CLIMED FIGURE 5
================

Historical: 2005-2014
Future: 2026-2035

SSPs:
SSP1-2.6
SSP2-4.5
SSP5-8.5

Climate forcing:
Three-GCM daily median forcing for each run.
One mechanistic CLIMED simulation per period.
Not an ensemble of three CLIMED outputs.

Optional gates must be OFF in all runs.
Verify this separately against run metadata.

Historical:
D:\CliMed_Methods_Audit_Candidate_CodeOnly_2026-10-21\app\output\research_historical\climed_main_results.nc

SSP1-2.6:
D:\CliMed_Methods_Audit_Candidate_CodeOnly_2026-10-21\app\output\research_ssp26\climed_main_results.nc

SSP2-4.5:
D:\CliMed_Methods_Audit_Candidate_CodeOnly_2026-10-21\app\output\research_future\climed_main_results.nc

SSP5-8.5:
D:\CliMed_Methods_Audit_Candidate_CodeOnly_2026-10-21\app\output\research_ssp85\climed_main_results.nc

India boundaries:
D:\dengue_final_report\admin2.shp

Annual means calculated for each year;
period means are equal-weight annual means.

Common valid grid cells: 8292

Hotspot definition:
Top 10 percent among strictly positive EIR
grid-cell values for each experiment.
Values at the threshold are retained.

Spatial means are unweighted cell means.
Percent change uses the ratio of period
spatial means, not the mean of cell ratios.

Jaccard overlap = intersection / union.
Three-SSP consensus counts classification
of each grid cell in future EIR hotspots.

Difference maps share one EIR scale
across all three scenarios.
FOI supplementary maps share a separate
common FOI scale.
Colour limits use 99.5th percentile of
absolute changes; no data are clipped
for numerical statistics.

Maps use the supplied India Admin2 file.
Administrative boundaries require independent
geographical/territorial verification.

No surveillance or case-count comparison.
All raster exports: 600 dpi.
