makes several functions available to access and ana- lyse pre-calculated back ..... in Environmental Modelling & Sof
source
OPEN
openair News The openair Project newsletter
Issue 10, October 2011
The openair project officially ends! But . . . The openair project was jointly funded by the UK Natural Environment Research Council (NERC) and Department for Environment, Food and Rural Affairs (Defra) until the end of September 2011. The 3 years have gone quickly and we have received some excellent feedback and support from users. It has become clear to us that developing an R package that is opensource, freely available and builds on R’s many capabilities is an effective mix. It has provided access to techniques that otherwise would have been unavailable and the interactions with users has led to many improvements — which is just what we wanted to happen. It is also good to see that we have a roughly equal split in users between academia, consultancies and governmental bodies. One of the aims of the original proposal was to make the project sustainable i.e. ensure that it develops enough momentum to continue beyond the direct funding. We have worked hard to develop systems and ways of working to ensure that this happens e.g. to automate many processes, using a proper version control system etc. All these developments will make it easier to maintain the momentum that has been gained. In some ways the project is just taking off. Every week there are enquiries from around the world — in fact it has been surprising the extent to which we have had international interest in the project. R itself even over the past 3 years has grown enormously
and continues to develop rapidly. For these reasons and more we will be doing what we can to keep the project going: both refining existing functions and developing new ones. We will of course be seeking additional funding to help things move along and would always be happy to hear of potential funding sources from users! Turning to this issue of the newsletter, there is a focus on calculating and using back trajectories. Back trajectories are widely used in the research community but less so elsewhere. One of the principal difficulties is being able to calculate and use back trajectories easily — and this is something we have taken steps towards addressing in openair. This is just the start however and in time we hope that these functions can be developed to become powerful tools for analysing air quality data. The sole openair contact will now be David Carslaw, who should be contacted concerning any openair matters. I would like to thank Karl for his many contributions over the past three years which have helped develop openair into a mature and widely used R package. Finally thanks to all of you for your support and input — which I hope will continue! This newsletter was produced using R version 2.13.2 and openair version 0.5-11.
Contents
Back trajectory functions in openair . . . . The corPlot function — correlation matrices Other openair developments . . . . . . . . .
The openair project officially ends! But . . . .
1
mailto:
[email protected]
2 7 9
Issue 10, October 2011
2
Back trajectory functions in openair Back trajectories are extremely useful in air pollution and can provide important information on air mass origins. Despite the clear usefulness of back trajectories, their use tends to be restricted to the research community. Back trajectories are used for many purposes from understanding the origins of air masses over a few days to undertaking longer term analyses. They are often used to filter air mass origins to allow for more refined analyses of air pollution — for example trends in concentration by air mass origin. They are often also combined with more sophisticated analyses such as cluster analysis to help group similar type of air mass by origin. Perhaps one of the reasons why back trajectory analysis is not carried out more often is that it can be time consuming to do. This is particularly so if one wants to consider several years at several sites. It can also be difficult to access back trajectory data. In an attempt to overcome some of these issues and expand the possibilities for data analysis, openair makes several functions available to access and analyse pre-calculated back trajectories. Currently these functions allow for the import of pre-calculated back trajectories are several pre-define locations and some trajectory plotting functions. In time all of these functions will be developed to allow more sophisticated analyses to be undertaken. Also it should be recognised that these functions are in their early stages of development and will may continue to change and be refined. This importTraj function imports pre-calculated back trajectories using the HYSPLIT trajectory model (Hybrid Single Particle Lagrangian Integrated Trajectory Model http://ready.arl.noaa.gov/HYSPLIT. php). Trajectories are run at 3-hour intervals and stored in yearly files (see below). The trajectories are started at ground-level (10m) and propagated backwards in time. The data are stored on web-servers at King’s College London in a similar way to importKCL, which makes it very easy to import pre-processed trajectory data for a range of locations and years. Note — the back trajectories have been pre-calculated for specific locations and stored as .RData objects. Users should contact David Carslaw to request the addition of other locations. When new trajectory locations are added, preference will be given to locations that are likely to be of interest widely e.g. an EMEP site, or a large city, a site with 1
lots of measurements that are publicly available etc. So far only a few receptors are available to users but in time the number will increase. It should be feasible for example to run back trajectories for the past 20 years at all the EMEP sites in Europe.1 These trajectories have been calculated using the Global NOAA-NCEP/NCAR reanalysis data archives. The global data are on a latitude-longitude grid (2.5 degree). Note that there are many different meteorological data sets that can be used to run HYSPLIT e.g. including ECMWF data. However, in order to make it practicable to run and store trajectories for many years and sites, the NOAA-NCEP/NCAR reanalysis data is most useful. In addition, these archives are available for use widely, which is not the case for many other data sets e.g. ECMWF. HYSPLIT calculated trajectories based on archive data may be distributed without permission (see http://ready. arl.noaa.gov/HYSPLIT_agreement.php). For those wanting, for example, to consider higher resolution meteorological data sets it may be better to run the trajectories separately. Users should see the help file for importTraj to get an up to date list of receptors where back trajectories have been calculated. As an example, we will import trajectories for London in 2010. Importing them is easy: traj
