From af9a707ae0ef3a1b25717b382269b147dee4d458 Mon Sep 17 00:00:00 2001
From: silasprincipe <53846571+silasprincipe@users.noreply.github.com>
Date: Mon, 4 Mar 2024 22:02:28 +0100
Subject: [PATCH] Corrected for use of mean

---
 codes/analysis_mhw_glorys.R      | 10 ++++++++--
 codes/analysis_occurrence_data.R |  3 +--
 codes/analysis_sst_species.R     |  2 +-
 3 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/codes/analysis_mhw_glorys.R b/codes/analysis_mhw_glorys.R
index 4263901..04a98ce 100644
--- a/codes/analysis_mhw_glorys.R
+++ b/codes/analysis_mhw_glorys.R
@@ -19,6 +19,7 @@ library(furrr)
 # Settings
 outdir <- "results/mhw/"
 fs::dir_create(outdir)
+set.seed(2024)
 
 
 # Load data ----
@@ -66,7 +67,7 @@ plan(multisession, workers = 4)
 
 retrieve_mhw <- function(id) {
   
-  depth_val <- 0
+  depth_val <- 0 # To try other depths change here
   
   mhs_sel <- mhs_info[mhs_info$area == id,]
   
@@ -90,6 +91,11 @@ retrieve_mhw <- function(id) {
     
     valid_cells <- as.data.frame(tvar[[1]], cell = T, xy = T)[,1:3]
     
+    if (nrow(valid_cells) > 2000) {
+      sa <- sample(1:nrow(valid_cells), 2000)
+      valid_cells <- valid_cells[sa,]
+    }
+    
     for (i in 1:nrow(valid_cells)) {
       wdf <- unname(unlist(tvar[valid_cells$cell[i]]))
       wdf <- data.frame(t = dates, temp = wdf)
@@ -115,6 +121,6 @@ retrieve_mhw <- function(id) {
   
 }
 
-future_map(unique(all_sites_filt$area), retrieve_mhw, .progress = T)
+future_map(unique(all_sites_filt$area), retrieve_mhw, .progress = T, .options = furrr_options(seed = T))
 
 ### END
\ No newline at end of file
diff --git a/codes/analysis_occurrence_data.R b/codes/analysis_occurrence_data.R
index 9e4c94a..06ae69f 100644
--- a/codes/analysis_occurrence_data.R
+++ b/codes/analysis_occurrence_data.R
@@ -21,13 +21,12 @@ speciestherm<-read_parquet("results/species_tsummaries.parquet")
 
 speciesthermdt<-data.table(speciestherm)
 
-speciesthermsst<-speciesthermdt[depth=="depthsurf" & variant=="max",,]
+speciesthermsst<-speciesthermdt[depth=="depthsurf" & variant=="mean",,]
 speciesthermsstc<-dcast(speciesthermdt,species~metric,value.var = "value",fun=mean)
 
 ## Site by species occurrence table
 
 speciesthermsite<-read_parquet("results/tsummaries_aggregated.parquet")
-speciesthermsite$where[is.na(speciesthermsite$where)] <- "Both" # correct NA that should be both
 
 speciesthermsitedt<-data.table(speciesthermsite)
 
diff --git a/codes/analysis_sst_species.R b/codes/analysis_sst_species.R
index 354805c..1872273 100644
--- a/codes/analysis_sst_species.R
+++ b/codes/analysis_sst_species.R
@@ -20,7 +20,7 @@ temp_sp <- open_dataset("results/species_tsummaries.parquet")
 
 temp_sp_filt <- temp_sp %>%
   filter(depth == "depthsurf") %>%
-  filter(variant == "max") %>%
+  filter(variant == "mean") %>%
   collect()
 
 # Temperature on sites