iiasa/ipcc_sr15_scenario_analysis

SSP data with zero carbon prices

masahirosugiyama opened this issue · 5 comments

I found zero carbon price for SSP from the IMAGE model
(for periods such as 2030). The following is an R code.
I heard that it is a reporting error. The world average was not properly
produced. I don't know how this should be resolved and I'm writing this issue
here. (I contacted some people from the IMAGE team.)

===== R code =======
library(tidyverse)

df_org <- readxl::read_xlsx("iamc15_scenario_data_world_r1.1.xlsx",sheet=2)
df <- df_org

names(df) <- tolower(names(df))
col_names <- names(df)
year_col_names <- col_names[grep("^[0-9].*?$", col_names)]
df <- gather_(df, "period", "value", year_col_names)

df %>% filter(variable=="Price|Carbon") %>%
filter( str_detect(scenario,"SSP") & (!
str_detect(scenario,"baseline")) &
str_detect(model,"IMAGE") &
variable == "Price|Carbon"& !is.na(value) &
period == 2030) -> df_IMAGE_temp
df_IMAGE_temp

Thanks @masahirosugiyama for raising this issue.

As you can see from the rendered notebook (cell [15]) or the source code on GitHub, these scenarios were excluded from the analysis in the SR15, Section 2.5.2. Therefore, there is no error in the assessment.

Unfortunately, there was not sufficient time to contact all teams for corrections when we identified the issue during the writing of the report due to the very stringent timeline.

If the IMAGE team (or any other team affected here) can generate correct carbon prices, we can include the data in the next release of the IAMC 1.5°C Scenario Ensemble.

please see related issue #19 - we attempted to just recompute correct regional prices and this proved far more effort than currently possible

We are working on more and better features for documentation in the ixmp Scenario Explorer.

The key question for us is usability - where to display additional information in a way that it is noted by users without being too annoying or making the user interface too convoluted? Another concern is that comments might relate to the entire timeseries data for a specific variable or only to specific years - and the interface should distinguish those.

We are grateful for any practical suggestions (but please start a new issue on this).