andrew-MET
commented
3 years ago There are a couple of issues here.
The meteogrid functions are to be applied on gridded data as they require projection information for wind direction, so need to be applied before interpolation in the read_forecast() step. Unfortunately this isn't built into the grib reading yet - hopefully @adeckmyn can address that soon.
harp actually doesn't know what do with wind direction when verifying at the moment as it won't take account of the cyclic nature of wind direction. I have opened an issue (https://github.com/harphub/harpPoint/issues/22) for that, but it might not be addressed until next year due to lack of harp budget for this year.
However, you can get wind speed from the data you have - it will require an update of harpPoint to be able to do this.
library(harp)
library(dplyr) # for rename, mutate, select & inner_join
library(purrr) # for map2
u10 <- read_point_forecast (
start_date = 2020082000,
end_date = 2020082100,
fcst_model = "ALAEF",
fcst_type = "eps",
parameter = "u10m",
file_path = "/work/users/ext005/sql/forecast",
)
v10 <- read_point_forecast (
start_date = 2020082500,
end_date = 2020082600,
fcst_model = "ALAEF",
fcst_type = "eps",
parameter = "v10m",
file_path = "/work/users/ext005/sql/forecast",
)
# Get all the members in a single column for easy manipulation
# and remove the parameter column so it's not used in the join
u10 <- gather_members(u10) %>%
rename(u = forecast) %>%
select(-parameter)
v10 <- gather_members(v10) %>%
rename(v = forecast) %>%
select(-parameter)
# get wind speed by joining and calculating the square root of the sum of the squares
# making sure that the output is of class "harp_fcst" and return to each member having its own column
# You can optionally add a new parameter column, but it's not actually used for anything.
wind_speed <- map2(u10, v10, inner_join) %>%
structure(class = "harp_fcst") %>%
mutate(forecast = sqrt(u ^ 2 + v ^ 2), parameter = "S10m") %>%
select(-u, -v) %>%
spread_members()
# Maybe get rid of u10 and v10 to save memory
rm(u10, v10)You can then do verification of wind speed.
adeckmyn
meteorolog90
Sorry if I bother you with such questions, I don't have much experience with the R language.
Does harp have a function to calculate the wind speed and direction from the u and v wind components ?I tried to use functions from meteogrid/R/geowind.R, but I didn't know how to work with it.
What i did: -read u,v components from SQlite:
and then I used the functions
I guess the problem is in object u10, v10, but I don't know how to change them.