mygetStationByID.r

mygetStationByID <- function(stationID, station.list, begin, end) {
    st <- station.list[station.list$USAF == stationID, ]
    st$BEGIN_Year <- as.numeric(substr(st$BEGIN, 1, 4))
    st$END_Year <- as.numeric(substr(st$END, 1, 4))
    location <- suppressMessages(revgeocode(as.numeric(geocode(paste(st$NAME,
        st$CTRY))), output = "more"))
    st$city <- paste(location$locality, location$country, sep = ", ")
    st$rank <- 1
    weatherDFs <- mydlStationData(st, begin, end)
    combined.list <- mycombineWeatherDFs(weatherDFs)
    return(combined.list)
}

mydlStationData <- function(kns, beg, end) {
    base.url <- "ftp://ftp.ncdc.noaa.gov/pub/data/noaa/"
    nstations <- nrow(kns)
    yrs <- seq(beg, end, 1)
    nyrs <- end - beg + 1
    usaf <- as.numeric(kns$USAF)
    wban <- as.numeric(kns$WBAN)
    status <- data.frame()
    temp <- as.data.frame(matrix(NA, nstations, 5))
    names(temp) <- c("File", "Status", "City", "rank", "kilo_distance")
    temp$City <- kns$city
    temp$rank <- kns$rank
    temp$kilo_distance <- kns$kilo_distance
    temp.list <- df.list <- list()
    city.names <- unlist(lapply(strsplit(kns$city, ", "), function(x) x[1]))
    df.names <- paste(city.names, kns$USAF, sep = "_")
    for (i in 1:nyrs) {
        for (j in 1:nstations) {
            temp[j, 1] <- paste(usaf[j], "-", wban[j], "-", yrs[i],
                ".gz", sep = "")
            tryCatch({
                gz.url <- paste(base.url, yrs[i], "/", temp[j,
                  1], sep = "")
                con <- gzcon(url(gz.url))
                raw <- textConnection(readLines(con))
                temp.list[[j]] <- read.fwf(raw, col.width)
                close(con)
                names(temp.list)[j] <- df.names[j]
                names(temp.list[[j]]) <- col.names
                temp.list[[j]]$LAT <- temp.list[[j]]$LAT/1000
                temp.list[[j]]$LONG <- temp.list[[j]]$LONG/1000
                #temp.list[[j]]$WIND.SPD <- temp.list[[j]]$WIND.SPD/10
                #temp.list[[j]]$TEMP <- temp.list[[j]]$TEMP/10
                #temp.list[[j]]$DEW.POINT <- temp.list[[j]]$DEW.POINT/10
                #temp.list[[j]]$ATM.PRES <- temp.list[[j]]$ATM.PRES/10
                #temp.list[[j]]$city <- city.names[j]
                #temp.list[[j]]$distance <- kns$kilo_distance[j]
                #temp.list[[j]]$rank <- kns$rank[j]
                temp[j, 2] <- "Success"
            }, error = function(cond) {
                return(NA)
                next
            }, finally = {
                if (is.na(temp[j, 2]) == "TRUE")
                  temp[j, 2] <- "Failed"
            })
        }
        status <- rbind(status, temp)
        status <- status[order(status[, 3], status[, 4], status[,
            1]), ]
        df.list <- append(df.list, temp.list)
    }
    output.list <- list(status, df.list)
    names(output.list) <- c("dl_status", "station_data")
    return(output.list)
}

mycombineWeatherDFs <- function(dfList) {
    combined.list <- list()
    trackNull <- which(names(dfList$station_data) == "" | names(dfList$station_data) %in%
        NA)
    if (length(trackNull)>0) dfList$station_data <- dfList$station_data[-trackNull]
    keys <- unique(names(dfList$station_data))
    nkeys <- length(keys)
    for (i in 1:nkeys) {
        track <- which(names(dfList$station_data) == keys[i])
        combined.list[[i]] <- suppressWarnings(Reduce(function(...) rbind(...),
            dfList$station_data[track]))
        names(combined.list)[i] <- keys[i]
    }
    output.list <- list(dl_status = dfList$dl_status, station_data = combined.list)
    return(output.list)
}