Lab 15 - Dasymetric Mapping

Final lab of the semester! This has been an interesting, if challenging, semester in Special Topics in GIS. We looked at a number of topics that were briefly covered in earlier courses, but this time with more depth and a critical thinking. Every now and then a totally new topic came up, like this final lab: dasymetric mapping, which is a mapping method used to estimate population in areas that do not correspond well to census enumeration units. Ancillary data such as road imperviousness or hydrology can be added to the analysis to refine how population numbers are geographically distributed (e.g not in water).

For our final lab, I first used spatial queries to determine population within a watershed basin using block groups instead of county level data. Upper, lower, and intermediate estimates of population were found using various Select by Location parameters. The problem with such a simple analysis is the inability to select only part of a block group. So I then attempted to improve the results by using areal weighting where the percent of block group that falls within the basin is the assumed same percentage of population (e.g. 30% of block group falls within basin, then 30% of population of that group falls within the basin). Results provided a similar population count to the Spatial query of finding block groups that had their centroids within the basin boundaries. 

The second part of the lab used areal weighting to determine the population of students ages 5-14 within high school attendance zones. Calculations of area before and after running the Intersect tool were used to calculate population estimates for sections of block groups that fell within each zone and the associated populations, minus areas with hydrological features. A reference population count was used to determine error/allocation accuracies. With a reference count of 54720 and a population estimate of 54719, with 10.95% of the 5-14 age population being allocated incorrectly, using areal weighting provides better results than just spatial selection used previously. 

The last part of the lab introduced imperviousness data to the analysis as imperviousness as increased imperviousness is an indicator of urban development and population. The steps were similar to before, however this time zonal statistics were used to find the total imperviousness value (0-100%) per tract within each school zone. The sum of all imperviousness values within the zone was calculated on the census tract feature class, which was then intersected with the high school zones to break up any bordering census tracts, then the population was recalculated using the new sums for each tract within each zone. With a reference count of 54720 and a population estimate of 54717, the estimate was slightly lower, however overall accuracy calculations result in 10.20% of the 5-14 age population being allocated incorrectly so using areal weighting provides slightly better results than area based areal weighting.

Screenshot of imperviousness raster layer (darker red cells are higher imperviousness values), census tracts (black) and high school zones (green) used to estimate population with dasymetric mapping.