambiguity: Visual analytics in avalanche forecasting. 2020 IEEE Visualization Conference (VIS).
Pourraz, F., Verjus, H., & Mauris, G. (2017). Visual analytics for aiding decisions of ski touring itinerary in a risky snow avalanche environment.2017 IEEE International Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications (CIVEMSA). Web.
Schindler, M., Wu, H. Y., & Raidou, R. G. VIS 2020. Web.
Schweizer, J., Mitterer, C., Techel, F., Stoffel, A., & Reuter, B. (2020). On the relation between avalanche occurrence and avalanche danger level. The Cryosphere, 14(2), 737–750.
Varga, M., Webb, H., Krilavičius, T., & Maiden, M. (2020). Visualization and Visual Analytics in Knowledge Landscapes. Navigating Digital Health Landscapes, 297–318.