batch calculate VI

R/calc_VI.R

@@ -0,0 +1,73 @@
+#' Batch calculate VI
+#'
+#' Calculate VI for the machine learning input
+#' data frame based on the spectral indices table
+#' (in csv format).
+#'
+#' @param df the machine learning data frame including the 7 MODIS NBAR bands
+#' @param indices path to the spectral indices file
+#'
+#' @return select spectral indices for all values in the data frame
+#' @export
+
+calc_VI <- function(
+ df,
+ indices = "data/spectral-indices-table.csv"
+ ){
+
+ # read in spectral indices file
+ vi <- readr::read_csv(indices) |>
+ dplyr::filter(
+ application_domain == "vegetation",
+ #grepl("drought|water", tolower(long_name)),
+ short_name != "MNLI",
+ !grepl("nexp|sla|slb|gamma|epsilon|alpha|beta|omega|A|fdelta|cexp|G1|RE1|RE2|PAR|lambda*",formula)
+ )
+
+ # read data and rename columns
+ df <- df |>
+ dplyr::rename(
+ R = Nadir_Reflectance_Band1,
+ N = Nadir_Reflectance_Band2,
+ B = Nadir_Reflectance_Band3,
+ G = Nadir_Reflectance_Band4,
+ S1 = Nadir_Reflectance_Band6,
+ S2 = Nadir_Reflectance_Band7
+ ) |>
+ dplyr::select(
+ site, date, year, doy,
+ flue, is_flue_drought, cluster,
+ R, N, B, G, S1, S2, LST_Day_1km
+ )
+
+ # loop over all indices
+ df_vi <- lapply(1:nrow(vi), function(i){
+
+ # create temporary object
+ tmp <- df
+
+ # set constants dynamically
+ if(grepl("EVI",vi$short_name[i])){
+ tmp$L <- 1
+ tmp$C1 <- 6
+ tmp$C2 <- 7.5
+ tmp$g <- 2.5
+ }
+
+ if(grepl("SAVI|SARVI",vi$short_name[i])){
+ tmp$L <- 0.5
+ }
+
+ # calculate and return index
+ tmp |>
+ mutate(
+ !!vi$short_name[i] := eval(rlang::parse_expr(vi$formula[i]))
+ ) |>
+ select(
+ !!vi$short_name[i]
+ )
+ }) |>
+ bind_cols()
+
+ return(df_vi)
+}