Fix CRU multi-site downscaling (#67)

CRU downscaling from monthly to daily was done wrongly for multiple sites. This now fixes it.

R/ingest_globalfields.R

@@ -346,8 +346,9 @@ ingest_globalfields <- function(
       # expand monthly to daily data
 
       if (length(cruvars)>0){
-        df_out <- expand_clim_cru_monthly( mdf, cruvars ) %>%
-          right_join( df_out, by = "date" )
+        df_out <- left_join(df_out,
+                            expand_clim_cru_monthly( mdf, cruvars ),
+                            by = c("date", "sitename") )
       }
 
       if ("vpd" %in% getvars){
@@ -924,15 +925,12 @@ ingest_globalfields_cru_byvar <- function( siteinfo, dir, varnam ){
 
 expand_clim_cru_monthly <- function( mdf, cruvars ){
 
-  # ensure this function is always called with a single site only
-  stopifnot(length(unique(mdf$sitename)) == 1)
-  # for multiple sites the code would need to be adapted, e.g.:
-  # ddf2 <- mdf |>
-  #   group_split(sitename, year) |>
-  #   purrr::map(\(df) expand_clim_cru_monthly_byyr(first(df$year), df, cruvars))
-
-  ddf <- purrr::map(as.list(unique(mdf$year)),
-      ~expand_clim_cru_monthly_byyr( ., mdf, cruvars ) ) %>%
+  ddf <- mdf |>
+    # apply it separately for each site and each year
+    group_split(sitename, year) |>
+    purrr::map(\(df) expand_clim_cru_monthly_byyr(first(df$year), df, cruvars) |>
+                 mutate(sitename = first(df$sitename)) #ensure to keep sitename
+                 ) |>
     bind_rows()
 
   return( ddf )

tests/testthat/test_CRU_WFDEI_NDEP.R

@@ -175,8 +175,8 @@ test_that("test CRU data multisite downscaling (monthly and downscaled daily)",
   #       getvars = c("temp", "ccov"), # , "prec" NOTE: we can't test prec, unless we specify a seed somewhere
   #       dir = "/data/archive/cru_harris_2024/data", 
   #       timescale = "m")}) |> bind_rows() |> ungroup() |> unnest(data)
-
-  # Illustration of failing test
+  # 
+  # # Illustration of failing test
   # library(ggplot2)
   # p1 <- ggplot(df_cru_monthly_combined, aes(y=temp, x=date, linetype=sitename, color=sitename)) + ggtitle("df_cru_monthly_combined") + geom_point() # CORRECT
   # p2 <- ggplot(df_cru_daily_separate, aes(y=temp, x=date, linetype=sitename, color=sitename))   + ggtitle("df_cru_daily_separate")   + geom_line()   # CORRECT DOWNSCALING