Filter TC tracks (lat-lon points) over a specified

Question

I am using R to filter Tropical cyclone tracks passing over a grid box. I have a csv file containing the tracks and convert them to a shapefile.

I wanted to filter only the points with the same identifier (the "SN" column in the sample data below) that passed over a specified grid box (5N to 25N and 115E to 135 E). Below is the code that I am using and the link to the data.

jtwc   <- read.csv("1979-1993_TC.csv",header=T,sep=",")

latmin <-5.00
latmax <- 25.00
lonmin <- 115.00
lonmax <- 135.00

jtwc.unique <- unique(jtwc[jtwc$Lat >= latmin & jtwc$Lat <= latmax & jtwc$Lon >= lonmin & jtwc$Lon <= lonmax,c(1,2)])
jtwc.filter <- merge(jtwc,jtwc.unique,all.x = F,all.y = T, sort = F)
jtwc.filter$Lon <- ifelse(jtwc.filter$Lon < 0, jtwc.filter$Lon + 360, jtwc.filter$Lon)
jtwc.filter <- jtwc.filter[with(jtwc.filter,order(Year,Month,Day,Hour,CY)),]

write.table(jtwc.filter,file = "test2_jul_par_1979-1993.csv", sep = ",", row.names = F)

Problem:

This code does not work properly. When I ran the script, I still see tracks outside the box.

Can anyone suggest any way to improve this?

I'll appreciate any help.

Answer 1

This code corrects the base issue, but I don't know if it fully solves your problem. Your original code seems to assume that the combination of CY and SN are unique in the dataset, which I believe is untrue. There must be combinations with different measurements for the same pair. This version saves the boundedvalues and then merges against this boundedtable

library(assertthat)

jtwc   <- read.csv("~/Downloads/1979-1993_TC.csv", header=T, sep=",")

latmin <-5.00
latmax <- 25.00
lonmin <- 115.00
lonmax <- 135.00

# adjust for negative lat
jtwc$Lon <- ifelse(jtwc$Lon < 0, jtwc$Lon + 360, jtwc$Lon)

# derive the bounded points 
jtwc.bounded <- jtwc[jtwc$Lat >= latmin & jtwc$Lat <= latmax & jtwc$Lon >= lonmin & jtwc$Lon <= lonmax,]

# all these are TRUE
assert_that (all(jtwc.bounded$Lat >= latmin))
assert_that (all(jtwc.bounded$Lat <= latmax))
assert_that (all(jtwc.bounded$Lon >= lonmin))
assert_that (all(jtwc.bounded$Lon <= lonmax))


jtwc.unique <- unique(jtwc.bounded[,c(1,2)])

# merge with bounded (
jtwc.filter <- merge(jtwc.bounded, jtwc.unique, all.x = F, all.y = T, sort = F)

assert_that (all(jtwc.filter$Lat >= latmin))
assert_that (all(jtwc.filter$Lat <= latmax))
assert_that (all(jtwc.filter$Lon >= lonmin))
assert_that (all(jtwc.filter$Lon <= lonmax))


jtwc.filter <- jtwc.filter[with(jtwc.filter,order(Year,Month,Day,Hour,CY)),]

write.table(jtwc.filter,file = "test2_jul_par_1979-1993.csv", sep = ",", row.names = F)

Answer 2

Here's an approach that uses data.table to do the data manipulation, then uses googleway to plot the tracks on google maps

library(googleway)
library(data.table) ## because I like working with data.table to do data manipulation

jtwc <- read.csv("~/Downloads/1979-1993_TC.csv")
setDT(jtwc)  ## set as data.table

latmin <-5.00
latmax <- 25.00
lonmin <- 115.00
lonmax <- 135.00

df_bounds <- data.frame(north = latmax, south = latmin, west = lonmin, east = lonmax)

## apply a logical column whether the point is in the box
jtwc[, inBounds := Lat >= latmin & Lat <= latmax & Lon >= lonmin & Lon <= lonmax]


## create a column that identifies if the SN at some point passes through the box
jtwc[SN %in% jtwc[inBounds == TRUE, unique(SN)], passesThroughBox := T ]
jtwc[is.na(passesThroughBox), passesThroughBox := F]

## adding a colour for plotting
jtwc[, colour := ifelse(passesThroughBox, "#4286F4", "#F44141") ]

## you need a google maps API key to plot on Google Maps
map_key <- 'your_api_key'

google_map(key = map_key) %>%
    add_polylines(data = jtwc, lat = "Lat", lon = "Lon", id = "SN", stroke_colour = 'colour',
                                mouse_over_group = 'passesThroughBox') %>%
    add_rectangles(data = df_bounds, north = 'north', south = 'south', west = 'west', east = 'east',
                                 fill_opacity = 0.1)

Then when hovering over the lines, you can see the ones that pass through

Answer 3

To filter you can use

library(dplyr)

dat %>% 
    filter(Lat>=5 & Lat <=25 & Lon>=115 & Lon<135)

conversely if you want to maintain the original data frame you could use

    dat %>% 
        mutate(boxed = ifelse(Lat>=5 & Lat <=25 & Lon>=115 & Lon<135, 1,0))

If you want to plot the tracks

library(ggplot2)

dat %>% 
    mutate(boxed = ifelse(Lat>=5 & Lat <=25 & Lon>=115 & Lon<135, 1,0)) %>% 
    ggplot(aes(Lon,Lat, color=factor(boxed)))+geom_point()