The Walk Accessibility map gives an accessibility score to each census tract in the state. The score is computed as a distance-weighted sum of total jobs or households (toggle between job and labor accessibility using the layer switcher in the top right corner) that can be reached in a 30 minute walk. This is a slightly modified version of the classic Hansen gravity measure because distances outside of a given threshold (here, 30 mins) are assigned a weight of zero. Results are only displayed for Maryland, but the calculations include jobs and housing data (from LEHD) from neighboring states. Clicking on any census tract will display its accessibility scores, but keep in mind that because they are sums weighted by an exponential decay function, the units are rather meaningless (which is why there is no legend). The color ramp is based on a logarithmic scale.
Also note that although the maps are displayed at the census tract level, accessibility is actually computed for each street intersection using the OpenStreetMap network and the fabulous pandana network analyis library from Synthicity. This means that any geographic unit could be used to display the maps (editors note: I initially tried to create these maps using census blocks for greater detail, but the 30mb topojson file was too complex and kept crashing my browser). Using pandana's network aggregation queries to calculate accessibility helps create smooth maps that are better reflections of reality by avoiding the dreaded 'modifiable areal unit problem'.
Bike accessibility scores are computed exactly the same as the walk access scores, although they assume that a bike can travel faster (average speed is assumed to be 15 km/h for biking and 5 km/h for walking).