MassGIS just released a ton of lidar data form MA, and we are now in the position of having three datasets covering some areas.
I don't have the LAS files yet, and have just been working with their gridded data, but this has shown a huge range of quality between the datasets. Some of this comes from the way it has been processed, some of it from how the data was collected. Its a great example that all lidar is not great, but when done well you can see an amazing amount of detail.
This image is a comparison of the unsharp data of boston common from three datasets.
The first panel is the old 2002 data we have been working with for a few years, and only had access to the gridded data instead of the LAS files. You can see that there are a lot of random holes in the data that must be artifacts from the processing. This makes it so that the data has to be overly smoothed before contours can be generated, or it just creates noisy junk. In some places I was having to apply a 25m box filter to get something usable out of it, which obviously removes a lot of fine detail that may be there. Post spacing on this data is 1.25m, and they have made available the LAS files, so when we get those it will be interesting to see how much better the data could be with proper processing.
The middle panel is from the 2010 FEMA data collection. Post spacing is 1.1m, again gridded by the collection company. All the pockmarks are not there, and the edge of frog pond is crisp
The right panel is a 2009 dataset that was done by the city of boston. You can see a lot of places where there are trees, and the bare earth data has been filled. I'm assuming there were still leaves on some of these (data was flown in november), and they didn't get a lot of ground returns through the trees. While areas where there are no trees are great, and you can see a lot of detail in the paths and even the edges of the baseball fields, this will take a lot of smoothing to get contours.