In nick-gauthier/tidyEOF: Tidy Empirical Orthogonal Functions and Spatial Downscaling

knitr::opts_chunk$set(echo = TRUE)
library(R.matlab)
library(tidyverse)
library(sf)

mat <- readMat('../../snow-wna/data/snow.mat')

mat

mat$snow

mat$snow %>% dim

115 * 1597

mat$snow[1,1,1]$X %>% as.matrix %>% as_tibble %>% mutate(time = 1:n()) %>%pivot_longer(-time) %>%
  ggplot(aes(time, value, group = name)) +
  geom_line(alpha = .1)

mat$snow[[1,1,1]] %>% as_tibble
mat$snow[[7,1,1]]  %>% c

mat$snow[[]] %>% as

mat$snow[7:9,1,1] %>% map_dfc(c)

dat <- mat$snow[c(2:6,10),1,1] %>% map_dfc(unlist) %>%
  mutate(id = str_sub(id, 1, -2),
         HUC = str_sub(HUC, 1, -2)) %>%
  bind_cols(map_dfc(mat$snow[7:9,1,1], c)) %>%
  st_as_sf(coords = c('lon', 'lat'), crs = 4326)
plot(dat)

years <- c(mat$snow[[11,1,1]])

swe <- mat$snow[[1,1,1]][,-1597]  # the last column is a blank duplicated one
swe[is.nan(swe)] <- NA


swe_dat <- swe %>% 
  `rownames<-`(years) %>% 
  `colnames<-`(dat$name) %>%
  as_tibble(rownames = 'year') %>%
  mutate(year = as.numeric(year)) 

colSums(swe, na.rm = TRUE)

swe %>% t %>% duplicated
swe_dat %>% select(where(function(x){sum(x, na.rm = TRUE) != 0})) %>%
  select(-year) %>%
  t %>% 
  duplicated() %>%
  sum

swe[,-1597]

Clusters

load the cluster excel file

library(readxl)

map_dfr(1:3, ~read_excel('../../snow-wna/Cluster_memberships.xlsx', sheet = .), .id = 'cluster')

so A) it looks like these clusters are for the tree ring data not the snow course data (duh, I guess) and b) the tree ring data don't have the raw site names attached. I'll have to try joining by coordinates.

clusters <- map_dfr(1:3, ~read_excel('../../snow-wna/Cluster_memberships.xlsx', sheet = .), .id = 'cluster')

clusters

so the above duplicated several rows, which means that there are rows in the clusters that have multiple matches in the coords

dat

nick-gauthier/tidyEOF documentation built on July 21, 2023, 8:25 a.m.