Skip to contents


This function plots back trajectories on a ggplot2 map. This function requires that data are imported using the openair::importTraj() function. It is a ggplot2 implementation of openair::trajLevel() with many of the same arguments, which should be more flexible for post-hoc changes.


  longitude = "lon",
  latitude = "lat",
  facet = "default",
  smooth = FALSE,
  statistic = "frequency",
  percentile = 90, = 1, = 1,
  min.bin = 1,
  .combine = NA,
  sigma = 1.5,
  alpha = 0.5,
  tile.border = NA,
  xlim = NULL,
  ylim = NULL,
  crs = sf::st_crs(4326),
  map = TRUE,
  map.fill = "grey85",
  map.colour = "grey75",
  map.alpha = 0.8,
  map.lwd = 0.5,
  map.lty = 1,



Data frame, the result of importing a trajectory file using openair::importTraj().

latitude, longitude

The decimal latitude/longitude.


Pollutant to be plotted. By default the trajectory height is used.


Used for splitting the trajectories into different panels. Passed to openair::cutData().


Should the trajectory surface be smoothed? Defaults to FALSE. Note that smoothing may cause the plot to render slower, so consider setting crs to sf::st_crs(4326) or NULL.


Statistic to use for trajLevel(). By default, the function will plot the trajectory frequencies (statistic = "frequency"). As an alternative way of viewing trajectory frequencies, the argument method = "hexbin" can be used. In this case hexagonal binning of the trajectory points (i.e., a point every three hours along each back trajectory). The plot then shows the trajectory frequencies uses hexagonal binning.

There are also various ways of plotting concentrations.

It is possible to set statistic = "difference". In this case trajectories where the associated concentration is greater than percentile are compared with the the full set of trajectories to understand the differences in frequencies of the origin of air masses. The comparison is made by comparing the percentage change in gridded frequencies. For example, such a plot could show that the top 10\ tend to originate from air-mass origins to the east.

If statistic = "pscf" then a Potential Source Contribution Function map is produced. This statistic method interacts with percentile.

If statistic = "cwt" then concentration weighted trajectories are plotted.

If statistic = "sqtba" then Simplified Quantitative Transport Bias Analysis is undertaken. This statistic method interacts with .combine and sigma.


The percentile concentration of pollutant against which the all trajectories are compared.,

The longitude and latitude intervals to be used for binning data.


The minimum number of unique points in a grid cell. Counts below min.bin are set as missing.


When statistic is "SQTBA" it is possible to combine lots of receptor locations to derive a single map. .combine identifies the column that differentiates different sites (commonly a column named "site"). Note that individual site maps are normalised first by dividing by their mean value.


For the SQTBA approach sigma determines the amount of back trajectory spread based on the Gaussian plume equation. Values in the literature suggest 5.4 km after one hour. However, testing suggests lower values reveal source regions more effectively while not introducing too much noise.


Opacity of the tiles. Must be between 0 and 1.


Colour to use for the border of binned tiles. Defaults to NA, which draws no border.

xlim, ylim

The x- and y-limits of the plot. If NULL, limits will be estimated based on the lat/lon ranges of the input data.


The coordinate reference system (CRS) into which all data should be projected before plotting. Defaults to latitude/longitude (sf::st_crs(4326)).


Should a base map be drawn? Defaults to TRUE.


Colour to use to fill the polygons of the base map (see colors()).


Colour to use for the polygon borders of the base map (see colors()).


Transparency of the base map polygons. Must be between 0 (fully transparent) and 1 (fully opaque).


Line width of the base map polygon borders.


Line type of the base map polygon borders. See ggplot2::scale_linetype() for common examples.


Arguments passed on to ggplot2::coord_sf


If TRUE, the default, adds a small expansion factor to the limits to ensure that data and axes don't overlap. If FALSE, limits are taken exactly from the data or xlim/ylim.


CRS that provides datum to use when generating graticules.


Character vector indicating which graticule lines should be labeled where. Meridians run north-south, and the letters "N" and "S" indicate that they should be labeled on their north or south end points, respectively. Parallels run east-west, and the letters "E" and "W" indicate that they should be labeled on their east or west end points, respectively. Thus, label_graticule = "SW" would label meridians at their south end and parallels at their west end, whereas label_graticule = "EW" would label parallels at both ends and meridians not at all. Because meridians and parallels can in general intersect with any side of the plot panel, for any choice of label_graticule labels are not guaranteed to reside on only one particular side of the plot panel. Also, label_graticule can cause labeling artifacts, in particular if a graticule line coincides with the edge of the plot panel. In such circumstances, label_axes will generally yield better results and should be used instead.

This parameter can be used alone or in combination with label_axes.


Character vector or named list of character values specifying which graticule lines (meridians or parallels) should be labeled on which side of the plot. Meridians are indicated by "E" (for East) and parallels by "N" (for North). Default is "--EN", which specifies (clockwise from the top) no labels on the top, none on the right, meridians on the bottom, and parallels on the left. Alternatively, this setting could have been specified with list(bottom = "E", left = "N").

This parameter can be used alone or in combination with label_graticule.


Method specifying how scale limits are converted into limits on the plot region. Has no effect when default_crs = NULL. For a very non-linear CRS (e.g., a perspective centered around the North pole), the available methods yield widely differing results, and you may want to try various options. Methods currently implemented include "cross" (the default), "box", "orthogonal", and "geometry_bbox". For method "cross", limits along one direction (e.g., longitude) are applied at the midpoint of the other direction (e.g., latitude). This method avoids excessively large limits for rotated coordinate systems but means that sometimes limits need to be expanded a little further if extreme data points are to be included in the final plot region. By contrast, for method "box", a box is generated out of the limits along both directions, and then limits in projected coordinates are chosen such that the entire box is visible. This method can yield plot regions that are too large. Finally, method "orthogonal" applies limits separately along each axis, and method "geometry_bbox" ignores all limit information except the bounding boxes of any objects in the geometry aesthetic.


Number of segments to use for discretising graticule lines; try increasing this number when graticules look incorrect.


Is this the default coordinate system? If FALSE (the default), then replacing this coordinate system with another one creates a message alerting the user that the coordinate system is being replaced. If TRUE, that warning is suppressed.


Should drawing be clipped to the extent of the plot panel? A setting of "on" (the default) means yes, and a setting of "off" means no. In most cases, the default of "on" should not be changed, as setting clip = "off" can cause unexpected results. It allows drawing of data points anywhere on the plot, including in the plot margins. If limits are set via xlim and ylim and some data points fall outside those limits, then those data points may show up in places such as the axes, the legend, the plot title, or the plot margins.


A ggplot2 plot

See also

the original openair::trajLevel()

trajLevelMap() for the interactive leaflet equivalent of trajLevelMapStatic()

Other static trajectory maps: trajMapStatic()