Introducing synopR

Standard workflow

show_synop_data() requires a character vector or a data frame column where each element is a SYNOP string.

library(synopR)
data_input_vector <- c("AAXX 04003 87736 32965 00000 10204 20106 39982 40074 5//// 333 10266 20158 555 64169 65090 =",
                       "AAXX 01094 87736 NIL=",
                       "AAXX 03183 87736 32965 12708 10254 20052 30005 40098 5//// 80005 333 56000 81270 =")

my_data <- show_synop_data(data_input_vector, wmo_identifier = '87736')

print(my_data)
#> # A tibble: 3 × 31
#>   wmo_id   Day  Hour Cloud_base_height Visibility Total_cloud_cover
#>   <chr>  <dbl> <dbl>             <dbl>      <dbl>             <dbl>
#> 1 87736      4     0                 9         65                 0
#> 2 87736      1     9                NA         NA                NA
#> 3 87736      3    18                 9         65                 1
#> # ℹ 25 more variables: Wind_direction <dbl>, Wind_speed <dbl>,
#> #   Wind_speed_unit <chr>, Air_temperature <dbl>, Dew_point <dbl>,
#> #   Relative_humidity <dbl>, Station_pressure <dbl>, MSLP_GH <dbl>,
#> #   Present_weather <dbl>, Past_weather1 <dbl>, Past_weather2 <dbl>,
#> #   Precipitation_S1 <dbl>, Precip_period_S1 <dbl>, Cloud_amount_Nh <dbl>,
#> #   Low_clouds_CL <dbl>, Medium_clouds_CM <dbl>, High_clouds_CH <dbl>,
#> #   Max_temperature <dbl>, Min_temperature <dbl>, Ground_state <dbl>, …

If a meteorological parameter isn’t present in any of the SYNOP messages, you can set remove_empty_cols = TRUE to remove the extra columns.

The optional wmo_identifier argument offers a significant advantage: it allows for automatic filtering in case the data contains messages from other stations.

While the following example uses a vector with only two messages for simplicity, if you are working with thousands of SYNOP strings from multiple stations, this built-in filtering becomes extremely convenient.

library(synopR)
# Messages from 87736 and 87016
mixed_synop <- c("AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 60001 82100 333 56600 82818=",
                 "AAXX 04033 87016 41460 83208 10200 20194 39712 40114 50003 70292 888// 333 56699 82810 88615="
                 )

colorado_data <- show_synop_data(mixed_synop, wmo_identifier = '87736', remove_empty_cols = TRUE)
#> Warning in show_synop_data(mixed_synop, wmo_identifier = "87736",
#> remove_empty_cols = TRUE): 1 message(s) do not contain the identifier '87736'
#> and will be discarded.
knitr::kable(t(colorado_data))

wmo_id	87736
Day	1
Hour	18
Cloud_base_height	4
Visibility	65
Total_cloud_cover	2
Wind_direction	0
Wind_speed	0
Wind_speed_unit	knots
Air_temperature	32.6
Dew_point	21.5
Relative_humidity	52.1
Station_pressure	997.4
MSLP_GH	1006.4
Precipitation_S1	0
Precip_period_S1	6
Cloud_amount_Nh	2
Low_clouds_CL	1
Medium_clouds_CM	0
High_clouds_CH	0

It is good practice to check the SYNOP messages for non-standard structures. The check_synop() function is designed to handle these. It will make sure every message starts with “AAXX” and ends with “=”, does not contain invalid characters (valid characters after removing “AAXX” and “=” are digits 0-9, ‘/’ and ‘NIL’), and verifies that all groups consist of 5 digits (except for the section identifiers ‘333’ and ‘555’).

The check_synop() function accepts either a string vector or a specific data frame column containing SYNOP strings. A data frame with multiple columns —where the SYNOP column is not explicitly specified— will be accepted if and only if that data frame is the direct output of parse_ogimet().

library(synopR)

my_df <- data.frame(syn = c("AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 60001 82100 333 56600 82818=",
                            "AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 60001 82100 333 56600 82818="),
                    second_column = c(5,7))

check_synop(my_df) # Bad
#> # A tibble: 2 × 2
#>   is_valid error_log                                                
#>   <lgl>    <chr>                                                    
#> 1 FALSE    Missing AAXX | Missing '=' terminator | Invalid groups: 5
#> 2 FALSE    Missing AAXX | Missing '=' terminator | Invalid groups: 7

check_synop(my_df$syn) # Good
#> # A tibble: 2 × 2
#>   is_valid error_log
#>   <lgl>    <chr>    
#> 1 TRUE     ""       
#> 2 TRUE     ""

So far, our messages have a correct structure (even the NIL ones). Now, let’s see what happens when they don’t.

library(synopR)

check_synop(c("AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 60001 82100 333 56600 82818=",
              "AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 6000182100 333 56600 82818=",
              "AAXX 01183 87736 12465 20000 10326 2021 39974 40064 5//// 60001 82100 333 56600 82818=",
              "AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 60001 82100 333 56600 82818",
              "Not a synop message="))
#> # A tibble: 5 × 2
#>   is_valid error_log                                              
#>   <lgl>    <chr>                                                  
#> 1 TRUE     ""                                                     
#> 2 FALSE    "Invalid groups: 6000182100"                           
#> 3 FALSE    "Invalid groups: 2021"                                 
#> 4 FALSE    "Missing '=' terminator"                               
#> 5 FALSE    "Missing AAXX | Invalid groups: Not, a, synop, message"

check_synop() returns a tibble where the first column indicates whether each SYNOP is valid (TRUE) or not (FALSE), and the second column describes the specific error found. In our example:

The first SYNOP is correct.
In the second, there is a missing space between groups 6 and 8 in Section 1.
In the third, group 2 of Section 3 contains only 4 digits.
The fourth message is missing the “=” terminator (remember that SYNOP messages must always start with “AAXX” and end with “=”).
The fifth is simply not a SYNOP string at all.

Workflow with Ogimet

The following SYNOP messages were retrieved from Ogimet for the Rio Colorado station, Argentina (WMO identifier: 87736). We will observe that these are not “pure” SYNOP strings; they include a prefix added by Ogimet that specifies the station ID (87736) along with the date and time of the observation.

However, this is not an issue, as we can use the parse_ogimet() function. This tool is specifically designed to separate these aggregates from the raw SYNOP message for processing.

library(synopR)

data_input <- data.frame(synops = c("87736,2026,02,01,03,00,AAXX 01034 87736 NIL=",
                                    "87736,2026,02,01,06,00,AAXX 01064 87736 NIL=",
                                    "87736,2026,02,01,09,00,AAXX 01094 87736 NIL=",
                                    "87736,2026,02,01,12,00,AAXX 01123 87736 12965 31808 10240 20210 39992 40082 5//// 60104 82075 333 10282 20216 56055 82360=",
                                    "87736,2026,02,01,15,00,AAXX 01154 87736 NIL=",
                                    "87736,2026,02,01,18,00,AAXX 01183 87736 12465 20000 10326 20215 39974 40064 5//// 60001 82100 333 56600 82818=",
                                    "87736,2026,02,01,21,00,AAXX 01214 87736 NIL="))

# Note that `parse_ogimet(data_input)` is incorrect
data_from_ogimet <- parse_ogimet(data_input$synops) 

print(data_from_ogimet)
#> # A tibble: 7 × 5
#>    Year Month Day_Ogimet Hour_Ogimet Raw_synop                                  
#>   <dbl> <dbl>      <dbl>       <dbl> <chr>                                      
#> 1  2026     2          1           3 AAXX 01034 87736 NIL=                      
#> 2  2026     2          1           6 AAXX 01064 87736 NIL=                      
#> 3  2026     2          1           9 AAXX 01094 87736 NIL=                      
#> 4  2026     2          1          12 AAXX 01123 87736 12965 31808 10240 20210 3…
#> 5  2026     2          1          15 AAXX 01154 87736 NIL=                      
#> 6  2026     2          1          18 AAXX 01183 87736 12465 20000 10326 20215 3…
#> 7  2026     2          1          21 AAXX 01214 87736 NIL=

# A 'Year' column is included!
parse_ogimet(data_input$synops) |> show_synop_data(wmo_identifier = 87736, remove_empty_cols = TRUE)
#> # A tibble: 7 × 24
#>   wmo_id  Year Month   Day  Hour Cloud_base_height Visibility Total_cloud_cover
#>   <chr>  <dbl> <dbl> <dbl> <dbl>             <dbl>      <dbl>             <dbl>
#> 1 87736   2026     2     1     3                NA         NA                NA
#> 2 87736   2026     2     1     6                NA         NA                NA
#> 3 87736   2026     2     1     9                NA         NA                NA
#> 4 87736   2026     2     1    12                 9         65                 3
#> 5 87736   2026     2     1    15                NA         NA                NA
#> 6 87736   2026     2     1    18                 4         65                 2
#> 7 87736   2026     2     1    21                NA         NA                NA
#> # ℹ 16 more variables: Wind_direction <dbl>, Wind_speed <dbl>,
#> #   Wind_speed_unit <chr>, Air_temperature <dbl>, Dew_point <dbl>,
#> #   Relative_humidity <dbl>, Station_pressure <dbl>, MSLP_GH <dbl>,
#> #   Precipitation_S1 <dbl>, Precip_period_S1 <dbl>, Cloud_amount_Nh <dbl>,
#> #   Low_clouds_CL <dbl>, Medium_clouds_CM <dbl>, High_clouds_CH <dbl>,
#> #   Max_temperature <dbl>, Min_temperature <dbl>

Limitations

General limitations

The validity of a SYNOP string doesn’t mean its content is correct. A quality-control of the derived data is not included. Data post-processing is on you.
Group 555 (reserved for national distribution) is currently ignored, as its content varies by country. However, future versions of synopR may include functions to extract data from this section based on user requirements.
There is no support for sections 222 y 444. show_synop_data() will incorrectly decode the message.

Specific limitations

The following meteorological parameters are not completely decoded, as they will not produce a strictly numeric vector, or the output would be too long:

Horizontal visibility VV
Lowest cloud base height h
Cloud cover N and Nh, but they can be directly interpreted as in oktas (octaves), except when it’s 9, which means the sky is not visible due to fog or other meteorological phenomenon
Present and past weather ww, W1, W2, cloud-related Cl, Cm and Ch. ground-related E and E'

However, Code tables are available in the section “Code Tables” for direct conversions!

You should also be aware of this:

Wind direction = 99 means “variable wind direction”
Wind speed greater than 99 units (m/s or knots) are not supported (the final result will be 99), but it’s expected it won’t break the function
If group 2 from section 1 informs relative humidity instead of dew point, the final value in the Dew_point column will be NA
For geopotential height, only pressure levels 850, 700 and 500 hPa are supported (others pressure levels will result in NA)
Groups 5 and 9 from section 1 are ignored
Imperceptible precipitation, codified as 990, is considered as 0.01 (mm), so it can be distinguished from a 0 value
A cloud description “/” (clouds not visible) is mapped to 10
Snow depth sss is assumed to be between 1 cm and 996 cm. ‘997’ means ‘less than 0.5 cm’, 998 “Snow cover, not continuous” and 999 “Measurement impossible or inaccurate”
Groups 5 (including 55, 56, 57, etc…), 7, 8 and 9 from section 3 are ignored