GetClosestForecasts: Get the forecast time closest to a given date for a given...

In rNOMADS: An R Interface to the NOAA Operational Model Archive and Distribution System

Get the forecast time closest to a given date for a given model

Description

This function returns which forecast precedes the date and which forecast follows the date for a given model product. Thus a user can average the two forecasts together to provide a precise forecast for a given date. It is optimized for grib file retrieval.

Usage

GetClosestForecasts(abbrev, forecast.date, model.date = "latest",
   depth = NULL, verbose = TRUE)

Arguments

abbrev	The requested model product
forecast.date	What date you want a forecast for, as a date/time object. It must be in the UTC time zone.
model.date	Which model run to use, in YYYYMMDDHH, where HH is 00, 06, 12, 18. Defaults to "latest", which gets the most recent model uploaded to the server.
depth	How many model instances to return. This avoids having to download the entire model list (sometimes several hundred) if only the first few instances are required. Defaults to NULL, which returns everything. This input only makes sense when model.date != "latest".
verbose	Gives a detailed account of progress. Defaults to TRUE.

Value

forecasts$model.url	URL to send to GribGrab for downloading data.
forecasts$model.run.date	When the model was run.
forecasts$back.forecast	Nearest forecast behind requested date.
forecasts$fore.forecast	Nearest forecast after requested date.
forecasts$back.hr	How many hours the back forecast is behind the requested date.
forecasts$fore.hr	How many hours the fore forecast is in front of the requested date.

Author(s)

Daniel C. Bowman [e-mail]

See Also

BuildProfile, GribGrab

Examples

#Get the exact temperature profile of Chapel Hill, NC
#by performing a weighted average of GFS model forecasts.

#Figure out which forecasts to use
forecast.date <- as.POSIXlt(Sys.time(), tz = "UTC")
abbrev <- "gfs_0p50"

## Not run: 
forecasts <- GetClosestForecasts(abbrev = abbrev, forecast.date)

## End(Not run)

#Get levels
pressure <- c(1, 2, 3, 5, 7,
10, 20, 30, 50, 70,
seq(100, 1000, by = 25))
levels <- paste(pressure, " mb", sep = "")

#Variables - temperature and height only
variables <- c("TMP", "HGT")

#Location
lon <- c(-79.052083)
lat <- c(35.907492)
model.domain <- c(lon - 1, lon + 1, lat + 1, lat - 1)

## Not run: 
#Get the data for each
grb.info <- GribGrab(forecasts$model.url, 
   c(forecasts$fore.forecast, forecasts$back.forecast), levels, variables, 
   model.domain = model.domain)

fore.data <- ReadGrib(grb.info[[1]]$file.name, levels, variables) 
back.data <- ReadGrib(grb.info[[2]]$file.name, levels, variables)

back.profile <- BuildProfile(back.data, lon, lat,
   spatial.average = TRUE, points = 8)

fore.profile <- BuildProfile(fore.data, lon, lat,
   spatial.average = TRUE, points = 8)

temps <- cbind(back.profile[[1]]$profile.data[, which(back.profile[[1]]$variables == "TMP"),],
    fore.profile[[1]]$profile.data[, which(fore.profile[[1]]$variables == "TMP"),])
 
heights <-  cbind(back.profile[[1]]$profile.data[, which(back.profile[[1]]$variables == "HGT"),],
    fore.profile[[1]]$profile.data[, which(fore.profile[[1]]$variables == "HGT"),])


time.gap <- forecasts$fore.hr - forecasts$back.hr
exact.temp <- (temps[,1] * abs(forecasts$fore.hr) + temps[,2] * abs(forecasts$back.hr))/time.gap
exact.hgt <- (heights[,1] * abs(forecasts$fore.hr) + heights[,2] * abs(forecasts$back.hr))/time.gap

#Plot results
plot(c(min(temps), max(temps)), c(min(heights), max(heights)), type = "n",
    xlab = "Temperature (C)", ylab = "Height (m)")
points(temps[,1], heights[,1], pch = 1, col = 1)
points(temps[,2], heights[,2], pch = 2, col = 2)
points(exact.temp, exact.hgt, col = 3, lty = 2, pch = 3)
legend("topleft", pch = c(1, 2, 3), col = c(1, 2, 3),
   legend = c(forecasts$back.forecast, forecasts$fore.forecast, as.character(Sys.time())))

## End(Not run)

rNOMADS documentation built on Sept. 6, 2022, 5:06 p.m.