[R-sig-Geo] Timezone boundaries (for adding to world map)?

[Martin Maechler]

>>>>> "RB" == Roger Bivand <Roger.Bivand at nhh.no>
>>>>>     on Fri, 20 Feb 2009 09:55:52 +0100 (CET) writes:

    RB> On Wed, 18 Feb 2009, Greg Snow wrote:
    >> There is a shapefile (works well with maptools and sp packages) with a 
    >> world map and time zone information at: 
    >> http://openmap.bbn.com/data/shape/timezone/
    >> 
    >> It has the info needed to color the countries or parts of countries 
    >> based on time zone, but does not have polygons for time zone stripes 
    >> over the oceans.

    RB> In fact, this shapefile provoked an update in sp, because it is really 
    RB> buggy. 

I'm happy to have provoked progress ;-) :-)

    RB> There are straight-line polygons and many others have multiple 
    RB> repeated coordinates (the sp update fixes these to retain finite label 
    RB> points), but it doesn't fix the absence of most of the data (several DBF 
    RB> columns are empty - checked in oocalc):

    >> library(rgdal)
    >> tz <- readOGR(".", "WrldTZA")
    RB> OGR data source with driver: ESRI Shapefile
    RB> Source: ".", layer: "WrldTZA"
    RB> with  1890  rows and  11  columns
    RB> Feature type: wkbPolygon with 2 dimensions
    RB> Warning messages:
    RB> 1: In Polygon(cbind(jG[[1]], jG[[2]])) :
    RB> Non-finite label point detected and replaced
    RB> 2: In Polygon(cbind(jG[[1]], jG[[2]])) :
    RB> Non-finite label point detected and replaced
    RB> 3: In Polygon(cbind(jG[[1]], jG[[2]])) :
    RB> Non-finite label point detected and replaced
    >> names(tz)
    RB> [1] "FIPS"     "NAME"     "COMMENTS" "REGION"   "LAT"      "LON"
    RB> [7] "TZ"       "GMTOFF"   "LOCALSUM" "OFFSET"   "CLASSES"
    >> summary(tz$GMTOFF)
    RB> NA's
    RB> 1890
    >> summary(tz$TZ)
    RB> NA's
    RB> 1890
    >> summary(tz$OFFSET)
    RB> 0     \xff0        -1    \xff-1    \xff+1       -10       +10
    RB> 25        57        13         7       181        24        38
    RB> \xff+10     +10.5       +11     +11.5       +12   \xff+12        +2
    RB> 24         1        19         3        12        13         1
    RB> \xff-2    \xff+2        +3    \xff-3    \xff+3      -3.5  \xff+3.5
    RB> 4        90        59       146        25         7         3
    RB> -4        +4    \xff-4    \xff+4  \xff+4.5        -5        +5
    RB> 56         4       165         8         1       130        13
    RB> \xff-5    \xff+5  \xff+5.5 \xff+5.75        -6        +6    \xff-6
    RB> 12         3        16         1        77        10        16
    RB> \xff+6  \xff+6.5        -7        +7    \xff+7        -8        +8
    RB> 12        15        63        98        14        61        68
    RB> \xff-8    \xff+8        -9        +9    \xff+9      +9.5      NA's
    RB> 3        56        82        97        13        11        33

    RB> where my locale is UTF-8.

    RB> I wrote to Martin offline that a more recent link is:

    RB> http://efele.net/maps/tz/world/

    RB> and I've put tz_world.rda on http://spatial.nhh.no/R/etc, but it is not 
    RB> line generalised, so it is very large and takes a long time to display - 
    RB> they've kept a lot of boundary detail.

indeed.  But thanks a lot to make it available!!

    RB> Because TZ follow administrative boundaries, they are actually hard to do 
    RB> right. Maybe a raster version might help more if vector precision isn't 
    RB> needed?

as long as we can add a couple of cities (using geographical
coordinates) to it, afterwards, that's fine.

When I load your (above URL) tz file and plot it,
I get the impression that much time is spent on thousands of
very small polygons;
consequently, if I'd want a  ``subset'' of tz that only contains
the polygons that correspond to areas larger than, say, 100 km^2,
how could I do that ?

Looking at the sp-object:

------------------------------------------------------------------------

load("tz_world.rda")

## Look a bit at the 'tz' object:
str(tz, max=2)
## Formal class 'SpatialPolygonsDataFrame' [package "sp"] with 5 slots
##   ..@ data       :'data.frame':	27704 obs. of  1 variable:
##   .. ..- attr(*, "data_types")= chr "C"
##   ..@ polygons   :List of 27704
##   ..@ plotOrder  : int [1:27704] 24471 111 14349 24641 1598 10291 22351 ...
##   ..@ bbox       : num [1:2, 1:2] -180 -90 180 83.6
##   .. ..- attr(*, "dimnames")=List of 2
##   ..@ proj4string:Formal class 'CRS' [package "sp"] with 1 slots

ltz <- sapply(slotNames(tz), slot, object=tz)
sapply(ltz, object.size)
##        data    polygons   plotOrder        bbox proj4string 
##     1690424   110706296      110856         704         704 

str(tz at data[,1])
##  Factor w/ 382 levels "Africa/Abidjan",..: 307 307 307 307 307 307 ...

str(head(tz at polygons[1:3]), max=1)
## List of 3
##  $ :Formal class 'Polygons' [package "sp"] with 5 slots
##  $ :Formal class 'Polygons' [package "sp"] with 5 slots
##  $ :Formal class 'Polygons' [package "sp"] with 5 slots

------------------------------------------------------------------------

and then ask for 

    showMethods(class = "Polygons")

the result is not so revealing to me; but of course, I'm really
an sp greenhorn.

How could I compute areas?

Regards,
Martin



    RB> Roger

    >> 
    >> Hope this helps,
    >> 
    >> 

    RB> -- 
    RB> Roger Bivand
    RB> Economic Geography Section, Department of Economics, Norwegian School of
    RB> Economics and Business Administration, Helleveien 30, N-5045 Bergen,
    RB> Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
    RB> e-mail: Roger.Bivand at nhh.no