DICEModel.jl

Documentation for DICEModel.jl, a Julia/JuMP port of the Nordhaus' DICE (Dynamic Integrated Climate-Economy model) and Nordhaus/Yang RICE (Regional Integrated Climate-Economy model) models.

This package currently implements "exactly" (in structure, data and hence output) the DICE2023-b-4-3-10.gms gams version and generalize it introducing a regional dimension that can be used to implement RICE-like models.

This program and output is not the original Barrage/Nordhaus version, which is currently only available in GAMS. Also, the RICE implementation here uses exogenous welfare weighths, default to equal ones, not to the Nagishi ones.

While DICEModel.jl is implemented in Julia it can easily be used in Python or R using the JuliaCall (Python) and JuliaCall (R) packages respectively.

Available documentation

To create the struct holding the parameters:

DICEParameters: List all the available parameters to use with the run_dice function
DICE2023(;kwargs...): Instantiate a DICEParameters struct with defaults parameters to DICE2023 (single region)
DICE2023_NREG(n;kwargs...): Build the parameters for a DICE2023 world partitioned in n equal regions (unless parameters are overrided)
RICE2023(;kwargs...): Build parameters calibrated to have the DICE2023 world totals and the 12-regions RICE2020 regional distribution.

To run the model:

run_dice_scenario(scenario_name): run one of the "official" 10 scenarios (browse code);
run_dice(pars;optimizer,bounds) or run_dice(;optimizer,bounds,kwargs...): run DICE with custom solver engine (and eventually options), custom variable constraints (bounds) or custom parameters (browse code);

Results

The output of the model (results) is a named tuple. Use keys(results) to find the available information (or just look at the source code) and results.VARIABLEX to obtain the values.

A summary of the main results, and a comparision with the official Barrage/Nordhaus DICE2023 version, is available on this page While DICE2023 output matches very well the "official" output, the same is not yet true for RICE. Some preliminary results of the multi-regional version of the model are available here, but they still clearly have some calibration issues.

An Excel version of all the results for DICE2023 is available here.

Example

using Pkg
Pkg.activate(@__DIR__)
Pkg.add(["DICEModel","Plots"]) # run only once, then comment out
using DICEModel, Plots

# CB Optimal scenario...
res_cbopt    = run_dice_scenario("cbopt")

# Base scenario...
res_base    = run_dice_scenario("base")

# Paris "extended" scenario...
tidx = 1:81
# upper limit to emissions mitigation rate
miuup = @. min( 0.05 + 0.04*(tidx-1) - 0.01*max(0,tidx-5)  ,1.00) 
res_parisext = run_dice(miuup = miuup) # or simply: run_dice_scenario("parisext")

# Max 2 °C scenario...
res_t2c = run_dice(bounds = Dict("TATM"=>("<=",2.0))) # or simply: run_dice_scenario("t2c")

# Plots
times = res_cbopt.times

# CO2 emissions plot...
plot(times[1:11],res_cbopt.ECO2[1:11],ylim=(0,70), title="CO₂ emissions",ylabel="GtCO₂/yr",label="C/B optimal", color=:blue4, markershape=:circle, markercolor=:white)
plot!(times[1:11],res_base.ECO2[1:11], label="Base", colour=:goldenrod3, markershape=:circle, markercolor=:goldenrod3)
plot!(times[1:11],res_parisext.ECO2[1:11], label="Paris ext", colour=:red, linestyle=:dash)
plot!(times[1:11],res_t2c.ECO2[1:11], label="T < 2 °C", colour=:green, markershape=:cross, markercolor=:green)

# Carbon price plot...
plot(times[1:9],res_cbopt.CPRICE[1:9],ylim=(0,300), title="Carbon price",ylabel="2019\$ / t tCO₂",label="C/B optimal", color=:blue4, markershape=:circle, markercolor=:white)
plot!(times[1:9],res_base.CPRICE[1:9], label="Base", colour=:goldenrod3, markershape=:circle, markercolor=:goldenrod3)
plot!(times[1:9],res_parisext.CPRICE[1:9], label="Paris ext", colour=:red, linestyle=:dash)
plot!(times[1:9],res_t2c.CPRICE[1:9], label="T < 2 °C", colour=:green, markershape=:cross, markercolor=:green)

Licence

The licence of the original GAMS code has never being specified. The Julia port itself (and only that) is MIT.