R/zchunk_L169.nonghg_NEI_scaling_USA.R
In JGCRI/gcamdata: Build the GCAM Input Datasets
Defines functions
module_gcamusa_L169.nonghg_NEI_scaling_USA
Documented in
module_gcamusa_L169.nonghg_NEI_scaling_USA
#' module_gcamusa_L169.nonghg_NEI_scaling_USA
#'
#' Further processes the NEI data to get a full timeseries for 1990 - 2017.
#' The NEI data is scaled to EPA Tier 1 for HIGHWAY VEHICLES, OFF-HIGHWAY, and FUEL COMB. ELEC. UTIL.
#' It is linearly interpolated between NEI years (2008 - 2017), and extrapolated back to 1990 based on
#' the 2008 sector and fuels shares.
#'
#'
#' @param command API command to execute
#' @param ... other optional parameters, depending on command
#' @return Depends on \code{command}: either a vector of required inputs,
#' a vector of output names, or (if \code{command} is "MAKE") all
#' the generated outputs: \code{L169.NEI_1990_2017_GCAM_sectors_unscaled}
#' @details Non-CO2 emissions from NEI interpolated and scaled to EPA Tier 1
#' @importFrom assertthat assert_that
#' @importFrom dplyr filter mutate select group_by
#' @importFrom tidyr gather spread
#' @author MAW December 2021
module_gcamusa_L169.nonghg_NEI_scaling_USA <- function(command, ...) {
if(command == driver.DECLARE_INPUTS) {
return(c(FILE="gcam-usa/emissions/state_tier1_caps",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2008",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2011",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2014",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2017",
FILE="gcam-usa/emissions/CEDS_to_USEPA_Tier1_Mapping",
FILE="gcam-usa/emissions/CEDS_to_GCAM_sector_mapping",
FILE="gcam-usa/emissions/NEI_pollutant_mapping",
"L102.ceds_GFED_nonco2_tg_C_S_F"))
} else if(command == driver.DECLARE_OUTPUTS) {
return(c("L169.NEI_1990_2017_GCAM_sectors_unscaled"))
} else if(command == driver.MAKE) {
# Silence package check.
X <- NULL
all_data <- list(...)[[1]]
# Load required inputs
state_tier1_caps <- get_data(all_data, "gcam-usa/emissions/state_tier1_caps", strip_attributes = T)
NEI_emissions_to_CEDS_adj_2008 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2008", strip_attributes = T) %>%
mutate(year = 2008)
NEI_emissions_to_CEDS_adj_2011 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2011", strip_attributes = T) %>%
mutate(year = 2011)
NEI_emissions_to_CEDS_adj_2014 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2014", strip_attributes = T) %>%
mutate(year = 2014)
NEI_emissions_to_CEDS_adj_2017 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2017", strip_attributes = T) %>%
mutate(year = 2017)
CEDS_to_USEPA_Tier1_Mapping <- get_data(all_data, "gcam-usa/emissions/CEDS_to_USEPA_Tier1_Mapping", strip_attributes = T)
CEDS_to_GCAM_sector_mapping <- get_data(all_data, "gcam-usa/emissions/CEDS_to_GCAM_sector_mapping", strip_attributes = T)
NEI_pollutant_mapping <- get_data(all_data, "gcam-usa/emissions/NEI_pollutant_mapping", strip_attributes = T)
L102.ceds_GFED_nonco2_tg_C_S_F <- get_data(all_data, "L102.ceds_GFED_nonco2_tg_C_S_F")
# Combine NEI data
NEI_emissions_to_CEDS_adj_all <- NEI_emissions_to_CEDS_adj_2008 %>%
bind_rows(NEI_emissions_to_CEDS_adj_2011,NEI_emissions_to_CEDS_adj_2014,NEI_emissions_to_CEDS_adj_2017)
# -----------------------------------------------------------------------------
# ======================================
# ======================================
# A. Interpolating 2008 - 2017
# ======================================
# ======================================
# ======================================
# 1. data and mapping preparation
# ======================================
# 1.1 preparing EPA Tier1 state level data
state_tier1_caps %>%
# changing units of the EPA Tier1 data to match NEI (Thousand Tons to Tons)
dplyr::mutate_if( is.numeric, list(~ . * 1000) ) -> state_tier1_tons
# NA entries become 0 in order for emissions sums to work (otherwise NA is returned)
state_tier1_tons[is.na(state_tier1_tons)] <- 0
# creating a dataframe that saves the three tier1 categories that NEI will be scaled to
state_tier1_tons %>%
filter( grepl( 'HIGHWAY VEHICLES|OFF-HIGHWAY|FUEL COMB. ELEC. UTIL.', tier1_description ) ) -> state_tier1_to_scale_to
# 1.2 preparing NEI data
NEI_emissions_to_CEDS_adj_all %>%
# remove pollutants not in tier1 data
filter( !grepl( 'Carbon Dioxide|Methane|Nitrous Oxide|PM10 Filterable|PM2.5 Filterable', pollutant ) ) %>%
# remove regions not in tier1 data
filter( state %in% gcamusa.STATES ) -> NEI_refined
# Separating combustion and noncombustion emissions (broadly) and assigning "Process" fuel to noncombustion
NEI_refined_combustion <- NEI_refined %>%
# CEDS sectors that begin with "1A" are the combustion sectors
filter( grepl( "^1A", CEDS_Sector ) )
NEI_refined_noncombustion <- NEI_refined %>%
filter( !grepl( "^1A", CEDS_Sector ) ) %>%
mutate( CEDS_Fuel = "Process" ) %>%
# sum emissions by state, sector, year, and pollutant now that they all have the same fuel
group_by( state, pollutant, year, CEDS_Sector, CEDS_Fuel, unit ) %>%
mutate( emissions = sum( emissions ) ) %>%
distinct()
NEI_refined_all <- NEI_refined_combustion %>%
bind_rows( NEI_refined_noncombustion )
# ======================================
# 2. "OTHER" Tier 1 interpolation
# ======================================
# Linearly interpolating the Tier 1 categories that are not being scaled
# 2.1 preparing the data for scaling and interpolation
# mapping the CEDS extended sectors to Tier 1 categories
NEI_refined_all %>%
left_join_error_no_match( CEDS_to_USEPA_Tier1_Mapping, by = c( "CEDS_Sector" ) ) -> NEI_tier1
NEI_tier1 %>%
# removing the three Tier1 categories that are not being linearly interpolated
filter( !grepl( 'HIGHWAY VEHICLES|OFF-HIGHWAY|FUEL COMB. ELEC. UTIL.', tier1_description ) ) -> NEI_othertier1
# NAs must become 0 here, or else values will later be interpolated for sectors that did not previously exist for some years
NEI_othertier1[is.na(NEI_othertier1)] <- 0
# Interpolate data linearly between years
NEI_othertier1 %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun(year, emissions, rule = 2) ) %>%
ungroup() %>%
filter( !is.na(emissions)) -> NEI_othertier1_2008_2017
# ======================================
# 3 NEI to Tier 1 Scaling
# ======================================
# 3.1 preparing the data for scaling and interpolation
# creating a dataframe that only include the three tier1 categories of interest (referred to as "hw tiers")
NEI_tier1 %>%
# keeping the three Tier1 categories that are being scaled
filter( grepl( 'HIGHWAY VEHICLES|OFF-HIGHWAY|FUEL COMB. ELEC. UTIL.', tier1_description ) ) -> NEI_to_scale
# 3.2 making adjustments to onroad sector names
NEI_to_scale[is.na(NEI_to_scale)] <- 0 #NA entries become 0 in order for emissions sums to work (otherwise NA is returned)
# Make dataframe long, remove onroad sectors (these will be processed differently), and interpolate.
NEI_to_scale %>%
filter( !grepl("1A3b", CEDS_Sector ) ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun(year, emissions, rule = 2) ) %>%
ungroup() %>%
filter( !is.na(emissions)) -> NEI_to_scale_nonroad_long
# Any CEDS Sector that begins with "1A3b" needs to be renamed to just "1A3bi_Road", and the emissions summed.
# This is because there are discrepancies in road sector names between NEI years that result in double counting
# when we interpolate. The specific CEDS sectors for Road are not used later, so this does not impact anything.
# User can change whether or not to include dust in constants.R
if( gcamusa.DUST == FALSE ){
NEI_to_scale_renamed_road <- NEI_to_scale %>%
filter( CEDS_Fuel != "Dust",
grepl("1A3bi", CEDS_Sector ) ) %>%
group_by( state, year, pollutant, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = sum(emissions) ) %>%
distinct() %>%
mutate( "CEDS_Sector" = "1A3bi_Road" ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
distinct()
} else {
# create a dataframe that has only onroad combustion emissions
NEI_to_scale_renamed_road_no_dust <- NEI_to_scale %>%
filter( grepl("1A3bi", CEDS_Sector ) ) %>%
group_by( state, year, pollutant, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = sum(emissions) ) %>%
ungroup() %>%
mutate( "CEDS_Sector" = "1A3bi_Road" ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
distinct()
# create a dataframe that has only dust emissions, which will be assigned fuel shares
NEI_to_scale_renamed_road_dust <- NEI_to_scale %>%
filter( CEDS_Fuel == "Dust",
# there are some non-PM pollutants that have 0 emissions, remove these
grepl( "PM", pollutant ) ) %>%
group_by( state, year, pollutant, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = sum(emissions) ) %>%
ungroup() %>%
mutate( "CEDS_Sector" = "1A3bi_Road" ) %>%
distinct()
# calculate fuel shares from the combustion onroad emissions
# Note: Dan (EPA-ORD) suggested miles based if possible, but that can't be done at this step
NEI_onroad_fuel_shares <- NEI_to_scale_renamed_road_no_dust %>%
filter( grepl( "PM", pollutant ),
# remove NA emissions. The shares will be applied to years we have, and emissions interpolatedv later
!is.na( emissions ) ) %>%
group_by( state, pollutant, year ) %>%
mutate( agg_emissions = sum( emissions ),
fuel_share = emissions / agg_emissions ) %>%
select( c( state, pollutant, CEDS_Fuel, year, fuel_share ) ) %>%
ungroup()
NEI_dust_emissions_to_fuels <- NEI_to_scale_renamed_road_dust %>%
# removing CEDS fuel since this will be replaces
select( -CEDS_Fuel ) %>%
# can't use left_join_error_no_match because we are adding rows due to adding fuels
left_join( NEI_onroad_fuel_shares, by = c( "state", "pollutant", "year" ) ) %>%
# sharing out the emissions
mutate( emissions = emissions * fuel_share ) %>%
select( -fuel_share ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
distinct()
# combine the combustion and dust dataframes, and aggregate emissions
# Now, PM emissions include dust
NEI_to_scale_renamed_road <- NEI_to_scale_renamed_road_no_dust %>%
bind_rows( NEI_dust_emissions_to_fuels ) %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit, year ) %>%
mutate( emissions = sum( emissions ) ) %>%
ungroup() %>%
distinct()
}
NEI_to_scale_renamed_road %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun(year, emissions, rule = 2) ) %>%
ungroup() %>%
distinct() -> NEI_to_scale_renamed_road_NAs
# There are NAs in some instances (1A3bi_Road emissions for ND natural gas (all pollutants),
# ID Dust (EV emissions) (CO, NOx, PM2.5, PM10), CT natural gas (all), and CA natural gas (SO2) ) for all but one year.
# In this case, using approx_fun results in all NAs for that given grouping, removing emissions from
# the year we have data for. We need to apply the value we have for all years or else we get emissions
# sum issues down the line.
NEI_to_scale_renamed_road_NAs %>%
# these instances can be found where emissions are NA
filter( is.na( emissions ) ) %>%
# we want to save only the unique state / pollutant / sector / fuel combinations from here, and then
# pull them from NEI_to_scale, retaining the values we do have
distinct( state, pollutant, CEDS_Sector, CEDS_Fuel ) %>%
left_join( NEI_to_scale_renamed_road, by = c( "state", "pollutant", "CEDS_Sector", "CEDS_Fuel" ) ) -> NEI_road_all_NAs
# Remove these entries from NEI_to_scale_renamed_road_NAs
NEI_to_scale_renamed_road_NAs %>%
anti_join( NEI_road_all_NAs,
by = c("state", "pollutant", "CEDS_Sector", "CEDS_Fuel",
"tier1_description", "unit", "year") ) -> NEI_to_scale_renamed_road_no_NAs
# To the dataframe that has several NAs, we want to use the approx_fun with the "constant" method
NEI_road_all_NAs %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun_constant(year, emissions, rule = 2) ) %>%
ungroup() -> NEI_to_scale_renamed_road_constant
# bind the two road and one nonroad tables
NEI_to_scale_renamed_road_no_NAs %>%
bind_rows( NEI_to_scale_renamed_road_constant, NEI_to_scale_nonroad_long ) -> NEI_to_scale_all
# 3.3 Scaling
# calculate shares for NEI years by year, state, pollutant, and tier1 description
NEI_to_scale_all %>%
group_by( state, tier1_description, pollutant, year ) %>%
mutate( share = emissions/sum( emissions ) ) %>%
ungroup() -> NEI_shares
# change NaN shares to 0- this happens when an entry has 0 emissions
NEI_shares$share[is.na(NEI_shares$share)] <- 0
# process the state level emissions data so that we can multiply total state emissions by NEI share values
# this will be done by state, year, pollutant, and tier1 description
state_tier1_to_scale_to %>%
group_by( state, tier1_description, pollutant_code ) %>%
# getting a total emissions value for each year, by group
mutate( total_emissions2008 = sum( emissions08 ),
total_emissions2009 = sum( emissions09 ),
total_emissions2010 = sum( emissions10 ),
total_emissions2011 = sum( emissions11 ),
total_emissions2012 = sum( emissions12 ),
total_emissions2013 = sum( emissions13 ),
total_emissions2014 = sum( emissions14 ),
total_emissions2015 = sum( emissions15 ),
total_emissions2016 = sum( emissions16 ),
total_emissions2017 = sum( emissions17 ) ) %>%
select( c(state, tier1_description, pollutant_code, total_emissions2008, total_emissions2009, total_emissions2010, total_emissions2011,
total_emissions2012, total_emissions2013, total_emissions2014, total_emissions2015, total_emissions2016, total_emissions2017 ) ) %>%
gather( id, total_emissions, -c( state, tier1_description, pollutant_code ) ) %>%
ungroup() %>%
separate( id, into = c("id", "year"), sep = "emissions" ) %>%
select( -id ) %>%
mutate( year = as.integer( year ) ) -> state_tier1_pollutant_emissions
# map the emissions from EPA state Tier 1 to NEI shares, then multiply scale emissions
NEI_shares %>%
left_join_error_no_match( NEI_pollutant_mapping, by = c( "pollutant" = "NEI_pollutant" ) ) %>%
mutate( Non.CO2 = gsub( "NMVOC", "VOC", Non.CO2 ),
Non.CO2 = gsub( "NOx", "NOX", Non.CO2 ),
Non.CO2 = gsub( "PM2.5", "PM25", Non.CO2 ) ) %>%
left_join_error_no_match( state_tier1_pollutant_emissions, by = c( "state", "tier1_description", "year", "Non.CO2" = "pollutant_code" ) ) %>%
mutate( emissions = ( total_emissions * share ) ) %>%
select( -c( share, total_emissions, Non.CO2 ) ) -> NEI_scaled_2008_2017
# ======================================
# 4. Combining Tables and Formatting
# ======================================
# rebinding all tiers
NEI_othertier1_2008_2017 %>%
bind_rows( NEI_scaled_2008_2017 ) -> NEI_2008_2017
# Map to GCAM sectors
NEI_2008_2017 %>%
# use left_join because some CEDS_Sectors map to NA GCAM sectors, as these emissions are not included in GCAM.
# Note: could map to "not used in GCAM" instead so LJENM works, and then filter out
left_join( CEDS_to_GCAM_sector_mapping, by = "CEDS_Sector" ) %>%
select( -c( CEDS_sector_code, IPCC_sector_code, IPCC_sector_description, Notes ) ) %>%
ungroup() -> NEI_2008_2017_GCAM
# assign Process CEDS_Fuel to process sectors
NEI_2008_2017_GCAM %>%
filter( !grepl( "^1A", CEDS_Sector ) ) %>%
mutate( CEDS_Fuel = "Process" ) -> GCAM_Process_2008_2017
NEI_2008_2017_GCAM %>%
filter( grepl( "^1A", CEDS_Sector ) ) -> GCAM_Combustion_2008_2017
GCAM_Process_2008_2017 %>%
bind_rows( GCAM_Combustion_2008_2017 ) %>%
distinct( state, year, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit, emissions, GCAM_sector ) -> NEI_2008_2017_final
# ======================================
# ======================================
# B. Extrapolating & Scaling 1990 - 2007
# ======================================
# ======================================
# ======================================
# 1. data and mapping preparation
# ======================================
# 1.1 preparing CEDS data
# Converting from Tg to short ton to match NEI units
L102.ceds_GFED_nonco2_tg_C_S_F %>%
mutate( emissions = emissions / CONV_TST_TG * 1000 ) %>%
mutate( units = "TON" ) -> CEDS_ton
# 1.2 preparing the NEI data
NEI_2008_2017_final %>%
# filtering for 2008, as the sector / fuel / pollutant distribution from this year will be carried back
filter( year == 2008,
# calculate shares for state and tier1 description
# because the state tier1 caps don't include domestic shipping and other natural (other natural is not used in GCAM),
# we have to remove these for the shares to work properly
# other natural is not used so this is fine, domestic shipping is addressed later
!grepl( "DOMESTIC SHIPPING|OTHER NATURAL", tier1_description ) ) %>%
group_by( state, pollutant, tier1_description ) %>%
mutate( share = emissions/sum( emissions ) ) %>%
ungroup() -> NEI_2008_tier1_shares
# 1.3 preparing the EPA Tier 1 data
# now we need to process the state level emissions data, so that we can multiply EPA Tier 1 emissions by NEI share values
# this will be done by state and tier1 description
# we need a value of total emissions, by state, by year, by tier1 description
# state_tier1 caps include Wildfires and Prescribed Fires, which is not included in the NEI output
# we are also removing storage & transport for now, as no CEDS sectors are mapped to it
# so we need to remove these from the state tier1 dataframe
state_tier1_tons %>%
filter( !grepl( "WILDFIRES|PRESCRIBED FIRES|STORAGE & TRANSPORT", tier1_description ) ) %>%
group_by( state, pollutant_code, tier1_description ) %>%
mutate( total_emissions2008 = sum( emissions08 ),
total_emissions2007 = sum( emissions07 ),
total_emissions2006 = sum( emissions06 ),
total_emissions2005 = sum( emissions05 ),
total_emissions2004 = sum( emissions04 ),
total_emissions2003 = sum( emissions03 ),
total_emissions2002 = sum( emissions02 ),
total_emissions2001 = sum( emissions01 ),
total_emissions2000 = sum( emissions00 ),
total_emissions1999 = sum( emissions99 ),
total_emissions1998 = sum( emissions98 ),
total_emissions1997 = sum( emissions97 ),
total_emissions1996 = sum( emissions96 ),
total_emissions1990 = sum( emissions90 )) %>%
distinct( state, pollutant_code, tier1_description, .keep_all = T ) %>%
select( c(state, pollutant_code, tier1_description, total_emissions2007, total_emissions2006, total_emissions2005,
total_emissions2004, total_emissions2003, total_emissions2002, total_emissions2001, total_emissions2000,
total_emissions1999, total_emissions1998, total_emissions1997, total_emissions1996, total_emissions1990 ) ) %>%
gather( id, total_emissions, -c( state, pollutant_code, tier1_description ) ) %>%
ungroup() %>%
separate( id, into = c("id", "year"), sep = "emissions" ) %>%
select( -id ) %>%
mutate( year = as.integer( year ) ) %>%
# interpolate emissions for missing years
complete( nesting( state, pollutant_code, tier1_description ), year = 1990:2007 ) %>%
group_by( state, pollutant_code, tier1_description ) %>%
mutate( total_emissions = approx_fun(year, total_emissions, rule = 2) ) %>%
ungroup() %>%
# make data long again
spread( year, total_emissions ) %>%
# changing some pollutant names for mapping purposes
mutate( pollutant_code = gsub( "NOX", "NOx", pollutant_code ),
pollutant_code = gsub( "PM25", "PM2.5", pollutant_code ),
pollutant_code = gsub( "VOC", "NMVOC", pollutant_code ) ) %>%
# map to NEI pollutant names
left_join_error_no_match( NEI_pollutant_mapping, by = c( "pollutant_code" = "Non.CO2" ) ) %>%
select( -pollutant_code ) -> state_tier1_total_emissions
# ======================================
# 2. Scaling
# ======================================
# 2.1 scaling the NEI data to EPA Tier 1
# map the total emissions value to NEI_tier1_shares, then multiply total_emissions by the share
NEI_2008_tier1_shares %>%
left_join_error_no_match( state_tier1_total_emissions, by = c( "state", "pollutant" = "NEI_pollutant", "tier1_description" ) ) %>%
# now we need to multiply every EPA year column by the NEI share value from 2008
mutate( `emissions2007` = ( `2007` * share ),
`emissions2006` = ( `2006` * share ),
`emissions2005` = ( `2005` * share ),
`emissions2004` = ( `2004` * share ),
`emissions2003` = ( `2003` * share ),
`emissions2002` = ( `2002` * share ),
`emissions2001` = ( `2001` * share ),
`emissions2000` = ( `2000` * share ),
`emissions1999` = ( `1999` * share ),
`emissions1998` = ( `1998` * share ),
`emissions1997` = ( `1997` * share ),
`emissions1996` = ( `1996` * share ),
`emissions1995` = ( `1995` * share ),
`emissions1994` = ( `1994` * share ),
`emissions1993` = ( `1993` * share ),
`emissions1992` = ( `1992` * share ),
`emissions1991` = ( `1991` * share ),
`emissions1990` = ( `1990` * share ) ) %>%
# deselect columns no longer needed
select( -c( "year", "emissions", "tier1_description", "share", "1990", "1991", "1992", "1993", "1994", "1995",
"1996", "1997", "1998", "1999", "2000", "2001", "2002", "2003", "2004", "2005", "2006", "2007" ) ) -> NEI_scaled_to_state_emissions_NAs
# TODO: scaling issue: some states/tier1 descriptions have emissions that do not have emissions in NEI
# At this point, we will remove all NAs from the table, but the scaling issue should be remediated in the future
# Entries with NA emissions in 2007 are removed. If there are NAs in one year, they are in all years.
NEI_scaled_to_state_emissions_NAs %>%
filter(!is.na(emissions2007),
# remove NA GCAM_sectors. This is CEDS 5C_Open-burning-land-clearing, which has no assocaited GCAM sector.
!is.na(GCAM_sector)) -> NEI_scaled_to_state_emissions
#Other NA entries (emissions) become 0 in order for emissions sums to work (otherwise NA is returned)
NEI_scaled_to_state_emissions[is.na(NEI_scaled_to_state_emissions)] <- 0
# 2.1 scaling NEI Domestic Shipping to CEDS
# We have emissions from domestic shipping from NEI for 2008 - 2017
# It is not included in the Tier1 state level data, so we scale to CEDS emissions instead
# Filter the CEDS data for domestic shipping, the years we need, and when there are non-zero emissions
CEDS_ton %>%
filter( sector == "1A3dii_Domestic-navigation",
year >= 1990,
year <= 2007,
emissions > 0 ) %>%
group_by( year, Non.CO2 ) %>%
mutate( emissions = sum(emissions) ) %>%
ungroup() %>%
distinct( year, Non.CO2, emissions ) -> CEDS_domestic_ship
# Calculate shares of Domestic Shipping emissions by pollutant
# This way we can share out the CEDS emissions by pollutant to each state
NEI_2008_2017_final %>%
filter( year == 2008 ) %>%
left_join_error_no_match( NEI_pollutant_mapping, by = c( "pollutant" = "NEI_pollutant" ) ) %>%
filter( CEDS_Sector == "1A3dii_Domestic-navigation (shipping)" ) %>%
group_by( Non.CO2 ) %>%
mutate( sum = sum(emissions),
share = emissions/sum ) %>%
select( -c( emissions, sum, year ) ) -> NEI_2008_ship
# Join the two tables, so that total US emissions from CEDS are matched for every share
CEDS_domestic_ship %>%
# calculate PM2.5 emissions based on BC and OC
spread( Non.CO2, emissions ) %>%
mutate( "PM2.5" = ( BC + OC * gcamusa.OC_TO_OM ) * gcamusa.PM1_TO_PM2.5 ) %>%
gather( Non.CO2, emissions, -year ) %>%
# Full join and removing NA "share" and NA "emissions" because we only can scale in instances when we have pollutants in both datasets
# The NA share removes BC, OC, CH4, and N2O (not in NEI)
# The NA emissions removes PM10 (not in CEDS)
full_join( NEI_2008_ship, by = c( "Non.CO2" ) ) %>%
filter( !is.na( emissions ),
!is.na( share ) ) %>%
mutate( emissions = emissions * share ) %>%
select( -c( "tier1_description", "share", "Non.CO2" ) ) -> domestic_ship_1990_2007
# ======================================
# 4. Combining Tables and Formatting
# ======================================
NEI_scaled_to_state_emissions %>%
gather( id, emissions, -c( state, pollutant, CEDS_Sector, CEDS_Fuel, unit, GCAM_sector ) ) %>%
separate( id, into = c("id", "year"), sep = "emissions" ) %>%
select( -id ) %>%
mutate( year = as.integer( year ) ) -> NEI_1990_2007_no_domestic_ship
# assign Process CEDS_Fuel to process sectors
NEI_1990_2007_no_domestic_ship %>%
filter( !grepl( "^1A", CEDS_Sector ) ) %>%
mutate( CEDS_Fuel = "Process" ) -> GCAM_Process_1990_2007
NEI_1990_2007_no_domestic_ship %>%
filter( grepl( "^1A", CEDS_Sector ) ) -> GCAM_Combustion_1990_2007
GCAM_Process_1990_2007 %>%
bind_rows( GCAM_Combustion_1990_2007 ) %>%
distinct( state, year, pollutant, CEDS_Sector, CEDS_Fuel, unit, emissions, GCAM_sector ) %>%
# add domestic shipping
bind_rows( domestic_ship_1990_2007 ) -> NEI_1990_2007_final
# Bind all of the historic year tables together
NEI_1990_2007_final %>%
bind_rows( NEI_2008_2017_final %>% select( -c( tier1_description ) ) ) %>%
# make sure emissions are grouped by state, year, CEDS_sector, CEDS_Fuel and pollutant
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, year ) %>%
mutate( emissions = sum( emissions ) ) %>%
distinct( state, pollutant, GCAM_sector, CEDS_Sector, CEDS_Fuel, unit, year, emissions ) %>%
ungroup() -> NEI_1990_2017_GCAM_sectors
# ===================================================
# Produce outputs
NEI_1990_2017_GCAM_sectors %>%
add_title("Non-CO2 emissions by US state / CEDS Sector / CEDS Fuel / GCAM sector / pollutant / year") %>%
add_units("TON") %>%
add_comments("Psuedo-NEI emissions that have been interpolated and scaled to state level EPA Tier 1 data") %>%
add_precursors("gcam-usa/emissions/state_tier1_caps",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2008",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2011",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2014",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2017",
"gcam-usa/emissions/CEDS_to_USEPA_Tier1_Mapping",
"gcam-usa/emissions/CEDS_to_GCAM_sector_mapping",
"gcam-usa/emissions/NEI_pollutant_mapping",
"L102.ceds_GFED_nonco2_tg_C_S_F") -> L169.NEI_1990_2017_GCAM_sectors_unscaled
return_data(L169.NEI_1990_2017_GCAM_sectors_unscaled)
} else {
stop("Unknown command")
}
}
JGCRI/gcamdata documentation built on March 21, 2023, 2:19 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions module_gcamusa_L169.nonghg_NEI_scaling_USA
Documented in module_gcamusa_L169.nonghg_NEI_scaling_USA
#' module_gcamusa_L169.nonghg_NEI_scaling_USA
#'
#' Further processes the NEI data to get a full timeseries for 1990 - 2017.
#' The NEI data is scaled to EPA Tier 1 for HIGHWAY VEHICLES, OFF-HIGHWAY, and FUEL COMB. ELEC. UTIL.
#' It is linearly interpolated between NEI years (2008 - 2017), and extrapolated back to 1990 based on
#' the 2008 sector and fuels shares.
#'
#'
#' @param command API command to execute
#' @param ... other optional parameters, depending on command
#' @return Depends on \code{command}: either a vector of required inputs,
#' a vector of output names, or (if \code{command} is "MAKE") all
#' the generated outputs: \code{L169.NEI_1990_2017_GCAM_sectors_unscaled}
#' @details Non-CO2 emissions from NEI interpolated and scaled to EPA Tier 1
#' @importFrom assertthat assert_that
#' @importFrom dplyr filter mutate select group_by
#' @importFrom tidyr gather spread
#' @author MAW December 2021
module_gcamusa_L169.nonghg_NEI_scaling_USA <- function(command, ...) {
if(command == driver.DECLARE_INPUTS) {
return(c(FILE="gcam-usa/emissions/state_tier1_caps",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2008",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2011",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2014",
FILE="gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2017",
FILE="gcam-usa/emissions/CEDS_to_USEPA_Tier1_Mapping",
FILE="gcam-usa/emissions/CEDS_to_GCAM_sector_mapping",
FILE="gcam-usa/emissions/NEI_pollutant_mapping",
"L102.ceds_GFED_nonco2_tg_C_S_F"))
} else if(command == driver.DECLARE_OUTPUTS) {
return(c("L169.NEI_1990_2017_GCAM_sectors_unscaled"))
} else if(command == driver.MAKE) {
# Silence package check.
X <- NULL
all_data <- list(...)[[1]]
# Load required inputs
state_tier1_caps <- get_data(all_data, "gcam-usa/emissions/state_tier1_caps", strip_attributes = T)
NEI_emissions_to_CEDS_adj_2008 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2008", strip_attributes = T) %>%
mutate(year = 2008)
NEI_emissions_to_CEDS_adj_2011 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2011", strip_attributes = T) %>%
mutate(year = 2011)
NEI_emissions_to_CEDS_adj_2014 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2014", strip_attributes = T) %>%
mutate(year = 2014)
NEI_emissions_to_CEDS_adj_2017 <- get_data(all_data, "gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2017", strip_attributes = T) %>%
mutate(year = 2017)
CEDS_to_USEPA_Tier1_Mapping <- get_data(all_data, "gcam-usa/emissions/CEDS_to_USEPA_Tier1_Mapping", strip_attributes = T)
CEDS_to_GCAM_sector_mapping <- get_data(all_data, "gcam-usa/emissions/CEDS_to_GCAM_sector_mapping", strip_attributes = T)
NEI_pollutant_mapping <- get_data(all_data, "gcam-usa/emissions/NEI_pollutant_mapping", strip_attributes = T)
L102.ceds_GFED_nonco2_tg_C_S_F <- get_data(all_data, "L102.ceds_GFED_nonco2_tg_C_S_F")
# Combine NEI data
NEI_emissions_to_CEDS_adj_all <- NEI_emissions_to_CEDS_adj_2008 %>%
bind_rows(NEI_emissions_to_CEDS_adj_2011,NEI_emissions_to_CEDS_adj_2014,NEI_emissions_to_CEDS_adj_2017)
# -----------------------------------------------------------------------------
# ======================================
# ======================================
# A. Interpolating 2008 - 2017
# ======================================
# ======================================
# ======================================
# 1. data and mapping preparation
# ======================================
# 1.1 preparing EPA Tier1 state level data
state_tier1_caps %>%
# changing units of the EPA Tier1 data to match NEI (Thousand Tons to Tons)
dplyr::mutate_if( is.numeric, list(~ . * 1000) ) -> state_tier1_tons
# NA entries become 0 in order for emissions sums to work (otherwise NA is returned)
state_tier1_tons[is.na(state_tier1_tons)] <- 0
# creating a dataframe that saves the three tier1 categories that NEI will be scaled to
state_tier1_tons %>%
filter( grepl( 'HIGHWAY VEHICLES|OFF-HIGHWAY|FUEL COMB. ELEC. UTIL.', tier1_description ) ) -> state_tier1_to_scale_to
# 1.2 preparing NEI data
NEI_emissions_to_CEDS_adj_all %>%
# remove pollutants not in tier1 data
filter( !grepl( 'Carbon Dioxide|Methane|Nitrous Oxide|PM10 Filterable|PM2.5 Filterable', pollutant ) ) %>%
# remove regions not in tier1 data
filter( state %in% gcamusa.STATES ) -> NEI_refined
# Separating combustion and noncombustion emissions (broadly) and assigning "Process" fuel to noncombustion
NEI_refined_combustion <- NEI_refined %>%
# CEDS sectors that begin with "1A" are the combustion sectors
filter( grepl( "^1A", CEDS_Sector ) )
NEI_refined_noncombustion <- NEI_refined %>%
filter( !grepl( "^1A", CEDS_Sector ) ) %>%
mutate( CEDS_Fuel = "Process" ) %>%
# sum emissions by state, sector, year, and pollutant now that they all have the same fuel
group_by( state, pollutant, year, CEDS_Sector, CEDS_Fuel, unit ) %>%
mutate( emissions = sum( emissions ) ) %>%
distinct()
NEI_refined_all <- NEI_refined_combustion %>%
bind_rows( NEI_refined_noncombustion )
# ======================================
# 2. "OTHER" Tier 1 interpolation
# ======================================
# Linearly interpolating the Tier 1 categories that are not being scaled
# 2.1 preparing the data for scaling and interpolation
# mapping the CEDS extended sectors to Tier 1 categories
NEI_refined_all %>%
left_join_error_no_match( CEDS_to_USEPA_Tier1_Mapping, by = c( "CEDS_Sector" ) ) -> NEI_tier1
NEI_tier1 %>%
# removing the three Tier1 categories that are not being linearly interpolated
filter( !grepl( 'HIGHWAY VEHICLES|OFF-HIGHWAY|FUEL COMB. ELEC. UTIL.', tier1_description ) ) -> NEI_othertier1
# NAs must become 0 here, or else values will later be interpolated for sectors that did not previously exist for some years
NEI_othertier1[is.na(NEI_othertier1)] <- 0
# Interpolate data linearly between years
NEI_othertier1 %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun(year, emissions, rule = 2) ) %>%
ungroup() %>%
filter( !is.na(emissions)) -> NEI_othertier1_2008_2017
# ======================================
# 3 NEI to Tier 1 Scaling
# ======================================
# 3.1 preparing the data for scaling and interpolation
# creating a dataframe that only include the three tier1 categories of interest (referred to as "hw tiers")
NEI_tier1 %>%
# keeping the three Tier1 categories that are being scaled
filter( grepl( 'HIGHWAY VEHICLES|OFF-HIGHWAY|FUEL COMB. ELEC. UTIL.', tier1_description ) ) -> NEI_to_scale
# 3.2 making adjustments to onroad sector names
NEI_to_scale[is.na(NEI_to_scale)] <- 0 #NA entries become 0 in order for emissions sums to work (otherwise NA is returned)
# Make dataframe long, remove onroad sectors (these will be processed differently), and interpolate.
NEI_to_scale %>%
filter( !grepl("1A3b", CEDS_Sector ) ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun(year, emissions, rule = 2) ) %>%
ungroup() %>%
filter( !is.na(emissions)) -> NEI_to_scale_nonroad_long
# Any CEDS Sector that begins with "1A3b" needs to be renamed to just "1A3bi_Road", and the emissions summed.
# This is because there are discrepancies in road sector names between NEI years that result in double counting
# when we interpolate. The specific CEDS sectors for Road are not used later, so this does not impact anything.
# User can change whether or not to include dust in constants.R
if( gcamusa.DUST == FALSE ){
NEI_to_scale_renamed_road <- NEI_to_scale %>%
filter( CEDS_Fuel != "Dust",
grepl("1A3bi", CEDS_Sector ) ) %>%
group_by( state, year, pollutant, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = sum(emissions) ) %>%
distinct() %>%
mutate( "CEDS_Sector" = "1A3bi_Road" ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
distinct()
} else {
# create a dataframe that has only onroad combustion emissions
NEI_to_scale_renamed_road_no_dust <- NEI_to_scale %>%
filter( grepl("1A3bi", CEDS_Sector ) ) %>%
group_by( state, year, pollutant, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = sum(emissions) ) %>%
ungroup() %>%
mutate( "CEDS_Sector" = "1A3bi_Road" ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
distinct()
# create a dataframe that has only dust emissions, which will be assigned fuel shares
NEI_to_scale_renamed_road_dust <- NEI_to_scale %>%
filter( CEDS_Fuel == "Dust",
# there are some non-PM pollutants that have 0 emissions, remove these
grepl( "PM", pollutant ) ) %>%
group_by( state, year, pollutant, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = sum(emissions) ) %>%
ungroup() %>%
mutate( "CEDS_Sector" = "1A3bi_Road" ) %>%
distinct()
# calculate fuel shares from the combustion onroad emissions
# Note: Dan (EPA-ORD) suggested miles based if possible, but that can't be done at this step
NEI_onroad_fuel_shares <- NEI_to_scale_renamed_road_no_dust %>%
filter( grepl( "PM", pollutant ),
# remove NA emissions. The shares will be applied to years we have, and emissions interpolatedv later
!is.na( emissions ) ) %>%
group_by( state, pollutant, year ) %>%
mutate( agg_emissions = sum( emissions ),
fuel_share = emissions / agg_emissions ) %>%
select( c( state, pollutant, CEDS_Fuel, year, fuel_share ) ) %>%
ungroup()
NEI_dust_emissions_to_fuels <- NEI_to_scale_renamed_road_dust %>%
# removing CEDS fuel since this will be replaces
select( -CEDS_Fuel ) %>%
# can't use left_join_error_no_match because we are adding rows due to adding fuels
left_join( NEI_onroad_fuel_shares, by = c( "state", "pollutant", "year" ) ) %>%
# sharing out the emissions
mutate( emissions = emissions * fuel_share ) %>%
select( -fuel_share ) %>%
complete( nesting( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ), year = 2008:2017 ) %>%
distinct()
# combine the combustion and dust dataframes, and aggregate emissions
# Now, PM emissions include dust
NEI_to_scale_renamed_road <- NEI_to_scale_renamed_road_no_dust %>%
bind_rows( NEI_dust_emissions_to_fuels ) %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit, year ) %>%
mutate( emissions = sum( emissions ) ) %>%
ungroup() %>%
distinct()
}
NEI_to_scale_renamed_road %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun(year, emissions, rule = 2) ) %>%
ungroup() %>%
distinct() -> NEI_to_scale_renamed_road_NAs
# There are NAs in some instances (1A3bi_Road emissions for ND natural gas (all pollutants),
# ID Dust (EV emissions) (CO, NOx, PM2.5, PM10), CT natural gas (all), and CA natural gas (SO2) ) for all but one year.
# In this case, using approx_fun results in all NAs for that given grouping, removing emissions from
# the year we have data for. We need to apply the value we have for all years or else we get emissions
# sum issues down the line.
NEI_to_scale_renamed_road_NAs %>%
# these instances can be found where emissions are NA
filter( is.na( emissions ) ) %>%
# we want to save only the unique state / pollutant / sector / fuel combinations from here, and then
# pull them from NEI_to_scale, retaining the values we do have
distinct( state, pollutant, CEDS_Sector, CEDS_Fuel ) %>%
left_join( NEI_to_scale_renamed_road, by = c( "state", "pollutant", "CEDS_Sector", "CEDS_Fuel" ) ) -> NEI_road_all_NAs
# Remove these entries from NEI_to_scale_renamed_road_NAs
NEI_to_scale_renamed_road_NAs %>%
anti_join( NEI_road_all_NAs,
by = c("state", "pollutant", "CEDS_Sector", "CEDS_Fuel",
"tier1_description", "unit", "year") ) -> NEI_to_scale_renamed_road_no_NAs
# To the dataframe that has several NAs, we want to use the approx_fun with the "constant" method
NEI_road_all_NAs %>%
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit ) %>%
mutate( emissions = approx_fun_constant(year, emissions, rule = 2) ) %>%
ungroup() -> NEI_to_scale_renamed_road_constant
# bind the two road and one nonroad tables
NEI_to_scale_renamed_road_no_NAs %>%
bind_rows( NEI_to_scale_renamed_road_constant, NEI_to_scale_nonroad_long ) -> NEI_to_scale_all
# 3.3 Scaling
# calculate shares for NEI years by year, state, pollutant, and tier1 description
NEI_to_scale_all %>%
group_by( state, tier1_description, pollutant, year ) %>%
mutate( share = emissions/sum( emissions ) ) %>%
ungroup() -> NEI_shares
# change NaN shares to 0- this happens when an entry has 0 emissions
NEI_shares$share[is.na(NEI_shares$share)] <- 0
# process the state level emissions data so that we can multiply total state emissions by NEI share values
# this will be done by state, year, pollutant, and tier1 description
state_tier1_to_scale_to %>%
group_by( state, tier1_description, pollutant_code ) %>%
# getting a total emissions value for each year, by group
mutate( total_emissions2008 = sum( emissions08 ),
total_emissions2009 = sum( emissions09 ),
total_emissions2010 = sum( emissions10 ),
total_emissions2011 = sum( emissions11 ),
total_emissions2012 = sum( emissions12 ),
total_emissions2013 = sum( emissions13 ),
total_emissions2014 = sum( emissions14 ),
total_emissions2015 = sum( emissions15 ),
total_emissions2016 = sum( emissions16 ),
total_emissions2017 = sum( emissions17 ) ) %>%
select( c(state, tier1_description, pollutant_code, total_emissions2008, total_emissions2009, total_emissions2010, total_emissions2011,
total_emissions2012, total_emissions2013, total_emissions2014, total_emissions2015, total_emissions2016, total_emissions2017 ) ) %>%
gather( id, total_emissions, -c( state, tier1_description, pollutant_code ) ) %>%
ungroup() %>%
separate( id, into = c("id", "year"), sep = "emissions" ) %>%
select( -id ) %>%
mutate( year = as.integer( year ) ) -> state_tier1_pollutant_emissions
# map the emissions from EPA state Tier 1 to NEI shares, then multiply scale emissions
NEI_shares %>%
left_join_error_no_match( NEI_pollutant_mapping, by = c( "pollutant" = "NEI_pollutant" ) ) %>%
mutate( Non.CO2 = gsub( "NMVOC", "VOC", Non.CO2 ),
Non.CO2 = gsub( "NOx", "NOX", Non.CO2 ),
Non.CO2 = gsub( "PM2.5", "PM25", Non.CO2 ) ) %>%
left_join_error_no_match( state_tier1_pollutant_emissions, by = c( "state", "tier1_description", "year", "Non.CO2" = "pollutant_code" ) ) %>%
mutate( emissions = ( total_emissions * share ) ) %>%
select( -c( share, total_emissions, Non.CO2 ) ) -> NEI_scaled_2008_2017
# ======================================
# 4. Combining Tables and Formatting
# ======================================
# rebinding all tiers
NEI_othertier1_2008_2017 %>%
bind_rows( NEI_scaled_2008_2017 ) -> NEI_2008_2017
# Map to GCAM sectors
NEI_2008_2017 %>%
# use left_join because some CEDS_Sectors map to NA GCAM sectors, as these emissions are not included in GCAM.
# Note: could map to "not used in GCAM" instead so LJENM works, and then filter out
left_join( CEDS_to_GCAM_sector_mapping, by = "CEDS_Sector" ) %>%
select( -c( CEDS_sector_code, IPCC_sector_code, IPCC_sector_description, Notes ) ) %>%
ungroup() -> NEI_2008_2017_GCAM
# assign Process CEDS_Fuel to process sectors
NEI_2008_2017_GCAM %>%
filter( !grepl( "^1A", CEDS_Sector ) ) %>%
mutate( CEDS_Fuel = "Process" ) -> GCAM_Process_2008_2017
NEI_2008_2017_GCAM %>%
filter( grepl( "^1A", CEDS_Sector ) ) -> GCAM_Combustion_2008_2017
GCAM_Process_2008_2017 %>%
bind_rows( GCAM_Combustion_2008_2017 ) %>%
distinct( state, year, pollutant, CEDS_Sector, CEDS_Fuel, tier1_description, unit, emissions, GCAM_sector ) -> NEI_2008_2017_final
# ======================================
# ======================================
# B. Extrapolating & Scaling 1990 - 2007
# ======================================
# ======================================
# ======================================
# 1. data and mapping preparation
# ======================================
# 1.1 preparing CEDS data
# Converting from Tg to short ton to match NEI units
L102.ceds_GFED_nonco2_tg_C_S_F %>%
mutate( emissions = emissions / CONV_TST_TG * 1000 ) %>%
mutate( units = "TON" ) -> CEDS_ton
# 1.2 preparing the NEI data
NEI_2008_2017_final %>%
# filtering for 2008, as the sector / fuel / pollutant distribution from this year will be carried back
filter( year == 2008,
# calculate shares for state and tier1 description
# because the state tier1 caps don't include domestic shipping and other natural (other natural is not used in GCAM),
# we have to remove these for the shares to work properly
# other natural is not used so this is fine, domestic shipping is addressed later
!grepl( "DOMESTIC SHIPPING|OTHER NATURAL", tier1_description ) ) %>%
group_by( state, pollutant, tier1_description ) %>%
mutate( share = emissions/sum( emissions ) ) %>%
ungroup() -> NEI_2008_tier1_shares
# 1.3 preparing the EPA Tier 1 data
# now we need to process the state level emissions data, so that we can multiply EPA Tier 1 emissions by NEI share values
# this will be done by state and tier1 description
# we need a value of total emissions, by state, by year, by tier1 description
# state_tier1 caps include Wildfires and Prescribed Fires, which is not included in the NEI output
# we are also removing storage & transport for now, as no CEDS sectors are mapped to it
# so we need to remove these from the state tier1 dataframe
state_tier1_tons %>%
filter( !grepl( "WILDFIRES|PRESCRIBED FIRES|STORAGE & TRANSPORT", tier1_description ) ) %>%
group_by( state, pollutant_code, tier1_description ) %>%
mutate( total_emissions2008 = sum( emissions08 ),
total_emissions2007 = sum( emissions07 ),
total_emissions2006 = sum( emissions06 ),
total_emissions2005 = sum( emissions05 ),
total_emissions2004 = sum( emissions04 ),
total_emissions2003 = sum( emissions03 ),
total_emissions2002 = sum( emissions02 ),
total_emissions2001 = sum( emissions01 ),
total_emissions2000 = sum( emissions00 ),
total_emissions1999 = sum( emissions99 ),
total_emissions1998 = sum( emissions98 ),
total_emissions1997 = sum( emissions97 ),
total_emissions1996 = sum( emissions96 ),
total_emissions1990 = sum( emissions90 )) %>%
distinct( state, pollutant_code, tier1_description, .keep_all = T ) %>%
select( c(state, pollutant_code, tier1_description, total_emissions2007, total_emissions2006, total_emissions2005,
total_emissions2004, total_emissions2003, total_emissions2002, total_emissions2001, total_emissions2000,
total_emissions1999, total_emissions1998, total_emissions1997, total_emissions1996, total_emissions1990 ) ) %>%
gather( id, total_emissions, -c( state, pollutant_code, tier1_description ) ) %>%
ungroup() %>%
separate( id, into = c("id", "year"), sep = "emissions" ) %>%
select( -id ) %>%
mutate( year = as.integer( year ) ) %>%
# interpolate emissions for missing years
complete( nesting( state, pollutant_code, tier1_description ), year = 1990:2007 ) %>%
group_by( state, pollutant_code, tier1_description ) %>%
mutate( total_emissions = approx_fun(year, total_emissions, rule = 2) ) %>%
ungroup() %>%
# make data long again
spread( year, total_emissions ) %>%
# changing some pollutant names for mapping purposes
mutate( pollutant_code = gsub( "NOX", "NOx", pollutant_code ),
pollutant_code = gsub( "PM25", "PM2.5", pollutant_code ),
pollutant_code = gsub( "VOC", "NMVOC", pollutant_code ) ) %>%
# map to NEI pollutant names
left_join_error_no_match( NEI_pollutant_mapping, by = c( "pollutant_code" = "Non.CO2" ) ) %>%
select( -pollutant_code ) -> state_tier1_total_emissions
# ======================================
# 2. Scaling
# ======================================
# 2.1 scaling the NEI data to EPA Tier 1
# map the total emissions value to NEI_tier1_shares, then multiply total_emissions by the share
NEI_2008_tier1_shares %>%
left_join_error_no_match( state_tier1_total_emissions, by = c( "state", "pollutant" = "NEI_pollutant", "tier1_description" ) ) %>%
# now we need to multiply every EPA year column by the NEI share value from 2008
mutate( `emissions2007` = ( `2007` * share ),
`emissions2006` = ( `2006` * share ),
`emissions2005` = ( `2005` * share ),
`emissions2004` = ( `2004` * share ),
`emissions2003` = ( `2003` * share ),
`emissions2002` = ( `2002` * share ),
`emissions2001` = ( `2001` * share ),
`emissions2000` = ( `2000` * share ),
`emissions1999` = ( `1999` * share ),
`emissions1998` = ( `1998` * share ),
`emissions1997` = ( `1997` * share ),
`emissions1996` = ( `1996` * share ),
`emissions1995` = ( `1995` * share ),
`emissions1994` = ( `1994` * share ),
`emissions1993` = ( `1993` * share ),
`emissions1992` = ( `1992` * share ),
`emissions1991` = ( `1991` * share ),
`emissions1990` = ( `1990` * share ) ) %>%
# deselect columns no longer needed
select( -c( "year", "emissions", "tier1_description", "share", "1990", "1991", "1992", "1993", "1994", "1995",
"1996", "1997", "1998", "1999", "2000", "2001", "2002", "2003", "2004", "2005", "2006", "2007" ) ) -> NEI_scaled_to_state_emissions_NAs
# TODO: scaling issue: some states/tier1 descriptions have emissions that do not have emissions in NEI
# At this point, we will remove all NAs from the table, but the scaling issue should be remediated in the future
# Entries with NA emissions in 2007 are removed. If there are NAs in one year, they are in all years.
NEI_scaled_to_state_emissions_NAs %>%
filter(!is.na(emissions2007),
# remove NA GCAM_sectors. This is CEDS 5C_Open-burning-land-clearing, which has no assocaited GCAM sector.
!is.na(GCAM_sector)) -> NEI_scaled_to_state_emissions
#Other NA entries (emissions) become 0 in order for emissions sums to work (otherwise NA is returned)
NEI_scaled_to_state_emissions[is.na(NEI_scaled_to_state_emissions)] <- 0
# 2.1 scaling NEI Domestic Shipping to CEDS
# We have emissions from domestic shipping from NEI for 2008 - 2017
# It is not included in the Tier1 state level data, so we scale to CEDS emissions instead
# Filter the CEDS data for domestic shipping, the years we need, and when there are non-zero emissions
CEDS_ton %>%
filter( sector == "1A3dii_Domestic-navigation",
year >= 1990,
year <= 2007,
emissions > 0 ) %>%
group_by( year, Non.CO2 ) %>%
mutate( emissions = sum(emissions) ) %>%
ungroup() %>%
distinct( year, Non.CO2, emissions ) -> CEDS_domestic_ship
# Calculate shares of Domestic Shipping emissions by pollutant
# This way we can share out the CEDS emissions by pollutant to each state
NEI_2008_2017_final %>%
filter( year == 2008 ) %>%
left_join_error_no_match( NEI_pollutant_mapping, by = c( "pollutant" = "NEI_pollutant" ) ) %>%
filter( CEDS_Sector == "1A3dii_Domestic-navigation (shipping)" ) %>%
group_by( Non.CO2 ) %>%
mutate( sum = sum(emissions),
share = emissions/sum ) %>%
select( -c( emissions, sum, year ) ) -> NEI_2008_ship
# Join the two tables, so that total US emissions from CEDS are matched for every share
CEDS_domestic_ship %>%
# calculate PM2.5 emissions based on BC and OC
spread( Non.CO2, emissions ) %>%
mutate( "PM2.5" = ( BC + OC * gcamusa.OC_TO_OM ) * gcamusa.PM1_TO_PM2.5 ) %>%
gather( Non.CO2, emissions, -year ) %>%
# Full join and removing NA "share" and NA "emissions" because we only can scale in instances when we have pollutants in both datasets
# The NA share removes BC, OC, CH4, and N2O (not in NEI)
# The NA emissions removes PM10 (not in CEDS)
full_join( NEI_2008_ship, by = c( "Non.CO2" ) ) %>%
filter( !is.na( emissions ),
!is.na( share ) ) %>%
mutate( emissions = emissions * share ) %>%
select( -c( "tier1_description", "share", "Non.CO2" ) ) -> domestic_ship_1990_2007
# ======================================
# 4. Combining Tables and Formatting
# ======================================
NEI_scaled_to_state_emissions %>%
gather( id, emissions, -c( state, pollutant, CEDS_Sector, CEDS_Fuel, unit, GCAM_sector ) ) %>%
separate( id, into = c("id", "year"), sep = "emissions" ) %>%
select( -id ) %>%
mutate( year = as.integer( year ) ) -> NEI_1990_2007_no_domestic_ship
# assign Process CEDS_Fuel to process sectors
NEI_1990_2007_no_domestic_ship %>%
filter( !grepl( "^1A", CEDS_Sector ) ) %>%
mutate( CEDS_Fuel = "Process" ) -> GCAM_Process_1990_2007
NEI_1990_2007_no_domestic_ship %>%
filter( grepl( "^1A", CEDS_Sector ) ) -> GCAM_Combustion_1990_2007
GCAM_Process_1990_2007 %>%
bind_rows( GCAM_Combustion_1990_2007 ) %>%
distinct( state, year, pollutant, CEDS_Sector, CEDS_Fuel, unit, emissions, GCAM_sector ) %>%
# add domestic shipping
bind_rows( domestic_ship_1990_2007 ) -> NEI_1990_2007_final
# Bind all of the historic year tables together
NEI_1990_2007_final %>%
bind_rows( NEI_2008_2017_final %>% select( -c( tier1_description ) ) ) %>%
# make sure emissions are grouped by state, year, CEDS_sector, CEDS_Fuel and pollutant
group_by( state, pollutant, CEDS_Sector, CEDS_Fuel, year ) %>%
mutate( emissions = sum( emissions ) ) %>%
distinct( state, pollutant, GCAM_sector, CEDS_Sector, CEDS_Fuel, unit, year, emissions ) %>%
ungroup() -> NEI_1990_2017_GCAM_sectors
# ===================================================
# Produce outputs
NEI_1990_2017_GCAM_sectors %>%
add_title("Non-CO2 emissions by US state / CEDS Sector / CEDS Fuel / GCAM sector / pollutant / year") %>%
add_units("TON") %>%
add_comments("Psuedo-NEI emissions that have been interpolated and scaled to state level EPA Tier 1 data") %>%
add_precursors("gcam-usa/emissions/state_tier1_caps",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2008",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2011",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2014",
"gcam-usa/emissions/NEI_emissions_to_CEDS_adj_2017",
"gcam-usa/emissions/CEDS_to_USEPA_Tier1_Mapping",
"gcam-usa/emissions/CEDS_to_GCAM_sector_mapping",
"gcam-usa/emissions/NEI_pollutant_mapping",
"L102.ceds_GFED_nonco2_tg_C_S_F") -> L169.NEI_1990_2017_GCAM_sectors_unscaled
return_data(L169.NEI_1990_2017_GCAM_sectors_unscaled)
} else {
stop("Unknown command")
}
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.