Puffin improves fine scale mapping from coarse geographic data
Puffin: Probabilistic Learning of Spatial Detail From Coarse Observations
Machine Learning
Summary
Getting detailed information like population or jobs in small areas is hard because data often comes only in big chunks. The authors created Puffin, a method that uses satellite images and statistics to guess how these big totals spread out into smaller areas. Instead of guessing one number, Puffin gives a range of possibilities and makes sure the small pieces add up exactly to the big total. This helps people make better maps when detailed data isn't available.
What this means in practice
- •For urban planners: Produce detailed maps of population and jobs within cities when only coarse census data is available.
- •For public health officials: Estimate fine-scale distributions of populations for targeted health interventions using high-level regional totals and satellite data.
Authors
Chaitanya Jobanputra, Sebastian Vollmer, Gerrit Großmann
Abstract
High-resolution socioeconomic variables are important for applications such as urban planning, public health, disaster response, and resource allocation. In practice, however, these variables are often observed only at a coarse spatial resolution. We introduce Puffin, a probabilistic framework for statistical disaggregation that raises the resolution of coarse totals using high-resolution satellite embeddings as covariates. Instead of predicting a single value for each fine-resolution subregion, Puffin learns a probability distribution and is trained through an aggregation-aware likelihood. At inference, Puffin conditions these predictions on the observed regional total and splits it among the subregions. The resulting fine-scale estimates are consistent with the observed aggregate and come with calibrated uncertainty, without requiring fine-resolution labels for training. We evaluate Puffin on German and US census, employment, and election data across population, jobs, and other count variables, and study when statistical disaggregation succeeds or fails across regions, countries, and targets.