Evaluate dasymetric refinement results

Evaluates dasymetric refinement results by comparing weighted and simple refinement rasters against observed census grid population (reference year 2021, treated as ground truth). Produces two evaluation datasets (one per refinement type) with absolute and percentage differences, accuracy metrics (bias, MAE, MdAE, MSE, RMSE, MAPE) for each, and a CORINE CLC raster restricted to cells overlapping census grid cells with positive observed population (R functions 'add_evaluations_to_censusgrid' and 'get_only_corineCLC_overlapping_positive_pop')

human population use case AquaINFRA dasymetric mapping evaluation accuracy metrics census grid CORINE

Inputs
Id Title Data Type Description
inputFile1_refinementWeightedReference_rds Weighted refinement raster (reference year) object Path to an .rds file containing the packed (terra::wrap) dasymetric refinement output raster from the 'weighted' refinement type, for the reference year (2021).
inputFile2_refinementSimpleReference_rds Simple refinement raster (reference year) object Path to an .rds file containing the packed (terra::wrap) dasymetric refinement output raster from the 'simple' refinement type, for the reference year (2021).
inputFile3_censusgrid_rds Census grid object Path to an .rds file containing the Eurostat census grid with observed population (column 'TOT_P_2021'), used as ground truth for evaluation.
inputFile4_corineCLC_rds CORINE CLC raster object Path to an .rds file containing the CORINE Land Cover raster (SpatRaster), used to derive the subset overlapping populated census grid cells.
Outputs
Id Title Description
evaluate_weighted Weighted refinement evaluation results Census grid with the weighted refinement's estimated population joined, including absolute difference ('dif1') and percentage difference ('dif_perc1') from observed population.
evaluate_simple Simple refinement evaluation results Census grid with the simple refinement's estimated population joined, including absolute difference ('dif2') and percentage difference ('dif_perc2') from observed population.
corineCLC_only_positive CORINE CLC overlapping positive population The CORINE CLC raster, masked to only the cells overlapping census grid cells with positive observed population.
metrics_weighted Accuracy metrics (weighted) Table of accuracy metrics for the weighted refinement: overall percentage error, overall average error, count of wrongly included/excluded cells, bias, MAE, MdAE, MSE, RMSE, and MAPE.
metrics_simple Accuracy metrics (simple) Table of accuracy metrics for the simple refinement: overall percentage error, overall average error, count of wrongly included/excluded cells, bias, MAE, MdAE, MSE, RMSE, and MAPE.

Educational resources and documentation

Jobs

Browse jobs

Links

Execution modes

  • Synchronous
  • Asynchronous