Gross vs Net CO2 emissions

1 Set Up

library(data.table)
library(hector)
library(dplyr)
library(tidyr)
library(ggplot2)

BASE_DIR <- here::here()

theme_set(theme_bw())

1.1 Objective

Run Hector under net vs gross CO2 emissions to determine does it really matter?

2 Run Hector Gross FFI Emissions

ini_list <- list.files(system.file("input", package = "hector"), pattern = "ini", full.names = TRUE)

lapply(ini_list, function(x){
  hc <- newcore(x, name = basename(gsub(pattern = "hector_|.ini", x = x, replacement = "")))
  run(hc)
  
  original_emissions <- fetchvars(hc, dates, vars = c(FFI_EMISSIONS(), DACCS_UPTAKE()))
  
  emiss_to_add <- c(0, 0.1, 0.5, 1, 1.5, 4)
  
  lapply(emiss_to_add, function(val){
    hc2 <- newcore(x, name = basename(gsub(pattern = "hector_|.ini", x = x, replacement = "")))
    
    # okay I think I can only set one variable at a time which is werid cause I think I do it all at 
    # once other places which might be bad news fuck 
    split(original_emissions, original_emissions$variable) %>% 
      lapply(function(xx){
        setvar(core = hc2, 
               dates = xx[["year"]],
               var  = unique(xx[["variable"]]), 
               values = xx[["value"]] + val, 
               unit =  unique(xx[["units"]]))
        reset(hc2)
      })
    
    run(hc2)
    
    out <- fetchvars(hc2, dates = dates, vars = vars)
    out$type <- paste0("gross ", val)
    return(out)
  }) %>%  
    bind_rows() -> 
    single_scn_different_vals
  
  return(single_scn_different_vals)
  
}) %>% 
  bind_rows() -> 
  gross_emission_ffi

When the 0 value is added to the ffi emissions we essentially running the default net emissions. For a single SSP scenario we would expect all the results (except for daccs and ffi) to be identical.

gross_emission_ffi %>% 
  filter(scenario == "ssp119") %>% 
  ggplot(aes(year, value, color = type)) + 
  geom_line() +
  facet_wrap("variable", scales = "free") + 
  labs(title = "ssp119", y = NULL, x = NULL)

gross_emission_ffi %>% 
  filter(scenario == "ssp585") %>% 
  ggplot(aes(year, value, color = type)) + 
  geom_line() +
  facet_wrap("variable", scales = "free") + 
  labs(title = "ssp585", y = NULL, x = NULL)

Doubble check that the net emissions are all the same as one another, for each SSP the net emissions shoudl be the same.

gross_emission_ffi %>%
  filter(variable %in% c(FFI_EMISSIONS(), DACCS_UPTAKE())) %>% 
  spread(variable, value) %>% 
  mutate(net = ffi_emissions - daccs_uptake) %>% 
  select(scenario, year, type, net) -> 
  net_emissions 

net_emissions %>% 
  ggplot(aes(year, net, color = scenario, linetype = type)) + 
  geom_line() + 
  labs(title = "Comparison of the net ffi emissions per test run", x = NULL, y = NULL)

The non emission inputs do look the same but now check the actual numbers, the only non-0 values should be daccs and ffi emissions.

gross_emission_ffi %>%
  spread(type, value) -> 
  wide_emission_ffi

wide_emission_ffi %>% 
  select(-scenario, -year, -variable, -units) %>% 
  apply(MARGIN = 1, FUN = sd) -> 
  sd_values

wide_emission_ffi$diff <- sd_values

wide_emission_ffi %>% 
  filter(diff > 1e-8) %>% 
  pull(variable) %>% 
  unique()

[1] "daccs_uptake"  "ffi_emissions"

So yes the ffi & daccs net and gross emissions are the same.

3 Run Hector Gross LUC Emissions

ini_list <- list.files(system.file("input", package = "hector"), pattern = "ini", full.names = TRUE)

lapply(ini_list, function(x){
  hc <- newcore(x, name = basename(gsub(pattern = "hector_|.ini", x = x, replacement = "")))
  run(hc)
  
  original_emissions <- fetchvars(hc, dates, vars = c(LUC_EMISSIONS(), LUC_UPTAKE()))
  
  emiss_to_add <- c(0, 0.1, 0.5, 1, 1.5, 4)
  
  lapply(emiss_to_add, function(val){
    hc2 <- newcore(x, name = basename(gsub(pattern = "hector_|.ini", x = x, replacement = "")))
    
    # okay I think I can only set one variable at a time which is werid cause I think I do it all at 
    # once other places which might be bad news fuck 
    split(original_emissions, original_emissions$variable) %>% 
      lapply(function(xx){
        setvar(core = hc2, 
               dates = xx[["year"]],
               var  = unique(xx[["variable"]]), 
               values = xx[["value"]] + val, 
               unit =  unique(xx[["units"]]))
        reset(hc2)
      })
    
    run(hc2)
    
    out <- fetchvars(hc2, dates = dates, vars = vars)
    out$type <- paste0("gross ", val)
    return(out)
  }) %>%  
    bind_rows() -> 
    single_scn_different_vals
  
  return(single_scn_different_vals)
  
}) %>% 
  bind_rows() -> 
  gross_emission_luc

When the 0 value is added to the luc emissions/uptake we essentially running the default net emissions. For a single SSP scenario we would expect all the results (except for emissions and uptake) to be identical.

gross_emission_luc %>% 
  filter(scenario == "ssp119") %>% 
  ggplot(aes(year, value, color = type)) + 
  geom_line() +
  facet_wrap("variable", scales = "free") + 
  labs(title = "ssp119", y = NULL, x = NULL)

gross_emission_luc %>% 
  filter(scenario == "ssp585") %>% 
  ggplot(aes(year, value, color = type)) + 
  geom_line() +
  facet_wrap("variable", scales = "free") + 
  labs(title = "ssp585", y = NULL, x = NULL)

Hmm the fact that the veg c pools are different from one another checks out intellectually, but is this a problem…

Double check that the net luc emissions/uptake are consistent for a scenario pathway. So yes the runs are set up correctly with the net luc emissions & uptake being the same.

gross_emission_luc %>%
  filter(variable %in% c(LUC_EMISSIONS(), LUC_UPTAKE())) %>% 
  spread(variable, value) %>% 
  mutate(net = luc_emissions - luc_uptake) %>% 
  select(scenario, year, type, net) -> 
  net_emissions 

net_emissions %>% 
  ggplot(aes(year, net, color = scenario, linetype = type)) + 
  geom_line() + 
  labs(title = "Comparison of the net luc emissions per test run", x = NULL, y = NULL)

Take a look at the Hector variables the only non-0 values should be luc emissions and uptake, howevver that is not the case.

gross_emission_luc %>%
  spread(type, value) -> 
  wide_emission_luc

wide_emission_luc %>% 
  select(-scenario, -year, -variable, -units) %>% 
  apply(MARGIN = 1, FUN = sd) -> 
  sd_values

wide_emission_luc$diff <- sd_values

wide_emission_luc %>% 
  filter(diff > 1e-8) %>% 
  pull(variable) %>% 
  unique()

[1] "luc_emissions"     "luc_uptake"        "CO2_concentration" "global_tas"        "NBP"               "RF_tot"            "veg_c"            

How much of a difference does this make on NBP, [CO2] and RF?

gross_emission_luc %>% 
  filter(type == "gross 0") %>% 
  select(scenario, year, variable, og_val = value) -> 
  original_output

gross_emission_luc %>% 
  filter(type != "gross 0") %>%
  right_join(original_output) %>% 
  mutate(percent_dif = 100 * (og_val - value)/value) -> 
  percent_dif_df

What is the impact (percent change)?

So it looks like it is about a 2% change in global temperature and [CO2] which eh it may or may not matter.