SurfaceVolume_SumSum [m3](grid_y, grid_x): Total integrated oil volume in time for each grid cell within individual spill scenario. Integrated oil volume over time, within individual spill scenario and for every grid location, reflects oiling exposure at a given location over the period of spill duration. Adding across spill scenarios sets us up for calculating the mean, and the mean of this value across runs reflects the average oil exposure for the given set of spill scenarios. (I think median is more what we want here, to capture most likely, but I'm not sure how to best calculate this across run sets. One idea I have for addressing this issue is to create a histogram of the integrated volume for each run set would have SurfaceVolume_Sum with dimensions of (volume_bin, grid_y, grid_x)

[SusanEAllen]

[SusanEAllen]

I believe (and doing a couple of your histogram plots might help) that the oil surface volume is not Gaussian but has a long tail to high values. This issue is exactly what you are raising with the idea of a median. However, median cannot be calculated incrementally. So I suggest we log transform oil surface volume which I believe will give us a more Gaussian distribution, find the mean of that (which will be the geometric mean), which will be closer to the median (if the transformed distribution is Gaussian, it will be the median).