Some OSM places included in Tilezen's places layer have a null population which mean's they get assigned a population_rank of 0.
This can make it difficult to style a map using population_rank alone, instead you have to dip into the kind_detail values to understand the relative importance of cities say 20k default, towns 5 to 10k default, villages, hamlets, isolated settlements and so on. The population_rank should be more reliable.
We can accomplish this by backfilling population at query time when it's missing in the OSM data, and then the population_rank transform will calculate a value based on that estimate.
We should choose a lowish estimate for that type of feature, so medium and large population features would get a larger size than a unknown population size. And the range of estimate values for city, town, village, hamlet and so on should gradually decrease in estimate so null population sizes between them fade off logically, too.
Some OSM places included in Tilezen's places layer have a
nullpopulation which mean's they get assigned a population_rank of0.This can make it difficult to style a map using population_rank alone, instead you have to dip into the kind_detail values to understand the relative importance of cities say 20k default, towns 5 to 10k default, villages, hamlets, isolated settlements and so on. The population_rank should be more reliable.
We can accomplish this by backfilling population at query time when it's missing in the OSM data, and then the population_rank transform will calculate a value based on that estimate.
We should choose a lowish estimate for that type of feature, so medium and large population features would get a larger size than a unknown population size. And the range of estimate values for city, town, village, hamlet and so on should gradually decrease in estimate so null population sizes between them fade off logically, too.