M_Map:

Users Guide v1.4


Table of Contents

(Note - a Chinese translation of an earlier version of this guide can be found here)


1. Getting started

First, get all the files, either as a zip archive or a gzipped tar-file and unpack them. If you are unpacking the zip file MAKE SURE YOU ALSO UNPACK SUBDIRECTORIES! Now, start up Matlab (version 5 or higher). Make sure that the toolbox is in your path. This can be done simply by cd'ing to the correct directory.

Alternatively, if you have unpacked them into directory /users/rich/m_map (and /users/rich/m_map/private), then you can add this to your search path:

path(path,'/users/rich/m_map');
or
addpath /users/rich/m_map

To follow along with this document, you would then use a Web-browser to open file:/users/rich/m_map/map.html, that is, this HTML document.

Note: you may want to install M_Map as a toolbox accessible to all users. To do this, unpack the files into $MATLAB/toolbox/m_map, add that directory to the list defined in $MATLAB/toolbox/local/pathdef.m, and update the cache file using

rehash toolboxcache

Instructions for installing an (optional) high-resolution bathymetry database are given in here, and instructions for installing the (optional) high-resolution GSHHS coastline database is given in here. However, we should first check that the basic setup is OK.

To see an example map, try this:

m_proj('oblique mercator');
m_coast;
m_grid;

This is a line map of the Oregon/British Columbia coast, using an Oblique Mercator projection (A few more complex maps can be generated by running the demo function m_demo).

The first line initializes the projection. Defaults are set for the different projection, so you can easily see what a specific projection looks like, but all projections have a number of optional parameters as well. To get the same map without using the defaults, you would use

m_proj('oblique mercator','longitudes',[-132 -125], ...
           'latitudes',[56 40],'direction','vertical','aspect',.5);
 

The exact meanings of the various options is given in Section 2. However, notice that longitudes are specified using a signed notation - East longitudes are positive, whereas West longitudes are negative (Also note that a decimal degree notation is used, so that a longitude of 120 30'W is specified as -120.5).

The second line draws a coastline, using the 1/4 degree database. Coastlines with greater resolution can be drawn, using your own database (see also Section 8). m_coast can be called with various line parameters. For example,

m_coast('linewidth',2,'color','r');

draws a thicker red coastline. Filled coastlines can also be drawn, using the 'patch' option (followed by any of the usual PATCH property/value pairs:

m_coast('patch',[.7 .7 .7],'edgecolor','none');

draws a coastline with a gray fill and no border.

The third line superimposes a grid. Although there are many possible options that can be used to customize the appearance of the grid, defaults can always be used (as in the example). These options are discussed in Section 4. You can get a list of the options using the GET syntax:

m_grid get

which acts somewhat like the get(gca) syntax for regular plots.

Finally, suppose you want to show and label the location of, say, a mooring at 129W, 48 30'N.

[X,Y]=m_ll2xy(-129,48.5);
line(X,Y,'marker','square','markersize',4,'color','r');
text(X,Y,' M5','vertical','top');

m_ll2xy (and its inverse m_xy2ll) convert from longitude/latitude coordinates to those of the projection. Various clipping options can also be specified in converting to projection coordinates. If you are willing to accept default clipping setting, you can use the built-in functions m_line and m_text :

m_line(-129,48.5,'marker','square','markersize',4,'color','r');
m_text(-129,48.5,' M5','vertical','top');

Finally (!), we may want to alter the grid details slightly. Note that, a given map must only be initialized once.

clf
m_proj('oblique mercator');  % repeated here so cut-n-paste simplified
m_coast('patch',[.7 .7 .7],'edgecolor','none');
m_grid('xlabeldir','end','fontsize',10);
m_line(-129,48.5,'marker','square','markersize',4,'color','r');
m_text(-129,48.5,' M5','vertical','top');

2. Specifying projections

In order to get a list of the current projections,

m_proj get

or

m_proj('set');

Which currently return the following list:

Available projections are:
     Stereographic          
     Orthographic 
     Azimuthal Equal-area    
     Azimuthal Equidistant
     Gnomonic
     Satellite
     Albers Equal-Area Conic
     Lambert Conformal Conic
     Mercator
     Miller Cylindrical
     Equidistant Cylindrical
     Oblique Mercator
     Transverse Mercator
     Sinusoidal
     Gall-Peters
     Hammer-Aitoff
     Mollweide
     Robinson
     UTM
  

If you want details about the possible options for any of these projections, add its name to the above command, e.g.

m_proj('set','stereographic');

which returns

     'Stereographic'                                   
     <,'lon<gitude>',center_long>                      
     <,'lat<itude>', center_lat>                       
     <,'rad<ius>', ( degrees | [longitude latitude] )>
     <,'rec<tbox>', ( 'on' | 'off' )>                 
 

You can also get details about the current projection. For example, in order to see what the default parameters are for the sinusoidal projection, we first initialize it, and then use the 'set' option:

m_proj('sinusoidal');
m_proj get

Current mapping parameters -
  Projection: Sinusoidal  (function: mp_tmerc)
  longitudes: -90  30 (centered at -30)    
  latitudes: -65  65             
  Rectangular border: off               
  

In order to initialize a projection, you usually specify some location parameters that define the geometry of the projection (longitudinal limits, central parallel, etc.), as well as parameters that define the extent of the map (whether it is in a rectangular axis, what the border points are, etc.). These vary slightly from projection to projection.

Two useful properties for projections are (1) the ability the preserve angles for differentially small regions, and (2) the ability to preserve area. Projections satisfying the first condition are called conformal, those satisfying the second are called equal-area. No projection can be both. Many projections (especially global projections) are neither, instead an attempt has been made to aesthetically balance the errors in both conditions.

Note: Most projections are currently spherical rather than ellipsoidal. UTM is an ellipsoidal projection, and both the lambert conformal conic and albers equal-area conic can be specified with ellipses if desired. This is sometime useful when you have data (e.g. from a GIS package) at scales of Canadian provinces or US states, which are often mapped using one of these projections. It is unlikely that using a spherical earth model will be a problem (or an advantage) in normal usage.

Let's go through the list of available projections:

  1. Azimuthal projections
  2. Azimuthal projections are those in which points on the globe are projected onto a flat tangent plane. Maps using these projections have the property that direction or azimuth from the center point to all other points is shown correctly. Great circle routes passing through the central point appear as straight lines (although great circles not passing through the central point may appear curved). These maps are usually drawn with circular boundaries. The following parameters specify the center point of an azimuthal projection map:

     
      <,'lon<gitude>',center_long>
      <,'lat<itude>', center_lat> 
    

    Maps are aligned so that the specified longitude is vertical at the map center, with its northern end at the top (but see option rotangle below in order to rotate this orientation). Then the extent of the map is defined by

       <,'rad<ius>', ( degrees | [longitude latitude] )> 
       

    Either an angular distance in degrees can be given (e.g. 90 for a hemisphere), or the coordinates of a point on the boundary can be specified. Then,

      <,'rec<tbox>', ( 'on' | 'off' | 'circle' )> 
      

    is used to specify the map boundary. The default is to enclose the map in a circular boundary (chosen using either of the latter two options), but a rectangular one can also be specified. However, rectangular maps are usually better drawn using a cylindrical or conic projection of some sort. Finally,

       <,'rot<angle>', degrees CCW> 
       

    rotates the figure so that the central longitude is not vertical.

    THe azimuthal projections include:

    1. Stereographic
    2. The stereographic projection is conformal, but not equal-area. This projection is often used for polar regions.

    3. Orthographic
    4. This projection is neither equal-area nor conformal, but resembles a perspective view of the globe.

    5. Azimuthal Equal-Area
    6. Sometimes called the Lambert azimuthal equal-area projection, this mapping is equal-area but not conformal.

    7. Azimuthal Equidistant
    8. This projection is neither equal-area nor conformal, but all distances and directions from the central point are true.

    9. Gnomonic
    10. This projection is neither equal-area nor conformal, but all straight lines on the map (not just those through the center) are great circle routes. There is, however, a great degree of distortion at the edges of the map, so maximum radii should be kept fairly small - 20 or 30 degrees at most.

    11. Satellite
    12. This is a perspective view of the earth, as seen by a satellite at a specified altitude. Instead of specifying a radius for the map, the viewpoint altitude is specified:

       <,'alt<itude>',altitude_fraction > 
         

      the numerical value assigned to this property represents the height of the viewpoint in units of earth radii. For example, a satellite in an orbit of radius 3 earth radii would have an altitude of 2.

  3. Cylindrical and Pseudo-cylindrical Projections
  4. Cylindrical projections are formed by projecting points onto a plane wrapped around the globe, touching only along some great circle. These are very useful projections for showing regions of great lateral extent, and are also commonly used for global maps of mid-latitude regions only. Also included here are two pseudo-cylindrical projections, the sinusoidal and Gall-Peters, which have some similarities to the cylindrical projections (see below).

    These maps are usually drawn with rectangular boundaries (with the exception of the sinusoidal and sometimes the transverse mercator).

    1. Mercator
    2. This is a conformal map, based on a tangent cylinder wrapped around the equator. Straight lines on this projection are rhumb lines (i.e. the track followed by a course of constant bearing). The following properties affect this projection:

       <,'lon<gitude>',( [min max] | center)> 

      Either longitude limits can be set, or a central longitude defined implying a global map.

       <,'lat<itude>', ( maxlat | [min max])> 

      Latitude limits are most usually the same in both N and S latitude, and can be specified with a single value, but (if desired) unequal limits can also be used. DO NOT USE MERCATOR FOR A MAP THAT DOES NOT CONTAIN THE EQUATOR!!!

    3. Miller Cylindrical
    4. This projection is neither equal-area nor conformal, but "looks nice" for world maps. Properties are the same as for the Mercator, above.

    5. Equidistant cylindrical
    6. This projection is neither equal-area nor conformal. It consists of equally-spaced latitude and longitude lines, and is very often used for quick plotting of data. It is included here simply so that such maps can take advantage of the grid generation routines. Also known as the Plate Carree. Properties are the same as for the Mercator, above.

    7. Oblique Mercator
    8. The oblique mercator arises when the great circle of tangency is arbitrary. This is a useful projection for, e.g., long coastlines or other awkwardly shaped or aligned regions. It is conformal but not equal area. The following properties govern this projection:

       
          <,'lon<gitude>',[ G1 G2 ]>  
          <,'lat<itude>', [ L1 L2 ]> 
          

      Two points specify a great circle, and thus the limits of this map (it is assumed that the region near the shortest of the two arcs is desired). The 2 points (G1,L1) and (G2,L2) are thus at the center of either the top/bottom or left/right sides of the map (depending on the 'direction' property).

       
         <,'asp<ect>',value> 
         

      This specifies the size of the map in the direction perpendicular to the great circle of tangency, as a proportion of the length shown. An aspect ratio of 1 results in a square map, smaller numbers result in skinnier maps. Aspect ratios >1 are possible, but not recommended.

       
          <,'dir<ection>',( 'horizontal' | 'vertical' ) 

      This specifies whether the great circle of tangency will be horizontal on the page (for making short wide maps), or vertical (for tall thin maps).

      WARNING - at SOME times, for certain combinations of G1/G2 L1/L2 endpoints, the coastline mapping algorithms fail because of a numerical instability affecting the mapping of points on the opposite side of the world - you can see weird lines going across your map when you try to plot a coastline. The work-around for this is to alter the G1/G2 or L1/L2 slightly.

    9. Transverse Mercator
    10. The Transverse Mercator is a special case of the oblique mercator when the great circle of tangency lies along a meridian of longitude, and is therefore conformal. It is often used for large-scale maps and charts. The following properties govern this projection:

          <,'lon<gitude>',[min max]> 
          <,'lat<itude>',[min max]> 
          

      These specify the limits of the map.

       
          <,'clo<ngitude>',value> 
          

      Although it makes most sense in general to specify the central meridian as the meridian of tangency (this is the default), certain map classification systems (noteably UTM) use only a fixed set of central longitudes, which may not be in the map center.

       
          <,'rec<tbox>', ( 'on' | 'off' )> 
          

      The map limits can either be based on latitude/longitude (the default), or the map boundaries can form an exact rectangle. The difference is small for large-scale maps. Note: Although this projection is similar to the Universal Transverse Mercator (UTM) projection, the latter is actually ellipsoidal in nature.

    11. Universal Transverse Mercator (UTM)
    12. UTM maps are needed only for high-quality maps of small regions of the globe (less than a few degrees in longitude). This is an ellipsoidal projection. Options are similar to those of the Transverse Mercator, with the addition of

           <,'zon<e>', value 1-60> 
          
           <,'hem<isphere>',value 0=N,1=S> 
           

      These are computed automatically if not specified. The ellipsoid defaults to 'normal', a spherical earth of radius 1 unit, but other options can also be chosen using the following property:

       
          <,'ell<ipsoid>', ellipsoid>
          

      For a list of available ellipsoids try m_proj('set','utm').

      The big difference between UTM and all the other projections is that for ellipsoids other than 'normal' the projection coordinates are in meters, so-called "easting" and "northing". To take full advantage of this it is often useful to call m_proj with 'rectbox' set to 'on' and not to use the long/lat grid generated by m_grid (since the regular matlab grid will be in units of meters).

    13. Sinusoidal
    14. This projection is usually called "pseudo-cylindrical" since parallels of latitude appear as straight lines, similar to their appearance in cylindrical projections tangent to the equator. However, meridians curve together in this projection in a sinusoidal way (hence the name), making this map equal-area.

    15. Gall-Peters
    16. Parallels of latitude and meridians both appear as straight lines, but the vertical scale is distorted so that area is preserved. This is useful for tropical areas, but the distortion in polar areas is extreme.

  5. Conic Projections
  6. Conic projections result from projecting onto a cone wrapped around the sphere. The vertex of the cone lies on the rotational axis of the sphere. The cone is either tangent at a single latitude, or can intersect the sphere at two separated latitudes. It is a useful projection for mid-latitude areas of large east-west extent. The following properties affect these projections:

     
      <,'lon<gitude>',[min max]> 
      <,'lat<itude>',[min max]> 
    

    These specify the limits of the map.

     
      <,'clo<ngitude>',value> 
      

    The central longitude appears as a vertical on the page. The default value is the mean longitude, although it may be set to any value (even one outside the limits).

     
      <,'par<allels>',[lat1 lat2]> 
      

    The standard parallels can be specified. Either one or two parallels can be given:

     
      <,'rec<tbox>', ( 'on' | 'off' )> 
    

    The default is a parallels at 1/3 and 2/3 of the latitudinal range.

    The map limits can either be based on latitude/longitude (the default), or the map boundaries can form an exact rectangle which contain the given limits. Unless the region being mapped is small, it is best to leave this 'off' .

    The default is to use a spherical earth model for the mapping transformations. However, ellipsoidal coordinates can also be specified. This tends to be useful only for doing coordinate transformations (e.g., if a particular gridded database in in this kind of projection, and you want to find lat/long data), since the difference would be impossible to see by eye on a plot. The particular ellipsoid used can be chosen using the following property:

     
        <,'ell<ipsoid>', ellipsoid>
        

    For a list of available ellipsoids try m_proj('set','albers').

    Finally, if you just want to use M_Map as an engine to transform between projected coordinates in some database and lat/long, it is useful to be able to explicitly specify the origin of the coordinate system that was originally used (origins are sometimes set at the lower boundary of a projection so that all Y values are positive). This can be done be setting:

     
        <,'ori<gin>', [long lat]>
        

    But note that "false eastings" and "false northings" are not handled by m_proj, instead you must correct for them explicitly using

         [long,lat]=m_xy2ll(x-false_easting,y-false_northing);
        

    if you are trying to do this.

    1. Albers Equal-Area Conic
    2. This projection is equal-area, but not conformal

    3. Lambert Conformal Conic
    4. This projection is conformal, but not equal-area.

  7. Miscellaneous global projections
  8. There are a number of projections which don't really fit into any of the above categories. Mostly these are global projections (i.e. they show the whole world), and they have been designed to be "pleasing to the eye". I'm not sure what use they are in general, but they make nice logos!

    1. Hammer-Aitoff
    2. An equal-area projection with curved meridians and parallels.

    3. Mollweide
    4. Also called the Elliptical or Homolographic Equal-Area Projection. Parallels are straight (and parallel) in this projection.

      Note that example 4 shows a rather sophisticated use for viewing the oceans, designed to reduce distortion.

      A more standard map can be made using

          m_proj('mollweide');
          m_coast('patch','r');
          m_grid('xaxislocation','middle');
          

    5. Robinson
    6. Not equal-area OR conformal, but supposedly "pleasing to the eye".

  9. Yeah, but which projection should I use?
  10. Well, it depends really on how large an area you are mapping. Usually, maps of the whole world are Mercator, although often the Miller Cylindrical projection looks better because it doesn't emphasize the polar areas as much. Another choice is the Hammer-Aitoff or Mollweide (which has meridians curving together near the poles). Both are equal-area. It's probably not a good idea to use these projections for maps that don't have the equator somewhere near the middle. The Robinson projection is not equal-area or conformal, but was the choice of National Geographic (for a while, anyway), and also appears in the IPCC reports.

    If you are plotting something with a large north/south extent, but not very wide (say, North and South America, or the North and South Atlantic), then the Sinusoidal or Mollweide projections will look pretty good. Another choice is the Transverse Mercator, although that is usually used only for very large-scale maps (i.e., ones covering a very small area).

    For smaller areas within one hemisphere or other (say, Australia, the United States, the Mediterranean, the North Atlantic) you might pick a conic projection. The differences between the two available conic projections are subtle, and if you don't know much about projections it probably won't make much difference which one you use.

    If you get smaller than that, it doesn't matter a whole lot which projection you use. One projection I find useful in many cases is the Oblique Mercator, since you can align it along a long (but narrow) coastal area. If map limits along lines of longitude/latitude are OK, use a Transverse Mercator or Conic Projection. The UTM projection is also useful if you want to make a grid in meters, since the projection coordinates (i.e., "eastings and northings") are in meters.

    Polar areas are traditionally mapped using a Stereographic projection, since everyone seems to think it looks nice to have a "bullseye" pattern of latitude lines (the fact that lines of longitude come together at the poles also makes them an almost irresistable destination for polar scientists and explorers).

    If you want to get a quick idea of what any projection looks like, it is simple to do so - default parameters for all functions are set for a "typical" usage. For example, to see a stereographic map, try:

      m_proj('stereographic');  % Example for stereographic projection
      m_coast;
      m_grid;
      

  11. Map scales
  12. M_Map usually scales the map so that it fits exactly within the current axes. If you just want a nice picture (which is mostly the case) then this is exactly what you need. On the other hand, sometimes you want to print things out at some exact scale (i.e. if you really much prefer sitting at your desk with a ruler and a piece of paper trying to figure out how far apart Bangkok and Tokyo are). Use the m_scale primitive for this - for a 1:250000 map, call

      m_scale(250000);
      

    after you have drawn everything (Be careful - a 1:250000 map of the world is a lot bigger than 8.5"x11" sheet of paper).

    This option is usually only useful for large-scale maps, i.e. maps of very small areas).

    If you wish to know the current scale, calling m_scale without any parameters will calculate and return that value.

    To return to the default scaling call m_scale('auto').

    (PS - If you do want to find distances from Bangkok to anywhere, plot an azimuthal equidistant projection of the world centered on Bangkok (13 44'N, 100 30'E), and choose a fairly small scale, like 1:200,000,000). Another option would be to use range rings, see example 11.

  13. Map coordinate systems - geographic and geomagnetic.
  14. Latitude/Longitude is the usual coordinate system for maps. In some cases UTM coords are also used, but these are really just a simple transformation based on the location of the equator and certain lines of longitude. On the other hand, there are occasions when a coordinate system based on some other set of axes is useful. For example, in space physics data is often projected in coordinates based on the magnetic poles. M_Map has a limited capability to deal with data in these other coordinate systems. m_coord allows you to change the coordinate system from geographic to geomagnetic. The following code gives you the idea - grids and boxes are shown in the two different coordinate systems:

    lat=[25*ones(1,100) 50*ones(1,100) 25]; % Outline of a box
    lon=[-99:0 0:-1:-99 -99];
    
    subplot(121);
    m_coord('IGRF2000-geomagnetic'); % Treat all lat/longs as geomagnetic
    m_proj('stereographic');
    m_coast('patch',[.5 1 .5],'edgecolor','none');
    m_grid;
    m_line(lon,lat,'color','r','linewi',3);   % "lat/ln" assumed geomagnetic on the geomagnetic map
    m_coord('geographic');                    % Switch to assuming geographic
    m_line(lon,lat,'color','b','linewi',3');  % Now they are treated as geographic
      
    subplot(122); 
    m_coord('geographic');                    % Define all in geographic 
    m_proj('stereographic'); 
    m_coast('patch',[.5 1 .5],'edgecolor','none'); 
    m_grid; 
    m_line(lon,lat,'color','b','linewi',3); 
    m_coord('IGRF2000-geomagnetic'); % Now assume that values are in geomagnetic 
    m_line(lon,lat,'color','r','linewi',3);
    

    Note that this option is not used very much, hence is not fully supported. In particular, filled coastlines may not work properly.


3. Coastlines and Bathymetry

M_Map includes two fairly simple databases for coastlines and global elevation data. Highly-detailed databases are not included in this release because they are a) extremely large and b) extremely time-consuming to process (loops are inherently involved). If more detailed maps are required, section 8 and section 9 give instructions on how to add some freely-available high-resolution datasets.

  1. Coastline options
  2. M_Map includes a 1/4 degree resolution coastline database. This is suitable for maps covering large portions of the globe, but is noticeably coarse for many large-scale applications. The M_Map database is accessed using the m_coast function. Coastlines can be drawn as simple lines, using

      m_coast('line', ...optional line arguments );
      

    or

      m_coast( optional line arguments );
      

    where the optional arguments are all the standard arguments for specifying line style, width, color, etc. Coastlines can also be drawn as filled patches using

      m_coast('patch', ...optional patch arguments );
      

    where the optional trailing arguments are the standard patch properties. For example,

      m_coast('patch',[.7 .7 .7],'edgecolor','g');
      

    draws gray land, outlined in green. When patches are being drawn, lakes and inland seas are given the axes background colour.

    Many older (ocean) maps are created with speckled land boundaries, which looks very nice in black and white. You can get a speckled boundary with

      m_coast('speckle', ....optional m_hatch arguments);
      

    which calls m_hatch. This only looks nice if there aren't too many very tiny islands and/or lakes in the image (see Example 13).

    Note that line coastlines are usually drawn rather rapidly. Filled coastlines take considerably more time to generate (because map limits are not necessarily rectangular, clipping must be accomplished in m-files).

  3. Topography/Bathymetry options
  4. M_Map can access a 1-degree resolution global elevation database (actually, this database is is included in the Matlab distribution, used by $MATLAB/toolbox/matlab/demos/earthmap.m). A contour map of elevations at default levels can be drawn using

    m_elev;
    Different levels can also be specified:
    m_elev('contour',LEVELS, optional contour arguments);
    For example, if you want all the contours to be dark blue, use:
    m_elev('contour',LEVELS,'edgecolor','b');
    Filled contours are also possible:
    m_elev('contourf',LEVELS, optional contourf arguments);
    Finally, if you want to simply extract the elevation data for your own purposes,
    [Z,LONG,LAT]=m_elev([LONG_MIN LONG_MAX LAT_MIN LAT_MAX]);
    returns rectangular matrices for depths Z at locations LONG,LAT.

4. Customizing axes

  1. Grid lines and labels
  2. In order to get the perfect grid, you may want to experiment with different grid options. Two functions are useful here, m_grid which draws a grid, and m_ungrid which erases the current grid (but leaves all coastlines and user-specified data alone). Try

      m_proj('Lambert');
      m_coast;
      m_grid;
      

    to get a Lambert conic projection of North America. Now try

      m_ungrid
      

    The coastline is still there, but the grid has disappeared and the axes shows raw X/Y projection coordinates. Now, try this:

      m_grid('xtick',10,'tickdir','out','yaxislocation','right','fontsize',7);
      

    The various options that can be changed include:

        'box',( 'on' | 'off' | 'fancy' ) 
        

    which specifies whether or not an outline box is drawn. Three types of outline boxes are available: 'on', the default, is a a simple line. Two types of fancy outline boxes are available. If 'tickdir' is 'in', then alternating black and white patches are made (see example 2). If 'tickdir' is set to 'out', then a more complex line pattern is drawn (see example 6). Fancy boxes are in general only available for maps bounded by lat/long limits (i.e. not for azimuthal projections), but if this option is chosen inappropriately a warning message is issued.

    The number/location of ticks on the longitude grid is given by

       'xtick',( num | [value1 value2 ...])
       

    If a single number is specified, grid lines/values are drawn for approximately that number of equally-spaced locations (the number is only approximate because the M_GRID attempts to find "nice" intervals, i.e. it rounds to even increments). Exact locations can be specified by using a vector of location values. There is an analogous 'ytick' property.

    Special labels can be specified using

       'xticklabels',[label1;label2 ...]
       

    Labels can either be numerical values (which are then formatted by M_GRID), or string values which are used without change. There is an analogous 'yticklabels' property.

    Longitude labels are either middled onto the ends of their respective grid lines (and drawn perpendicular to those lines), or are drawn extending outwards from the ends of those lines.

       'xlabeldir', ( 'middle' | 'end' )
       

    There is an analogous 'ylabeldir' property.

    The lengths of tickmarks (as a fraction of plot width) is specified using

       'ticklen',value  
      

    and their direction is set inwards or outwards with

       'tickdir',( 'in' | 'out' )
       

    If 'box' is set to 'fancy', the tickdir specifies the form of the fancy outline box (their is the 'in' version and the 'out' version).

    Axis labels can be produced either in decimal degrees ('dd') or degrees-minutes ('dm', default). The 'da' option is an abbreviated degrees-minutes format without degree marks or the N/S/E/W letters appended:

      'tickstyle',( 'dd' | 'da' |  'dm' )
      

    A number of other (obvious) properties can also be set:

    'color',colorspec  
    'gridcolor',colorspec
    'backgroundcolor',colorspec  
    'linewidth', value  
    'linestyle', ( linespec | 'none' )  
    'fontsize',value  
    'fontname',name
    

    Finally,

     
      'xaxisLocation',( 'bottom' | 'middle' | 'top' ) 
      

    specifies where the X-axis will be drawn, either at the bottom (southermost) end, at the top (northernmost) end, or in the middle (note - for southern hemisphere maps, especially those containing the south pole, setting this to 'top' is recommended), and

       'yaxisLocation',( 'left' | 'middle' | 'right' ) 
      

    specifies where the Y-axis will be drawn, either at the left (westernmostmost) end, at the right (easternmost) end, or in the middle.

    Note: if you are using the UTM projection, and you want to have a grid with eastings and northings, call m_utmgrid AFTER calling m_grid. This will draw a UTM grid, independent of the lat/long grid provided by a call to m_grid. Again, you can customize many aspects of the grid by setting appropriate properties.

  3. Titles and x/ylabels
  4. Titles and x/ylabels can be added to maps using the title and x/ylabel functions in the usual way (this is a change from v1.0 in which the 'visible' property had to be explicitly set to 'on'; this is now done within m_grid).

  5. Legend Boxes
  6. A legend box can be added to the map using m_legend. Only some of the functionality of legend is currently included. The legend box can be dragged and dropped using the mouse button.

  7. Scale Bars
  8. A scale bar can be added to the map using m_ruler. The bar is drawn horizontally or vertically, and will create a 'nice' number of ticks (although this can be changed with another calling parameter). The location is specified in normalized coordinates (i.e. between 0 and 1) so you can adjust placement on the map, see Examples 9 and 10. It is probably best to call m_ruler AFTER calling m_grid since m_grid resets the normalization.

    WARNING - the scalebar is probably not useful for any global (i.e. whole-world) or even a significant-part-of-the-globe map, because the actual ground scale is often quite distorted in some parts of the map, but I won't stop you using it. Caveat user!


5. Adding your own data

The purpose of M_Map is to allow you to map your own data! Once a suitable grid and (possibly) a coastline have been chosen, you can add your own lines, text, or contour plots using built-in M_Map drawing functions which handle the conversion from longitude/latitude coordinates to projection coordinates. These drawing functions are very similar to the standard Matlab plotting functions, and are described in the next section.

Sometimes you may want to convert between longitude/latitude and projection coordinates without immediately plotting the data. This might happen if you want to interactively select points using ginput, or if you want to draw labels tied to a specific point on the screen rather than a particular longitude/latitude. Projection conversion routines are described in sections 5.5 and 5.6. Once raw longitude/latitude coordinates are converted into projection coordinates, standard Matlab plotting functions can be used.

Maps are drawn to fit within the boundaries of the plot axes. Thus their scale is somewhat arbitrary. If you are interested in making a map to a given scale, e.g. 1:200000 or something like that, you can do so by using the m_scale primitive, see section 2.6 . The data units are the projection coordinates, which are distances expressed as a fraction of earth radii. To get a map "distance" between two points, use the Cartesian distance between the points in the projection coordinate system and multiply by your favourite value for the earth's radius, usually around 6370 km (exception - the UTM projection uses coordinates of northing and easting in meters, so no conversion is necessary).

CAUTION: One problem that sometimes occurs is that data does not appear on the plot due to ambiguities in longitude values. For example, if plot longitude limits are [-180 180], a point with a longitude of, say, 200, may not appear in cylindrical and conic projections. THIS IS NOT A BUG. Handling the clipping in "wrapped around" curves requires adding points (rather than just moving them) and is therefore incompatible with various other requirements (such as keeping input and output matrices the same size in the conversion routines described below).

  1. Drawing lines, text, arrows, patches, hatches, speckles and contours
  2. For most purposes you do not need to know what the projection coordinates actually are - you just want to plot something at a specified longitude/latitude. Most of the time you when you want to plot something on a map you want to do so by specifying longitude/latitude coordinates, instead of the usual X/Y locations. To do so in M_Map, replace calls to plot, line, text, quiver, patch, annotationcontour, and contourf with M_Map equivalents that recognize longitude/latitude coordinates by prepending "m_" to the function name. For example,

      m_plot(LONG,LAT,...line properties)      % draw a line on a map (erase current plot) 
      m_line(LONG,LAT,...line properties)      % draw a line on a map 
      m_text(LONG,LAT,'string')                % Text   
      m_quiver(LONG,LAT,U,V,S)                 % A quiver plot 
      m_patch(LONG,LAT,..patch properties)     % Patches.  
      m_annotation('line',LON,LAT)             % Annotation
      

    Each of these functions will handle the coordinate conversion internally, and will return a vector of handles to the graphic objects if desired. The only difference between these functions and the standard Matlab functions is that the first two arguments MUST be longitude and latitude.

    One caveat applies to m_patch. For compatibility reasons this uses the same code that applies to coastline filling. Coastlines come either as either "islands" or "lakes", and M_Map keeps track of the difference by assuming curves are oriented so that the filled area ("land") is always on the right as we go around the curve. This is slightly different than the convention used in patch which always fills the inside. Keeping track of this difference is relatively straightforward in a Cartesian system, but not so easy in spherical coordinates. In the absence of other information m_patch tries to do the right thing, but (especially when the patch intersects a map boundary) it can get confused. If a patch isn't filling correctly, try reversing the order of points using flipud or fliplr.

    Data gridded in longitude and latitude can also be contoured:

      m_contour(LONG,LAT,VALUES)
      m_contourf(LONG,LAT,VALUES)
      

    Again, these functions will return handles to graphics objects, allowing (for example) the drawing of labelled contours:

      [cs,h]=m_contour(LONG,LAT,VALUES)
      clabel(cs,h,'fontsize',6);
      

    Fancy arrows (i.e. with width, head shape, and colour specifications) can be generated using m_vec.m . See the on-line help for more details about the use of m_vec.

    You can also get hatched areas by calling m_hatch:

      
      m_hatch('single',LONG,LAT,...hatch properties)    % Interior Single Hatches. 
      m_hatch('cross',LONG,LAT,...hatch properties)     % Interior Crossed Hatches. 
      

    Note that this call does not generate the edge lines (an additional m_line is required for this. In addition, we can speckle the inside edges of patches using:

     
      m_hatch('speckle',LONG,LAT,...speckle properties)  % Speckled edges.
      

    See the on-line help and/or Example 12 for more details about using m_hatch.

  3. Drawing images and pcolor
  4. Gridded data in matlab can be handled with either a) pcolor, usually for small grids, whose vertices are specified in matrices the same size as that of the data, and where you might want to shade across each grid cell, or b) image, often for much larger pixellated datasets where each value will be mapped to a colour in rectangular cells all of which are the same size. pcolor is often used for datasets that one might also handle with contour or contourf, while image is used for photographs and the like.

    1. In M_Map, if you have a georeferenced image in lat/long coordinates (i.e. each data row is along a line of constant latitude, each column a line of equal longitude), then you can use m_image. That is,

      m_image(LON,LAT,DATA)
      
      will accept data in which LON and LAT are either two-element vectors which give the coordinates of the first and last row or column, or are vectors the same size as the number of rows or columns of DATA (this is a little more flexible than image since it means that rows can be unequally spaced in longitude). The data matrix will be regridded into map coordinates and displayed using image. Generally this will be a good idea only when the data pixels are already too small to see (see this example). Note that DATA can either by an NxM double precision floating point matrix (say, a bathymetry database), or an NxMx3 uint8 "truecolor" image (say, one derived from optical remote sensing).

      It is also sometimes useful to use m_image solely to transform the data in an NxM array, following it with a m_shadedrelief call to actually display the data
      [IM,X,Y]=m_image(LON,LAT,DATA);
      m_shadedrelief(X,Y,IM,'coords','map')
      
      Note that you should set the figure background colour (e.g., set(gcf,'color','w')) before calling m_image because pixels that appear outside the map boundaries are set to that background colour - see This example
    2. If your georeferenced image is in lat/long coordinates, but all points in a row do NOT have the latitude, then use m_pcolor with shading flat. This is reasonably satisfactory (although it can be slow for large images), but you SHOULD offset your coordinates by one-half of the pixel spacing. This is because of the different behaviors of pcolor and image when given the same data. (see this example).

      Note: you must be careful with m_pcolor near map boundaries. Ideally one would want data to extend up to (but not across) a map boundary (i.e. polygons are clipped). However, due to the way in which matlab handles surfaces this is not easily done. Instead - unless you are using a simple cylindrical or conic projection - you will probably get a ragged edge for the coloured surface. 

      Also, be aware that image will center the (i,j) pixel at the location of the (i,j)th entry of the X/Y matrices,as long as these matrices are regular in their spacing. If they are irregular, then pixels are spaced evenly between the 1st and last entries in the matrix. However p_color with shading flat will draw a panel between the (i,j),(i+1,j),(i+1,j+1),(i,j+1) coordinates of the X/Y matrices with a color corresponding to the data value at (i,j). Thus everything will appear shifted by one half a pixel spacing. This can be avoided with shading interp but then the computational time to create an image is greatly increased.

    3. If your figure has already been placed in some projection, and if you know the exact parameters of that projection, you can probably use a straight image call and then overplot an M_Map map. For example, polar satellite images are often given on an equally-spaced grid in a polar stereographic projection. In this case you should use m_ll2xy to get the screen coordinates of the image corners, then use those points in an image() call before overplotting your data. See in particular this example.

      HINT - check to see that coastlines overplot in the right place to make sure this is working correctly.

  5. Drawing shaded relief maps
  6. Shaded relief is a visualization feature used a lot by geologists and geophysicists, but generally not by oceanographers (although perhaps it should be). The general idea is that instead of just contouring a map (with different heights shown as different colours), as in the left example below, you additionally shade the surface as if it was a 3D object, lit from a single direction, say, (as a default) from the upper left corner of the map, as in the right example below. Slopes facing towards the upper left corner are brighter, and those facing away are darker.

    Shaded relief thus lets you visually represent both the low-wavenumber variation (blobs of different colours), as well as the high-wavenumber variation in an image (since slopes are a derivative, they amplify high-wavenumber information).

    in M_MAP, m_shadedrelief is used to construct such a map. However, it is generally only useful for large scale maps covering a small area. This is because it is (essentially) a drop-in replacement for image showing a true-colour image (like a photograph) and hence can only be used when all the pixels are the same size (this means that all points are on a GRID in screen coordinates, and the spacing between points is always EQUAL), and the map axes limits are a RECTANGLE. That is:

    1. You should call m_shadedrelief only AFTER calls to colormap and caxis, because these are needed to determine the shading in the true-colour image.
    2. m_shadedrelief most straightforward use is as a backdrop to maps with a rectangular outline box - either a cylindrical projection, or some other projection with m_proj(...'rectbox','on'). The simplest way of not running into problems is
      1. if your elevation data is in LAT/LON coords, use m_proj('equidistant cylindrical',...)
      2. if your elevation data is in UTM coords (meters E/N), use m_proj('utm',...)
    3. Alternatively, you can use m_shadedrelief AFTER calling m_image (see Section 5.2); this can be done with ANY projection or outline box.

    m_shadedrelief accepts several options. The essence of the mapping is that slopes are calculated relative to a light source, and then a shading function is applied:

           Shading = clipval*tanh(slope/gradient) 
     

    where the shading increases with slope angle, up until a limiting value of about gradient degrees, after which shading saturates. The saturated amount of shading is set by clipval which is between 0 and 1, where 0 means no shading correction and 1 means shading to white or black.

    The direction of the lighting can also be changed, but the default upper left source direction seems to be an accepted standard.

    Finally, the slope calculation only works correctly if we know how to interpret x/y changes with Z changes. For UTM coordinate data, all 3 would be in meters. However, a lot of high-resolution data is provided in lat/long coordinates, and hence the x/y directions would have to be properly scaled. FInally, pixel locations may be in map coordinates. The coords option can let you specify which of the cases is in use.

            'coo<rds>', ( 'g<eographic>'  |  'z'  |  'm<ap>' ) 
    

    For more information, see this example, this one, or this one.

  7. Drawing tracklines
  8. It is sometimes useful to annotate lines representing the time-varying location of a ship, aircraft, or satellite with time and date annotations. This can be done using m_track:

      m_proj('UTM','long',[-72 -68],'lat',[40 44]);
      m_gshhs_i('color','k');
      m_grid('box','fancy','tickdir','out');
      
      % fake up a trackline
      lons=[-71:.1:-67];
      lats=60*cos((lons+115)*pi/180);
      dates=datenum(1997,10,23,15,1:41,zeros(1,41));
      
      m_track(lons,lats,dates,'ticks',0,'times',4,'dates',8,...
                   'clip','off','color','r','orient','upright');  
       

    See the on-line help for more details about the use of m_track, and the different options for setting fontsize, tick spacing, date formats, etc.

    While fiddling with the various parameters, it is often handy to be able to erase the plotted tracks without erasing the coastline and grid. This can be done using

      m_ungrid track
      

    or

      m_ungrid('track')
      

  9. Drawing range rings and geodesics
  10. One nifty thing that is sometimes useful is the ability to draw circles at a given range or ranges from a specific location. This can be done using m_range_ring, which has 3 required calling parameters: LONG, LAT, RANGE, followed by any number of (optional) line specification property/value pairs. Example 11 illustrates how to use m_range_ring.

    If you want to plot circular geodesics (i.e. curves which are perpedicular to the range rings at all ranges), m_lldist can find both distances and points along the geodesics between points. Example 13 illustrates how to use m_lldist.

    If you care about the difference between great circle and ellipsoidal geodesics (a very very small proportion of users I would bet) then m_fdist (which computes the position at a given range/bearing from another), m_idist (distance and bearings between points), and m_geodesic (points along the geodesic) can be used with a variety of (user-specified) ellipses. The calling sequence for these is different than for m_lldist for historical reasons.

  11. Converting longitude/latitude to projection coordinates
  12. If you want to use projection coordinates (perhaps you want to compute map areas, or distances, or you want to make a legend in the upper left corner), the following command converts longitude/latitude coordinates to projection coordinates.

     
      [X,Y]=m_ll2xy(LONG,LAT, ...optional clipping arguments )
      

    where LONG, LAT, X, and Y are matrices of the same size. Projection coordinates are equal to true distances near the center of the map, and are expressed as fractions of an earth radius. To get a distance, multiply by the radius of the earth (about 6370km). The exception is the UTM projection which provides coordinates of northing and easting in meters.

    The possible clipping arguments are

       'clip','on' 
       

    This is the default. Columns of LONG and LAT are assumed to form lines, and these are clipped to the map limits. The first point outside the map is therefore moved to the map edge, and all other points are converted the NaN.

       'clip','off' 
       

    No clipping is performed. This is sometimes useful for debugging purposes.

      'clip','point' 
      

    Points are tested against the map limits. Those outside the limits are converted to NaN, those inside are converted to projection coordinates. No points are moved. This option is useful for point data (such as station locations).

       'clip','patch' 
       

    Points are tested against the map limits. Those outside the limits changed into a point exactly on the limits. Those inside are converted to projection coordinates. This option may be useful when trying to draw patches, however it probably won't work well.

  13. Converting projection to longitude/latitude coordinates
  14. Conversion from projection coordinates to longitude/latitude is straightforward:

      [LONG,LAT]=m_xy2ll(X,Y)
      

    There are no options.

  15. Computing distances between points
  16. Geodesic (great circle) distances on a spherical earth can be computed between pairs of either geographic (long/lat) or map (X/Y) coordinates using the functions m_lldist and m_xydist. For example,

      DIST=m_lldist([20 30],[44 45])
      

    computes the distance from 20E, 44N to 30E, 45N. Alternatively, if you want to compute the distance between two points selected by the mouse:

      [X,Y]=ginput(2);
      DIST=m_xydist(X,Y)
      

    will return that distance. Because of the inaccuracies implicit in a spherical earth approximation the true geodesic distances may differ by 1% or so from the computed distances.

    If you want greater accuracy, then you must calculate geodesics on an ellipsoidal earth. There is a very accurate numerical algorithm for doing so ( Vincenty's algorithm), which is implemented in the functions m_idist, m_fdist, and m_geodesic. For example,

     [distance,a12,a21] = m_idist(lon1,lat1,lon2,lat2,spheroid)
    

    computes the distance in meters between two points (lon1,lat1) and (lon2,lat2) on the specified spheroid ('wgs84' is the default, for other options see the code or use on of the options shown by m_proj('get','utm')). Forward and reverse azimuths (bearings for the rest of us) a12 and a21 in degrees are also computed.

    m_fdist is used to get the location of a point at a given bearing and distance from a specified point. Example 6 shows the use of these functions to find a midpoint of a drifter track.

    Finally, if you want to plot a geodesic on a map, then m_geodesic can be used to generate a vector of points along the elliptical geodesic between two specified points. If you ever find yourself needing this, I'd be interested in knowing about it!

  17. Annotation
  18. Adding labels and boxes to your map may be simpler with m_annotation, which lets you input arrow and text location using latitude/longitude coordinates and otherwise passes arguments through to the built-in MATLAB function annotation.m. Note that this function is fragile with respect to window resizing. Since the annotation coordinates are normalized to the window size, and the map's boundaries may move with respect to the window edges if you change the aspect ratio of the plot window (by, e.g., resizing the window with the mouse), it is best to either

    • refrain from changing the window size between the m_annotation call and a subsequent print call, or
    • carry out operations in the order:
      m_proj('miller');   % Map first
      m_coast;
      m_grid;
      ...
      orient tall;        % Fix the orientation, and...
      wysiwyg;            % make the screen version match the (future) printed version
      m_annotation('textarrow',[-140 -122.6],[20 49.35],'string','Vancouver')
      ...
      print
      

  19. Colour and Colourmaps
  20. The use of colour to show bathymetry/topography on a map, or values of other parameters displayed on a map, is often critical, and m_colmap contains a number of useful colourmaps to do this. These include several maps with good perceptual qualities, as well as a number which `look good' on geographic maps. You can generate the plot above, which gives examples of different calls, using:

    m_colmap demo

    What does it mean to have 'good perceptual qualities'? Nowadays most people know that Matlab's jet colormap has a number of shortcoming. In particular it is NOT PERCEPTUALLY UNIFORM - we can see fine details around the yellow colour band, but the wide range of blues are hard to separate.

    More modern colourmaps are designed to overcome this problem, by designing a gradation in colour-space which looks uniform to the human eye. One school of thought does away with colours altogether, in favour of what is essentially a PERCEPTUALLY UNIFORM single-colour scale that goes from dark to light. Examples b and h above are this kind of colormap (adapted from the ColorBrewer), but so is Matlab's parula (more or less), hot and gray. These kinds of colourmap let us distinguish 2 levels: 'high' (light) and 'low' (dark).

    Such simple colourmaps are not helpful for data that has a +/- sign (e.g., current velocities), and so for this kind of data a DIVERGING colormap (like example d above) is often used. Diverging colourmaps let us distinguish 3 levels: 'negative','zero', and 'positive'. They are also good for 'below-average', 'average', and 'above-average'.

    What if we want more gradations? There is no reason why we can't have the multiple colours of a JET-like colourmap, which makes it much easier to identify several levels within an image, once we design in a perceptually uniform luminance scale. Thus we can create a PERCEPTUALLY UNIFORM JET colourmap. Example a above shows a smooth version (taken from the CET Perceptually Uniform Colour Maps toolbox), which shows about 5 different levels, but examples j and k also show versions with an arbitrary (but small) number of distinct colours which are all, to the eye, about equally different to one another.

    Example j is useful for false-colour satellite imagery, where values are contaminated by noise. The idea of the gradual blurring at boundaries between colours is that it will de-emphasize speckles that would otherwise arise.

    Example k is useful for numerical models, or other data that is already smooth, so quantizing to a small number of levels will not be a problem. The sharp boundaries will then act a little like contours on the final image.

    Finally, m_colmap includes several specialized colormaps (examples c, f, i, and k) designed to show land; these are based on the ETOPO1 land colormap from the Cartographic Palette Archive.

  21. Colourbars with Contourmaps
  22. Matlab has the colorbar command which can be used to add a scale to smoothly-plotted data (see Satellite examples 1, 2 or 4) It can also be useful if you try to simulate contours used a stepped colourmap as in Example 14). But if you are actually showing filled contours, the continuous set of colours in the default colormap does not correspond with the small number of discrete colours shown in the contourmap. This may be important in figures like Example 15, 7 or 8.

    For these latter situations use m_contfbar. Typical usage is

       m_contourf(LON,LAT,DATA,LEVELS);
       m_contfbar(Xloc,Yloc,DATA,LEVELS);
    

    or

       [CS,CH]=m_contourf(LON,LAT,DATA,LEVELS);
       m_contfbar(Xloc,Yloc,CS,CH);
    
    or other calls that involve contourf like
       [CS,CH]=m_etopo2('contourf',LEVELS);
       m_contfbar(Xloc,Yloc,CS,CH);
    

    where Xloc=[X1,X2] and Yloc=Y for a horizontal scale bar at normalized height Y, with sides at normalized x-locations X1 and X2 (normalized means between 0 and 1). For a vertical scale bar, Xloc=X1 and Yloc=[Y1,Y2]. By using the same data and levels information, the scale bar shows the same levels as that of the filled contour image itself.

    m_contfbar can also be used with M_shadedrelief:

       caxis([CMIN CMAX])
       colormap(map)
       m_shadedrelief(LON,LAT,DATA)
       m_contfbar(Xloc,Yloc,DATA,LEVELS);
    

    where you are free to choose the LEVELS as you wish.

    There are 4 additional parameter/value pairs that can be used to modify the appearance of the scale bar. First, the width of the bar can be set to 0.03 of the full axis width using:

       'axfrac',.03
    

    and triangular endpieces (signifying data outside the region of the colour scale) can be added with

       'endpiece',[ 'yes' | 'no']
    

    If you don't like having a black edging line between the colours, choose the 'none' option:

       'edgecolor',[colorspec | 'none']
    

    and finally, it is possible to set the LEVELS vector to have levels that are outside the range of values in DATA. Do you want the scale bar show these levels, or to show only the levels that match those in the range of DATA? Specify with:

       'levels',['set' (default) | 'match']
    

6. More complex maps

For ideas on how to make more complex maps, see the Examples. Some of these maps are also included in the function m_demo.


7. Removing features from a map

Once a given map includes several elements a certain amount of fiddling is usually necessary to satisfy the natural human urge to give the image a certain aesthetic quality. If the image includes complicated coastlines which take a long time to draw (e.g. those discussed below) than clearing the figure and redrawing soon becomes tedious. The m_ungrid command introduced above can be used to selectively remove parts of the figure. For example:

m_proj('lambert','long',[-160 -40],'lat',[30 80]);
m_coast;
m_range_ring(-123,49,[1e3:1e3:10e3],'color','r');

draws range rings at 1000km increments from my office. But I am unsatisfied with this, and want to redraw using only 200km increments. I can remove the effects of m_range_ring and redraw using:

m_ungrid range_ring
m_range_ring(-123,49,[200:200:2000],'color','r');

In general the results of m_ANYTHING can be deleted by calling m_ungrid ANYTHING. m_ungrid can recognize and delete specific elements by searching the 'tag' property of all plot elements, which is set by M_Map routines.

The 'tag' property is also useful if you want to make your own modifications. For example, say you want to REMOVE every 2nd xticklabel (you like the extra lines in a grid, but not that many labels). You can do this with:

handles=findobj('gca,'tag','m_grid_xticklabel');
delete(handles(2:2:end));

8. Adding your own coastlines

If you are interested in a particular area and want a higher-resolution coastline than that used by m_coast, the best procedure is to get one of the high-resolution databases I describe below. If this doesn't work, first I give are some hints on how to deal with your own coastlines.

  1. Reading and Handling coastline data
  2. If you have data is stored in 2 columns (longitudes then latitudes, with line segments separated by a row of NaNs) in a file named "coast.dat", you can plot it (as lines) using the following:

    load coast.dat
    m_line(coast(:,1),coast(:,2));
    

    Filled coastlines will require more work. First, if the coastline is in a number of discrete segments, you have to join them all together to make complete "islands" and "lakes". If you are lucky, (i.e. no lakes or anything else), you may achieve success with

    load coast.dat
    [X,Y]=m_ll2xy(coast(:,1),coast(:,2),'clip','patch');
    k=[find(isnan(X(:,1)))];
    for i=1:length(k)-1,
         x=coast([k(i)+1:(k(i+1)-1) k(i)+1],1);
         y=coast([k(i)+1:(k(i+1)-1) k(i)+1],2);
         patch(x,y,'r');
     end;
     

    and then try replacing patch with m_patch.

    If this does not work (e.g., because your coastline includes "lakes"), read the comments in private/mu_coast, orient the curves in the desired fashion, and use m_usercoast to load your own data.

  3. ESRI Shapefiles
  4. A de facto standard for the interchange of vector data are ESRI shapefiles. A dataset comes in (at minimum) 3 files, each with the same root name but with .dbf, .shp, and .shx extensions. Files can contain point, line or polygon information, as well as other fields in a self-describing way. For more information see this description .

    Many (all?) shapefiles can be read into Matlab using m_shaperead, which returns a data structure containing the information in the files. However, figuring out what to do with the contents requires you to examine the contents of the data structure.

    You can usually at least create a simple plot of the data stored in files datafile.shp, datafile.shx and datafile.dbf using

    M=m_shaperead('datafile'); 
    clf; 
    for k=1:length(M.ncst), 
         line(M.ncst{k}(:,1),M.ncst{k}(:,2)); 
    end; 
    

    If the data is already in lat/long coordinates, change the line to m_line.

  5. Projection Conversions
  6. Sometimes coastline data is already provided in the coordinates of some projection. Usually you will want to convert this data back to lat/long by a) calling m_proj with the specifications of that projection, and b) calling m_xy2ll with the data you read in.

    For data in UTM coordinates, this is particularly easy. For example, for data in an area around Vancouver, Canada, we are told the ellipse parameters (grs80) for a particular dataset, so:

    m_proj('utm','ellipse','grs80','lat',[49+15.7/60 49+21/60],...
            'long',[-123-15/60 -123-3/60]);      
    [LONG,LAT]=m_xy2ll(eastings,northings);
    

    In other cases, first find the projection information which is usually provided somewhere - in a README, or (perhaps) a .prj file (this is especially true for shapefile information). Examine the .prj text file. As an example, the Cascadia DEM contains data in coordinates defined by a file cascadia.prj:

    --------------cascadia.prj--------------
    Projection    LAMBERT                                                           
    Zunits        NO                                                                
    Units         METERS                                                            
    Spheroid      CLARKE1866                                                        
    Xshift        0.0000000000                                                      
    Yshift        0.0000000000                                                      
    Parameters                                                                      
     41 30  0.000 /* 1st standard parallel                                          
     50 30  0.000 /* 2nd standard parallel                                          
    -124 30  0.000 /* central meridian                                              
     38  0  0.000 /* latitude of projection's origin                                
    0.00000 /* false easting (meters)                                       
    0.00000 /* false northing (meters) 
    ---------------end of file----------------
    

    so to convert from projection coordinates back to lat/long you would use:

    m_proj('lambert','parallels',[41.5 50.5],'long',[-133 -116],...
                 'lat',[39 53],'false',[-124.5 38],'ellipsoid','clrk66');
    

    In another example, data from WA, USA is provided in a projection specified using:

    -------------------beginning of .prj file----------
     PROJCS["NAD_1983_HARN_StatePlane_Washington_South_FIPS_4602_Feet",
     GEOGCS["GCS_North_American_1983_HARN",
     DATUM["D_North_American_1983_HARN",
     SPHEROID["GRS_1980",6378137.0,298.257222101]],
     PRIMEM["Greenwich",0.0],
     UNIT["Degree",0.0174532925199433]],
     PROJECTION["Lambert_Conformal_Conic"],
     PARAMETER["False_Easting",1640416.666666667],
     PARAMETER["False_Northing",0.0],
     PARAMETER["Central_Meridian",-120.5],
     PARAMETER["Standard_Parallel_1",45.83333333333334],
     PARAMETER["Standard_Parallel_2",47.33333333333334],
     PARAMETER["Latitude_Of_Origin",45.33333333333334],
     UNIT["Foot_US",0.3048006096012192]]
     -----------------end of file-----------------------
    

    and we convert data back using:

     m_proj('lambert conformal','ellipsoid','grs80','par',[45.83333333333334 47.33333333333334],...
            'clong',-120.5,'false',[-120.5 45.33333333333334]);
    
      [LONG,LAT]=m_xy2ll( (X-1640416.666666667)*0.3048006096012192,Y*0.3048006096012192);
     

  7. Coastline Extractor
  8. In the past one could get high-resolution data from The Coastline Extractor, but as of 2015 this web site has been decommissioned.

  9. DCW political boundaries
  10. As of 2011 the DCW web site has been decommissioned. The following information is retained for historical reasons only. New users see the next section on Natural Earth.

    Files containing political boundaries for various countries and US states can be downloaded from http://www.maproom.psu.edu/dcw/. Select an area and choose the "download points" option (rather than "download data"). Once downloaded to your machine use m_plotbndry to access and plot the desired boundary. For example, if you downloaded various US states into a subdirectory "states:,

      m_plotbndry('states/arizona','color,'r')
      

    would plot arizona on the current map.

  11. Natural Earth Political Boundaries
  12. Political Boundary info is available in shapefile format from Natural Earth. Download the shapefiles for areas you are interested in and use m_shaperead as described above.

  13. GSHHS(G) high-resolution coastline database
  14. When drawing maps there is always a tradeoff between the execution time of the generating program and the resolution of the resulting map. Included in M_Map is a 1/4 degree coastline database which can be used to generate very fast maps, with adequate resolution for many purposes.

    However, it is often desirable to be able to make detailed maps of limited geographic areas. For this purpose a higher-resolution coastline database is necessary. I have not included such a database in M_Map because it would greatly increase the size of the package. However, I have included m-files to access and use a popular high-resolution database called GSHHS (as of 2016 now called GSHHG).

    As distributed, GSHHG consists of a hierarchical set of databases at different resolutions. The lowest or "crude" resolution is not as good as the M_Map database, although it contains many more inland lakes. The "high" resolution consists of points about 200m apart. There is also an even finer "full" resolution. You can install part or all of the database (depending on how much disk space you have available). The "full" resolution occupies 90Mb of disk space, and successively coarser resolutions are smaller by about 1/4. Thus "high" resolution occupies 21Mb, "intermediate" uses 6Mb, and "low" uses 1.1Mb (one reason for not always using "high" resolution is that the entire 90Mb database must be read and processed each call, which may take some time).

    1. Installing GSHHS
      1. Go to http://www.ngdc.noaa.gov/mgg/shorelines/data/gshhs/.

      2. Get gshhg-bin-2.3.6.zip (as of late 2017) and uncompress any or all of the files there - gshhs_*.b, wdb_borders_*.b, and wdb_rivers_*.b for coastlines, borders, and rivers respectively, in a convenient directory. One useful place is in m_map/private. GSHHS data format has changed between v1.2 and 1.3, and again for v2.0, but m_map should be able to figure this out.

      3. If the database files are not in subdirectory m_map/private, you must edit the FILNAME setting in m_gshhs.m to point to the appropriate files.

    2. Using GSHHS effectively
    3. The simplest calling mechanism is identical to that for m_coast (Section 3). For example, to draw a gray-filled high-resolution coastline, you just need

        m_gshhs_h('patch',[.5 .5 .5]);
        

      However, execution times may be very, very long, as the entire database must be searched and processed. I would not recommend trying to draw world maps with the intermediate or high-resolution coastlines! There are two ways to speed this up. The first is to use a lower-resolution database, with fewer points. The second is useful if you are going to be repeatedly drawing a map (because, for example, it's the base figure for your work). In this case I recommend that you save an intermediate processed (generally smaller) file as follows:

        m_proj ...  % set up projection parameters
        
        % This command does not draw anything - it merely processes the 
        % high-resolution database using the current projection parameters 
        % to generate a smaller coastline file called "gumby"
        
        m_gshhs_h('save','gumby');
        
        % Now we can draw a few maps of the same area much more quickly
        
        figure(1);
        m_usercoast('gumby','patch','r');
        m_grid;
        
        figure(2);
        m_usercoast('gumby','linewidth',2,'color','b');
        m_grid('tickdir','out','yaxisloc','left');
        
        etc.
        

      Note that there another way of getting this info, as well as river and border info, is with the m_gshhs.m function, whose first argument can be used to specify different options.

      m_gshhs('lc','patch','r');  % Low resolution filled coastline
      m_gshhs('fb1');             % Full resolution national borders
      m_gshhs('ir');              % Intermediate resolution rivers
      

9. Adding your own topography/bathymetry

A number of global and regional topography databases are available at NCAR . Several are available for free from their ftp site.

As long as the data is stored in a mat-file as a rectangular matrix in longitude/latitude, then m_contour or m_contourf can be used to plot that data.

  1. Sandwell and Smith Bathymetry
  2. A recent new bathymetry with approximately 1km resolution in lower latitude areas is being used by many people. This dataset is described at http://topex.ucsd.edu/marine_topo/  and is available as a 134Mb binary file at ftp://topex.ucsd.edu/pub/global_topo_2min/ (get the file topo_X.Y.img where X.Y is the version number) - note as of 2017 this is now a 729Mb binary at ftp://topex.ucsd.edu/pub/global_topo_1min/. I have included an m-file (mygrid_sand2.m) which can extract portions of the data (you will have to modify path names within the code). Once this database (and the m-file) is installed on your computer, you can use it in M_Map very easily. A typical usage is as follows:

      % Extract data
      [elevations,lat,lon]=mygrid_sand([long_west long_east lat_south lat_north ]);
      % Use in M_Map command
      m_contour(lon,lat,elevations);
      

    For some projections, you must make sure that the 'lon' values returned by mygrid_sand2.m fall within the range used in this projection (i.e. you may have to add/subtract 360). This seems to happen all the time for areas in the west (i.e. negative longitudes), if you forget this you often end up with bewildering error messages about empty vectors!

  3. TerrainBase 5-minute global bathymetry/topography
  4. THIS INFO IS KEPT FOR HISTORICAL REASONS - USE ETOPO1 (see below)

    For many purposes the elevation database accessed by M_Map provides adequate resolution. However, there are also many cases when more detail is desired. I have not included a higher-resolution database because it would greatly increase the size of the package. However, v1.2 includes m-files to access and plot a popular global 5-minute bathymetry/topography database, after a few minutes of work.

    This section provides instructions on how to download TerrainBase, and convert it from a 56Mb ASCII file to a 18Mb binary file using m_tba2b.m. It is then straightforward to access and plot bathymetry from this file using m_tbase.m, which is in every way functionally identical to m_elev (see Section 3.2).

    TerrainBase is also available on CDrom, and is also commonly stored in netcdf (or other) binary format somewhere on many academic networks. If you modify m_tbase.m to access data from one of these sources, let me know!

    How to install TerrainBase:

    1. get and uncompress the tbase.Z file from http://dss.ucar.edu/datasets/ds759.2/ into the m_map directory.
    2. Run m_tba2b('PATHNAME') to store the resulting 18Mb binary file as PATHNAME/tbase.int.
    3. Delete the original ASCII file tbase.
    4. Edit the PATHNAME setting in m_tbase to point to the location of this file.

    That's it! Test things out with this map of the western mediterranean:

    m_proj('lambert','lon',[-10 20],'lat',[33 48]);
    m_tbase('contourf');
    m_grid('linestyle','none','tickdir','out','linewidth',3);
    

  5. ETOPO 2 or 1-minute global bathymetry/topography
  6. ETOPO is a useful database, but it has undergone a number of changes in the last few years. Since it seems to be released primarily as a netCDF file now it is possible m_etopo2 should be completely rewritten, but instead I have just modified the "traditional" method of using the database from binary files. Read all of the points below and then follow the instructions in (4):

    1. (2004-2014 instructions: now obsolete), there is a corrected higher-resolution (2 minute) database ETOPO2. Download http://rda.ucar.edu/dsszone/ds759.3/etopo2_2006apr/etopo2_2006apr.raw.gz (a gzipped binary), gunzip it into a 116Mb file, edit the PATHNAME setting in m_etopo2 to point to the location of this file, and then use it in the same way as m_tbase and m_elev. UCAR requires users to register and the second link won't work without you doing this (go to first link and follow instructions).

    2. (2014-2017 instructions: mostly obsolete) In 2014, it was pointed out to me that the above is obsolete. First, there is a corrected 2-minute ETOPO database - ETOPO2v2 which you should be using instead. Now, ETOPO2v2 is a little more complicated, because it comes in 4 version - big-endian and little-endian, in both cell-centered and grid-centered versions.

      It doesn't particularly matter if you get big- or little-endian since you can modify the fopen line in m_etopo2 to account for this. I recommend getting the grid-centered version, since it works "better" when you are contouring the elevations (it will be more likely to extend all the way up to the map edge without weird little 'gaps').

      In any case, download one of the zipped binaries, unzip it, and then edit 4 lines in m_etopo2 to set the PATHNAME, the filename in the fopen line, as well as setting the last option to 'b' or 'l' for big-endian or little-endian formats. Then make sure the grid and resolution parameters are set appropriately. If you forget (or get them wrong), code may run but it won't give the right bathymetry!

    3. If you want even higher resolution bathymetry, you can also use the 1-minute ETOPO1. This appears to come in two versions: grid or cell-referenced, both little-endian. Again, I recommend the grid-referenced version etopo1_ice_g_i2.bin. Modify the relevant lines in m_etopo2 in the same way as for ETOPO2v2.

    4. (2017-present instructions: use these) As of 2017, notice that the file you want for ETOPO1, etopo1_ice_g_i2.bin is not available as a link from that page - instead they just reference a netCDF and a geo-referenced tiff version. Also, you will see two different versions that handle differences between the true surface and the top of the permanent icepacks in Greenland and Antarctica. Fear not! If you click on one of those, you fall into a directory page - click PARENT DIRECTORY, then binary, and you get into the place you want! For example, https://www.ngdc.noaa.gov/mgg/global/relief/ETOPO1/data/ice_surface/grid_registered/binary/ for the top-of-the-ice data.

      As in the above instructions, download the zipped binary "etopo1_ice_g_i2.zip", unzip it, and then edit 4 lines in m_etopo2 to set the PATHNAME, the filename in the fopen line, as well as setting the last option to 'b' or 'l' for big-endian or little-endian formats (as of 2017 this file seems to be available in little-endian format only). Then make sure the grid and resolution parameters are set appropriately. If you forget (or get them wrong), code may run but it won't give the right bathymetry - try a quick map to test it!


10. M_Map toolbox contents and description

  1. Contents.m - toolbox contents
  2. m_demo.m - demonstrates a few maps.

User-callable functions

  1. m_proj.m - initializes projection
  2. m_coord - geomagnetic to geographic coords

  3. m_grid.m - draws grids
  4. m_utmgrid.m - draws a UTM grid for UTM projection maps
  5. m_scale.m - forces map to a given scale
  6. m_ruler - draws a scale bar
  7. m_ungrid.m - erases map elements (if you want to change parameters)

  8. m_coast.m - draws a coastline
  9. m_elev.m - draws elevation data
  10. m_tbase.m - draws elevation data from high-resolution database
  11. m_etopo2.m - draws elevation data from (another) high-resolution database
  12. m_gshhs_c.m - draws coastline from GSHHS crude database
  13. m_gshhs_l.m - draws coastline from GSHHS low-resolution database
  14. m_gshhs_i.m - draws coastline from GSHHS intermediate-resolution database
  15. m_gshhs_h.m - draws coastline from GSHHS high-resolution database
  16. m_gshhs_f.m - draws coastline from GSHHS full resolution database
  17. m_plotbndry.m - draws a political boundary from the DCW
  18. m_usercoast.m - draws a coastline using a user-specified subset database.
  19. m_shaperead.m - reads ESRI shapefiles
  20. m_plot.m - draws line data in map coords
  21. m_line.m - draws line data in map coords
  22. m_text.m - adds text data in map coords
  23. m_legend.m - Draw a legend box
  24. m_patch.m - adds patch data in map coords
  25. m_pcolor.m - draws pcolor surface
  26. m_annotation.m- annotation lines/boxes/text

  27. m_quiver.m - draws arrows for vector data
  28. m_contour.m - draws contour lines for gridded data
  29. m_contourf.m - draws filled contours
  30. m_track.m - draws annotated tracklines
  31. m_range_ring.m - draws range rings
  32. m_streamline.m - draws streamlines
  33. m_ll2xy.m - converts from long/lat to map coordinates
  34. m_xy2ll.m - converts from map coordinates to long/lat
  35. m_geo2mag.m - converts from magnetic to geographic coords
  36. m_mag2geo.m - the reverse

  37. m_lldist.m - distance between long/lat points
  38. m_xydist.m - distance between map coordinate points

  39. m_fdist.m - location of point at given range/bearing along ellipsoidal earth
  40. m_idist.m - range/bearings between points on ellipsoidal earth
  41. m_geodesic.m - points on geodesics between given points on ellipsoidal earth
  42. m_tba2b.m - used in installing high-resolution elevation database.

  43. m_vec.m - fancy arrows
  44. m_windbarb.m - fancy arrows for meteorologists

  45. m_contfbar.m - draws colorbars for contourf plots
  46. m_colmap.m - useful perceptually uniform colourmaps
  47. m_shaperead.m - reads ESRI shapefiles
  48. mygrid_sand2.m - reads Sandwell and Smith bathymetry file

  49. wysiwyg.m - Sets figure window to match size/aspect of printed output

Internal functions (not meant to be user-callable)

  1. private/mp_azim.m - azimuthal projections
  2. private/mp_cyl.m - cylindrical projections (equatorial)
  3. private/mp_conic.m - conic projections
  4. private/mp_tmerc.m - transverse cylindrical projections
  5. private/mp_utm.m - elliptical universal transverse cylindrical projections
  6. private/mp_omerc.m - oblique cylindrical projection

  7. private/mu_util.m - various utility routines
  8. private/mu_coast.m - routines to handle coastlines.
  9. private/mc_coords - coordinate conversion.
  10. private/clabel.m - patched version of clabel (matlab v5.1 version does not contain capabilities for different text properties).

  11. private/m_coasts.mat - coastline data

HTML Documentation

  1. map.html - documentation intro
  2. mapug.html - users guide
  3. various .png in doc/ - examples.

11. Known Problems and Bugs

  1. Lakes come out black! If plotted data is coloured white, the print changes it to black in output figures. In order to avoid this, set the figure background to white, i.e.
        set(gcf,'color','white')
        

  2. Filled coastlines go weird. If you try to use azimuthal projections covering nearly half the globe, or oblique mercator projections over large areas, sometimes coastline patching fails (this is due to rounding errors in the projection math fpor points very near the map boundary). To solve this you can either
    • Adjust the projection parameters (lat/lon limits, radius, etc.) VERY SLIGHTLY so that the problematic points are no longer as close to the map boundary, or
    • Use line coastlines instead of patches.
  3. Generally weird-looking stuff that happens when you use filled contours. For some reason this has been a glory-hole for all kinds of weird bugs in MATLAB. Most of them relate somehow to the way in which the map background interacts with contourf patches, and how the 'renderer' (the internal matlab code that figures out what goes on top of what) works, or doesn't work. Unfortunately I can't think of way that works around the problem in all cases, but if you see something weird, try:
     set(findobj('tag','m_grid_color'),'facecolor','none') 

    after the

    m_grid
    call, or
     set(gcf,'renderer','opengl'); 
    (under Unix you may have to do this one on starting MATLAB)
  4. Things not appearing correctly in tiff output. Pre 2014b Matlab uses ghostscript to covert from ps to many other formats. But their version has some problems. It may be better to print to postscript and do the conversion (say, to tiff) yourself.

12. OCTAVE Compatibility Issues

From their website: "GNU Octave is a high-level interpreted language, primarily intended for numerical computations. [...] The Octave language is quite similar to Matlab so that most programs are easily portable."

M_Map currently runs under Octave...mostly. m_demo runs pretty much perfectly (depending on which version of Octave you are using). However, there are features that either don't work, or don't work properly. Unfortunately the rather obscure details of different functionalities that M_Map relies on can change from release to release; issues that were "fixed" may "break" in later releases. Feel free to contact me if you have problems; I may have solutions (or I may not).

Features that don't work, or don't work well when I last tested this under Octave 4.4.1, include:

  1. Everything seems to run fine as long as you are in the m_map home directory, but Octave crashes for me if I try to run m_map from other directories, relying on the load path.

  2. m_contourf, which is really just a call to contourf, doesn't work correctly sometimes, because the Octave contourf does not handle NaNs properly, and NaNs can appear if data fields extend beyond the map boundaries and must be clipped. Even when it does work, it can take a VEERRRYYY LOOOONNNNGG TIIIIMMMMEEEE to run. Staggeringly long, sometimes.

  3. Latitude and Longitude labels don't appear to rotate, even if you specify a rotation.

  4. X/YLabels don't appear (they seem to be tied to the axes 'visible' property, which M_Map sets to 'off')

  5. m_image won't work for projections other than equidistant cylindrical.

13. Changes since last release

  1. m_shadedrelief.m changed and m_image.m added. Lots of small changes in an attempt to improve Octave compatibility.