Contemporary Computing Technologies for Processing Big Spatiotemporal Data - Space-Time Integration in Geography and GIScience - page 333

18.5.1 2D Analytical Visualization

18.5.1.1 Pixel-Based Visualization Technique for Space-Aware Images

Pixel-based visualization uses 1 pixel to represent a data record, the whole dataset
is visualized using a flat two dimensional image. In this image, the pixel color
represents the data value, and the pixel location (sequence) in this image represents
the two dimensions of this dataset. A legend image is generated accordingly to
the value-color mapping. This technique is extremely useful for visualizing space
patterns of spatiotemporal data. For 2D data, each dataset is visualized by an image
with latitude, longitude and data value mapping to pixel row, column and color (Eq.
18.1).

Equation 18.1: Transformation of 2D dataset to image

data . latitude ; longitude / D value ! pixel . row ; col / D color . RGB /

While the generated images could be simply displayed in a separated window,
overlaying them to a geo-referenced map provides more intuitiveness. This is
generally achieved by overlaying the images to an existing map application, such
as OpenLayers, Google Maps, Bing Maps and Others. To improve the performance,
the images of frequently accessed datasets and parameters could be pre-generated.
Figure 18.6 shows the monthly average net radiation of Earth in December 1956
simulated by ModelE visualized as an RGB image. This image is overlapped on
Microsoft Bing Maps to provide geospatial reference.

18.5.1.2 Multiple Windows to Display Multiple Parameters

Simultaneously

For data with many parameters, displaying multiple parameters in multiple windows
(one parameter each window) provides an effective way for end users to easily
compare different parameters. For example, displaying net thermal radiation and
net solar radiation in two map windows for the same study area and same time
(period) enables scientists efficiently examine the relationship between the two
climatic parameters.
This technique should allow the users to customize how many parameters they
want to compare simultaneously and select the parameters, time periods and study
areas they want to compare. How to organize the data to optimize the parameters,
time periods and area selections in different datasets is a great challenge due to the
intrinsic characteristics of big data. Figure 18.7 is a screenshot showing four climatic
variables overlapped in four Google Earth windows. When users interact with one
map window, such as zoom in, the other three map windows will simultaneously
zoom to the same point for better comparing selected variables.