Getting climate data in gcamwrapper

[rodrigoms95]

Hello, what would be the string that I should pass to g.get_data() in gcamwrapper in order to get global temperature and radiative forcing?

[pralitp]

Hello @rodrigoms95

Unfortunately there remains a gap with interacting with JGCRI/hector and GCAM via gcamwrapper. In principle Hector supports interaction from R and Python directly. However there are complications to be able to leverage this with the instance of Hector which would be directly coupled within GCAM.

A couple of work around are available however:

If you need full control over setting parameters in GCAM and Hector (in addition to checking model outputs) the best approach would be to avoid using the version of Hector coupled within GCAM and instead load each as their own R / Python package and facilitate the interactions manually. The downside here is there is some non-trivial amount of processing of GCAM results to prepare them for input into Hector. We've certainly accomplished this in the past but I do not think the scripts have been published anywhere.

On the other hand if you just need to inspect climate outcomes as a result of changes in GCAM the following approach may be easier. The general idea is that each time hector runs it will be outputting its full output to a CSV file (configurable via <Value write-output="1" append-scenario-name="0" name="hector-output">../output/gcam-hector-outputstream.csv</Value>) appending new results each time. We can then take advantage to parse the climate outcomes out of that CSV file in our R / Python session. Attached is an example where we run GCAM under various assumptions via gcamwrapper and plot out the implications for total climate forcing:

library(gcamwrapper)
library(dplyr)
library(tidyr)
library(ggplot2)
library(readr)

g = create_and_initialize("configuration_ssp4.xml", "~/gcam-core/exe")

get_forcings <- function() {
  # we don't have a clean way of getting forcings directly but parsing them from the
  # hector outputstream is easy enough
  read_csv("logs/gcam-hector-outputstream.csv", comment = "#") %>%
    # TODO: Only interested in 2100 forcing?
    filter(year == 2100, component == "forcing") %>%
    # the output stream is going to accumulate results from each run so here we
    # just want the last values
    group_by(year, component, variable) %>%
    summarize(value = last(value)) %>%
    ungroup()
}

run_price <- function(initial_price, g) {
  # update the globalCO2_LTG market with the new initial_price + 3%/yr staring in 2030
  price_query = get_query("market", "price")
  get_data(g, price_query, list("market" = c("=", "globalCO2_LTG"), "year" = c(">=", 2030))) %>%
    rename(year = `market-period`) %>%
    mutate(price = initial_price * (1.03)^(year-2030)) %>%
    set_data(g, ., price_query)
  # rewind back to 2030
  run_period(g, convert_year_to_period(g, 2030))
  # and run through 2100
  run_period(g, convert_year_to_period(g, 2100))
}

run_test <- function(initial_price, g) {
  run_price(initial_price, g)
  get_forcings() %>%
    mutate(price=initial_price) %>%
    select(price, year, forcing = variable, value)
}

# loop over some prices to try and collect forcings
price_trials = seq(0, 1800, by=200)
out = lapply(price_trials, run_test, g)
mac = bind_rows(out)