Part II: The Bishamon Web Project
While the static maps provide a snapshot of the ongoing radiation exigency, a more dynamic and interactive solution was sought out to provide a greater user interface that exposes the data effectively to the public. To this end, the Bishamon team has collaborated with the Institute of Digital Research and Education, my team here at UCLA, where we have been asked to provide web based GIS and visualization solutions for their system. The goal was to provide a seamless, dynamic, interactive and accurate web based mapping platform that effectively communicated the gravity of the radiation situation by displaying the readings collected by the hybrid monitoring devices.
Early on, a critical component was on how to display the data points. While the data to be displayed is inexplicably basic (just a bunch of points on a map), the sheer quantity of points posed a daunting challenge. The “need for speed”, ie, the speed to access the information on the site from both computers and mobile devices, was of paramount importance. To achieve this, it was imperative to find a solution to instantly display hundreds of thousands of data collection points onto a web based mapping interface. Furthermore, the interactivity must allow users to pan, zoom, and quickly change the parameters of what gets displayed on the map.
Serious considerations were taken in addressing the question on whether or not to provide estimations of radiation levels on areas that were not directly measured, but inferred via GIS spatial statistical techniques. An early version of the web platform provided a surface estimation of radiation levels for areas around the actual measured coordinates using a spatial interpolation method called inverse distance weighting (IDW). IDW assigns values to un-measured locations based on the values of surrounding points, taking into account the weight (radiation level) and proximity of those points. Applying this technique to create a visual representation of the collected data reveals a seamless picture of radiation levels in a community as it estimates levels for all areas. However, this method does not take into account a variety of factors. As all measurements were conducted in an outdoor environment, usually on the road, it is not indicative of what measurements may reveal if measured indoors. By providing estimates of areas, inclusive of homes and businesses, it implies that all measurements are equal, regardless of how and where they were taken. While the estimated values were interpoloated via widely accepted statistical methods, in the end, the Bishamon team made the decision to exclude all estimations for areas not directly recorded.
Grid based approach
Still, the challenge remained on how to communicate the vast amount of data points onto a single map without sacrificing performance measures. After considering various clustering algorithms, the team elected not to display each and every reading location, but instead, to use a grid based approach, where a map canvas would be subdivided into a manageable array of cells in a grid, and each grid would be colored by the radiation level governed by the average of all points recorded within that grid.
The grids would scale depending on the zoom level chosen by the user, allowing for a uniform number of data objects to be displayed at any location and at any scale, resulting in a smooth and reactive user experience.
Bells and whistles