Scratch Work

Intro

See https://github.com/JGCRI/stitches/pull/9#issuecomment-796800511 for more details, but the basic idea was that we wanted to do a test run to emulate data from an existing (middle of the road) scenario with archive data from an upper and lower bound set scenarios.

Our target scenanrio is CanESM ssp245

Archive data used CanESM ssp585 and ssp126.

The challange is that there are some target values that do not have corresponding data values. Just eye balling here but it seems like this might be a problem what we want to address now or eventually.

Options * continue the nearest neighboor analysis with limited data? * add other ssp scenarios to the archive of data? * do we have some coditions we want to test for before emulating?

0. Set Up

library(data.table)
library(ggplot2)
library(magrittr)
BASE_DIR <- "/Users/dorh012/Documents/2021/stitches/notebooks/stitches_dev"

Import the archive data

archive_data <- read.csv(file.path(BASE_DIR, "inputs", "archive_data.csv"), 
                         stringsAsFactors = FALSE) %>% 
  as.data.table

Import the target data

target_data <- read.csv(file.path(BASE_DIR, "inputs", "target_data.csv"), 
                         stringsAsFactors = FALSE)

1. Plots

Compare what the data sets look like! When only the two boundry ssp scenarios are read in. With that happens when all of the data excluding the target data archive is read in.

ggplot() + 
  geom_point(data = archive_data[experiment %in% c("ssp585", "ssp126")], 
             aes(fx, dx, color = "archive data", color = "archive data")) + 
  geom_point(data = target_data, aes(fx, dx,  color = "target data")) + 
 scale_color_manual(values = c("archive data" = "grey", "target data" = "blue"))+
  theme_bw() + 
  labs(title = "Archive data only using ssp585 and ssp126", subtitle = "target scenario ssp245")

Duplicated aesthetics after name standardisation: colour

ggplot() + 
  geom_point(data = archive_data[!experiment %in% c("ssp245")], 
             aes(fx, dx, color = "archive data", color = "archive data")) + 
  geom_point(data = target_data, aes(fx, dx,  color = "target data")) + 
 scale_color_manual(values = c("archive data" = "grey", "target data" = "blue"))+
  theme_bw() + 
  labs(title = "Archive data for all spps execpt ssp345")

Duplicated aesthetics after name standardisation: colour