Data visualisation plays an important role in the communication of spatial information, especially to non-experts who are unfamiliar with existing cartographic conventions. SunSpot was built with two base maps as well as a number of overlays including annual insolation, annual sun hours, surface temperatures for a selection of dates, and tree canopy cover. The ESRI Online World Topographic Base Map was used as a reference for topography, streets, buildings, addresses, and basic land cover features such as trees and parkland (ESRI, 2012g). The Bing Maps Aerial World Base Map was used as a visual reference for surface features including land cover and buildings. These layers were included as they let users visually investigate the data separately from the workflows associated with each widget. It also gives users the freedom to compare any of these layers with data or visual components from the widgets. Layers that display neighbourhood boundaries, property parcels, selected buildings, and drawn geometry are set to only be visible at specific stages of each widget’s workflow.
The dynamic nature of Web-based maps presents a cartographic design challenge. Properties such as layer order, visibility, symbology and scale levels can all change based on user-input and are not restricted to a finite number of combinations. Unlike traditional paper maps, considerations must be made for how each map element is represented and what control the user has over their display (Veregin, 2011). It is important to let the user investigate these layers as freely as possible while mitigating the possibility of user confusion or over- restriction of visualisation capabilities. Through investigation of existing web-applications that are visually and interactively effective as well as research on best cartographic practices, a number of key concepts were identified for consideration when developing the data layers
and layer interface (TYDAC, 2004; Van den Worm, 2000; Veregin, 2011; City of Boston, 2012; City of New York, 2012).
The first concept is to ensure that layer symbology intuitively communicates the data but is also complementary to other layers that might be simultaneously visible. For example, vegetation is best identified as a shade of green, but the selected colour should contrast from the colours found in the base maps so that it is easily distinguishable. The second concept is to govern which layers can be made visible at any time and which are only visible at specific stages of the workflow. This mitigates potential user confusion and directs focus to the visual components relevant to the current workflow. The third concept is to control which layers can be visible in conjunction with each other. This is accomplished by determining which map layers may commonly be used together and which can be designed to be exclusive of each other. The last concept is that layer controls should be designed in a way that these restrictions are intuitive to the user in order to minimize confusion. This also passively communicates the relationship that exists between layers.
The solar potential rasters were symbolised using a red-to-blue dichromatic colour ramp that contrasts between high and low insolation values. Clipped versions of both the yearly insolation and annual sun hour rasters were produced, showing only rooftop areas. The unclipped rasters give users the freedom to investigate solar potential and the effects of local shading on non-rooftop surfaces. The clipped rasters display the minimum amount of data necessary for rooftop investigation, making the map easier to navigate and helps the user retain visual reference of place. The difference between clipped and unclipped solar rasters and the colour gradient employed is evident in Figure 3.17.
Figure 3.17: Example (a) clipped and (b) unclipped solar layers
As described in section 3.2.4, the calibrated surface temperature data layers were adjusted to represent temperature in degrees Celsius above or below the average. This was done to improve the ability for the data to be compared across time despite varying conditions that affect surface temperature. These breaks are visualised in classes reflecting temperatures within one standard deviation, which is represented by beige, as well as two and three standard deviations above and below of the mean represented in different shades of red and green respectively. The data are then spatially interpolated using an on-the-fly bilinear interpolation technique. While this increases radiometric uncertainty between pixels, it serves dampen sharp breaks between cells, which are counter-intuitive to what the user expects to see for surface temperature. Describing the details behind this technical issue would require the introduction of additional concepts, which is counter to the objective of making the application simple and intuitive for non-experts. The difference between interpolated and non-interpolated data are displayed in Figure 3.18.
Figure 3.18: Surface temperature layer (a) with and (b) without bilinear interpolation
Tree canopy data was extracted from the previously described land cover data and displayed as a discontinuous raster that can be displayed on top of any existing layer. Represented in dark green, the colour does not interfere with the symbology of any other layer that can be visible at the same time. Unlike the other layers, the transparency level of the tree canopy layer cannot be changed. An example of the appearance of tree canopy data overlaid with the topographic base map is provided in Figure 3.19.