CRIO

Risk and gap analysis for the CGIAR Research Investment Opportunities (RIO) report.

CRIO reads IMPACT model scenario results (through GFSir) and builds:

• hunger, diet (F&V, fats, sugar, calories), nutrient adequacy and blue water indicators against targets;
• pathway analysis: share of scenarios hitting each target and classification trees on socioeconomic pathway and warming;
• 4-panel indicator maps;
• target trajectories with climate-model ranges by SSP, at CG-region or country level;
• distance-to-target overviews across indicators;
• comparisons of Breeding For Tomorrow (B4T) scenarios against the baseline (heatmaps, gap closure).

Installation

# GFSir and DOORMAT from the IFPRI drat repository
install.packages(c("GFSir", "DOORMAT"), repos = c("https://ifpri.github.io/drat", getOption("repos")))

# gamstransfer ships with GAMS (see the GAMS documentation)

# CRIO from a local source folder
devtools::install("C:/EPTD/package_sources/CRIO")

Scenarios

Any scenario named SSP-GCM-RCP-suffix.gdx is supported. Current RIO set:

Set	Files	Pathway (default_pathways)
Climate ensemble	SSP1-{GFDL,IPSL,MPI,MRI,UKESM}-126-379	Green world
	SSP1-{...}-370-379	Sustainability
	SSP2-{...}-370-379	Middle of the road
	SSP3-{...}-370-379	Reference
Baseline	SSP3-MRI-370-379	crio_baseline_id
B4T	SSP3-MRI-370-{CWANA,ESA,LAC,SA,SEA,WCA,CG6}	suffix = CG region where B4T is implemented; CG6 = all CG regions

All runs are without CO2 fertilization (-379). SSP x RCP combinations not in default_pathways are named automatically ("SSPx-RCPy"). Each B4T run is compared with the climate run with the same SSP, GCM and RCP (climate_counterpart()), so B4T runs under other scenarios also work.

# Only RCP 7.0 runs, without re-reading GDXs
fulldata_70 <- select_scenarios(fulldata, rcp = "7.0")

# Rename pathways / change classifications
my_pathways <- default_pathways
my_pathways$pathway[1] <- "SSP1 low emissions"
fulldata <- relabel_scenarios(fulldata, pathways = my_pathways)

Usage

Two worked workflows ship with the package:

system.file("scripts", "risk_analysis.R", package = "CRIO")
system.file("scripts", "gap_analysis.R", package = "CRIO")

Minimal example:

library(CRIO)

files    <- list_scenarios("path/to/Scenarios/RIO")
fulldata <- load_crio_data(files, nutrients_gdx, faostat_csv, sets_xlsx)
specs    <- indicator_specs(fulldata)

# Hunger trajectories for selected countries
plot_spec_trajectories(specs$HNGR, regions = c("NGA", "ETH", "IND"), b4t_show = "own")

# B4T effects on all indicators
b4t <- build_b4t_comparisons(specs)

Package layout

File	Contents
R/constants.R	baseline id, B4T regions, food groups, rest-of-world countries
R/reference_data.R	RNIs, nutrient labels, fat calorie shares, aquatic nutrient placeholders
R/scenarios.R	listing and reading scenarios, scenario descriptor columns
R/regions.R	country to CG region lookup
R/indicators.R	get_*() indicator readers and load_crio_data()
R/specs.R	indicator specs (targets, labels)
R/pathways.R	pathway analysis and tree plots
R/maps.R	4-panel indicator maps
R/trajectories.R	trajectories and distance-to-target overviews
R/b4t.R	B4T comparisons, heatmaps, gap closure
R/export.R	RDS / CSV export

Known limitations

• Hunger: the hunger module floors the share at risk at 1%, scaled to each country's FAO 2020 numbers, so countries flatten at a country-specific minimum and scenario differences below it are not resolved.
• Nutrients: availability (supply), not intake. Aquatic food nutrient values are generic placeholders (aquatic_nutrient_per_kg).
• B4T runs exist only under SSP3-MRI.

Authors

Abhijeet Mishra (IFPRI) and Claude (Anthropic, AI coding assistant).