These are definitely better examples than you previously used. That said, I think the tax assessments don't accurately state the value of the land vs the building. I don't think OnlineGladiator is correct about the universality of >10x land:building, but I don't think you're right about ~0.2x in these neighborhoods either. I do think OG is correct about housing as an appreciating asset being primarily about the land value.

1) The nominal land cost has huge jumps up and down. From 2016->2017, all of these properties had their land value lowered 60-70%: $150k->$44k; $166k->$59k; $75k->$32k; $279k->$96k. I find it hard to believe that property prices swung that much, especially since the total appraised value remained constant.

2) The price of the land depends on the zoning of the land. A lot which can't be subdivided further or which can't have multi-unit housing will be constrained by what a single family is willing to pay for the lot. If you compare the last property ($140k land assessment / 19k ft2) to a smaller lot nearby [0] ($50k land / 2.4k ft2), you can see a huge difference in the land price per ft2: $6.7/ft2 vs $20.4/ft2. This price difference reflects legal constraints on what can be done with the lot. If the larger lot could be subdivided, it should be closer in price per ft2 to the smaller lot and would be worth ~$390k. If the total assessed value remained constant, that would be 1.25:1 for land:building.

[0] https://property.phila.gov/?p=344121100

3) Furthermore, the price of the building itself depends on the scarcity of housing. With plentiful land to build new housing, old housing loses value quicker. An old house in a constricted market has more value because it's existing housing on a scarce plot of land. Even though the value is assigned to the building, that building is more valuable because of the plot of land it's associated with.