This project aimed to make the spatial and temporal patterns of strike events in Ukraine accessible and understandable for a broad audience. During the design process we considered the content of the course, in order to keep the visual structure simple and focused, so that the map, the time filter, and the thematic layers are understandable for a broad audience, without requiring advanced GIS knowledge. Nevertheless, during the implementation we ran into some technical and visual issues revealing several limitations.
The first point relates to the temporal slider in the ArcGIS Map. In our understanding, the current time slider is not fully intuitive, since individual months are not marked with clear ticks. As a result, selecting a specific month feels imprecise. Additionally, the platform used does not allow to implement arrows to move step by step through consecutive months. Instead, each month must be selected manually with the mouse, which makes it hard to see consecutive spatial strike patterns in the map, resulting in a less intuitive month-to-month comparison. Furthermore, the expanded time slider cannot be permanently displayed in a clean way. Instead, it covers part of the Map while switching between individual months. It is also not obvious to the observer that the timeline is accessed through "Month", which is why we had to explain this application in the text over the map.
A further issue relates to the concept of change blindness. Since a lot of point data is visualized in the map, the temporal interpretation can be challenging. Smaller changes between months may be hard to notice, especially since users have to switch manually between the individual months, due to the limitations of the time scale. This causes shifts in strike density or local changes in the data to be less visible. We thought about solving this issue with an animation that automatically moves through the monthly time steps in relation to the respective magnitude of change (MOC), where periods with a larger MOC would be visualized more slowly, giving the observer more time to notice relevant differences between months. However, this would also reduce the interactivity of the map, since users would no longer be able to select individual months directly.
Furthermore, we had an issue regarding the strike density displayed as a heatmap. Our ArcGIS Online license only allowed us to visualize vector data but no raster data. However, for this project, a heat map derived from raster data would have been more intuitive and visually pleasing. As a result, we had to perform additional data processing in order to change the raster-based heat map to a heatmap-like visualization using vector-based data. This made the workflow more complex and limited the appearance of the final result. We also would habe liked to adapt the map extend when first loaded to the borders of Ukraine, so that the whole country is shown at the beginning. However, this turned out to be muh more complicated than expected, which is why the user now has to first scroll out once to see the whole Ukraine. With further improvements, this defenitely would be something we would like to adapt.
Additionally, we had some issues with the performance of the webpage itself. Due to the combination of animations, embedded Map elements, and the interactive map, the webpage can sometimes feel slightly laggy, especially when several visual elements load at the same time. We tried to address these performance issues by optimizing the webpage for Google Chrome and limiting unnecessary visual effects where possible. Although these adjustments improved the webpage performance, minor delays and some visual lags cannot be fully excluded at this stage. We are aware of these issues and are therefore aiming at improving the performance and stability of the webpage.
Overall, the ArcGIS Dashboard provided a useful framework for building an interactive map. However, we would choose a different platform for another project, since ArcGIS Dashboard came with some restrictions regarding the design and interaction options. A more flexible platform such as R Shiny or Python would likely be more suitable for future work, since they provide more possibilities and are less restrictive in terms of map design and visualization, allowing more control over the layout, the temporal interface, the visual representation of density patterns and the overall user experience.
During the process of this project, ChatGPT has been used to help with programming the webpage and to correct for grammar.
