R/get_dayflow.R

In goertler/inundation: Calculate Delta Inundation Metrics

#' Get Yolo bypass dayflow data
#'
#' Downloads all dayflow data and selects Yolo Bypass and Sacramento River.
#' Returns data frame of daily values.
#'
#' @return data.frame of dayflow data
#'
#' @importFrom rlang .data
#'
#' @export
#'
#' @examples
#' \dontrun{
#'
#' dayflow <- get_dayflow()
#'
#' }
get_dayflow <- function(){

    # create cache dir if it doesn't exist
    if (!(dir.exists(rappdirs::user_cache_dir("inundation")))) {
        dir.create(rappdirs::user_cache_dir("inundation"), recursive = TRUE)
    }

    # read in weir data from cache if it exists
    if (file.exists(file.path(rappdirs::user_cache_dir("inundation"), "dayflow.csv"))) {
        message("Reading dayflow data from cache. To clear cache, use clear_cache().")
        return(readr::read_csv(file.path(rappdirs::user_cache_dir("inundation"), "dayflow.csv"),
                        progress = FALSE,
                        show_col_types = FALSE))
    }


    # get metadata
    m <- jsonlite::fromJSON("https://data.cnra.ca.gov/dataset/06ee2016-b138-47d7-9e85-f46fae674536.jsonld")

    file_table <- m$`@graph`

    file_table <- subset(file_table, `dct:format` == "CSV")

    t <- grep("Results", file_table$`dct:title`)
    urls <- file_table$`dcat:accessURL`$`@id`[t]


    # read in the data

    # set the column type to 'character'
    col_types <- readr::cols(.default = readr::col_character())
    # make an empty list
    dat <- list()
    # for every url...
    for (i in 1:length(urls)){
        # if the url corresponds to the dayflowcalculations2019 csv file,
        if (grepl(pattern = "dayflowcalculations2019", x = urls[i])){
            # read in the csv with some special parsing (read in the columns from 'Year' to 'X2' for the first 365 lines)
            suppressWarnings(dat_file <- readr::read_csv(urls[i], n_max = 365, col_select = c(Year:X2),
                                                         col_types=col_types, show_col_types = T, progress = T))
            # add the csv to our list of files
            dat[[i]] <- dat_file
        } else {
            # else for every other url, read in the csv normally
            dat_file <- readr::read_csv(urls[i], col_types=col_types, show_col_types = T, progress = T)
            # add the csv to our list of files
            dat[[i]] <- dat_file
        }
    }
    suppressWarnings(dat <- lapply(dat, function(x){
        if (is.null(x$YOLO)){
            x$YOLO <- NA
        }
        return(x[, c("Date", "SAC", "YOLO")])
    }))



    dayflow <- do.call(rbind, dat)

    dayflow$Date <- lubridate::parse_date_time(dayflow$Date, orders = c("mdy", "ymd", "dmy"))
    dayflow$YOLO <- as.numeric(dayflow$YOLO)
    dayflow$SAC <- as.numeric(dayflow$SAC)

    names(dayflow) <- c("date", "sac", "yolo")

    # duplicates & triplicates beginning in 2018
    # identifies duplicate dates and takes yolo value that is most similar to previous date's value
    dayflow <- as.data.frame(dayflow)

    dayflow <- dayflow[order(dayflow$date), ]

    for(i in 2:length(unique(dayflow$date))){
        if(nrow(dayflow[dayflow$date==dayflow$date[i],])==1)
            next
        if(nrow(dayflow[dayflow$date==dayflow$date[i],]) > 1)
            min.val <- which.min(abs(dayflow[dayflow$date == dayflow$date[i], "yolo"]-dayflow$yolo[i-1]))
        rnames <- row.names(dayflow[dayflow$date == dayflow$date[i],])[-min.val]
        dayflow <- dayflow[!(row.names(dayflow) %in% rnames),]

    }

    # write out
    utils::write.csv(dayflow, file.path(rappdirs::user_cache_dir("inundation"), "dayflow.csv"), row.names = FALSE)

    return(dayflow)

}