You would have to do a series of approximations to get the length from that.
You do not know the size of the cloud. You do not know the distance to the cloud. You do not know the size of the lightining bold or the distance to it.
You do not know the size of the city.
Sometimes there might happen to be a focal distance in the metadata of pictures. However we can see clearly that the focal distance is going to be beyond the autofocus distance of any camera you are likely to encounter, so even if that information is present you are not going to be able to use it.
You can use your knowledge of typical city distances between lit buildings to estimate the distance represented by the visible city lights.
But you are still stuck not knowing the sizes or distances to cloud or lightening.
So what else do you know?
This: you can see what kind of clouds you are dealing with, and you can look up reference material on the height range of that kind of clouds: https://en.wikipedia.org/wiki/Cumulonimbus_cloud and https://www.metoffice.gov.uk/weather/learn-about/weather/types-of-weather/clouds/low-level-clouds/cumulonimbus
Then knowing the approximate height of the clouds and having estimated the distance along the foreground, and assuming that the pixels are square, measure the vertical distance to the cloud in the image, figure out how far that is according to the ground estimate... and with some trig you can figure out how far away the cloud would have to be for the vertical distance to have been shortened that much.
Then you can skeletonize the lightning, and do a bw geodesic distance measurement to get the pixel length of the branches of the lightning.
And now you should have enough information to estimate the physical length of the lightning strikes.
... Thus all would have been a lot easier if you knew the time before the thunderbolt was heard so you could estimate the distance...
(Mumble mumble... the lower parts of the lightning bolt are going to be at different angles so the distance per pixel is not actually going to stay constant, so you probably should take that into account...)
