To make Fig. 8: This uses quite a lot of generated LLT data. I have uploaded the relevant original network (after running ValleyNetwork.m) and the processed workspaces (after running ProcessNetwork.m on the 'original' workspaces), with 20 realisations for each parameter set, to the following subfolders. The original network is in folders the names of which end with fill factor, and the processed folders have an extra 'P' on the end.

For the blue circles:
SimCity/L250W50ff0_02(P), L250W50ff0_04(P), L250W50ff0_06(P), L250W50ff0_08(P), L250W50ff0_1(P), L250W50ff0_12(P), L250W50ff0_14(P), L250W50ff0_16(P).

For the red diamonds:
SimCity2/L50W50ff0_02(P), L50W50ff0_04(P), L50W50ff0_06(P), L50W50ff0_08(P), L50W50ff0_1(P), L50W50ff0_12(P), L50W50ff0_14(P), L50W50ff0_16(P).

To compute the localisation length from this data, you need to use the script GetXi.m. You run it once per point on the curves and record \xi with it's error, which then allows you to plot the curves. The plotting script is provided: fit_xiE_to_xi2D.m, which includes the fitting.