Merge pull request #200 from edoddridge/paper-update

Updated author affiliations for JOSS paper

paper/paper.md

@@ -12,9 +12,9 @@ authors:
  - name: Alexey Radul
    affiliation: 2
 affiliations:
- - name: Massachusetts Institute of Technology, Earth, Atmospheric and Planetary Sciences
+ - name: Earth, Atmospheric and Planetary Science, Massachusetts Institute of Technology, Cambridge, MA, USA
    index: 1
- - name: Massachusetts Institute of Technology
+ - name: Brain and Cognitive Science, Massachusetts Institute of Technology, Cambridge, MA, USA
    index: 2
 date: 5 December 2017
 bibliography: paper.bib
@@ -26,7 +26,7 @@ Aronnax is a highly idealised model for simulating large-scale (100-1000km) flow
 
 Aronnax is an _isopyncal_ model: it approximates the ocean as a number of discrete homogeneous layers with spatially variable thicknesses. These layers are stacked vertically and the density difference between neighbouring layers is specified by the user. Other widely used vertical coordinates require solving the equations of motion at specified vertical levels where the density is allowed to vary [@Griffies2000]. Representing the large-scale ocean circulation in layers contributes to Aronnax's speed: one needs only a few layers to achieve the same fidelity as 50 or more fixed vertical levels [@Stewart2017].
 
-Aronnax serves three distinct purposes. Firstly, many of the studies that use a model like Aronnax do not provide the source code, see e.g. [@Davis2014,@Fevrier2007,@Johnson2002a,@Stern1998]. This increases the likelihood that coding errors go undetected, and requires that each research group spend time developing their own idealised model. Aronnax solves these problems by providing an open source, tested model for the community to use. Secondly, Aronnax furthers the goals of scientific reproducibility since a simulation can be entirely specified with a set of input files and a version number. Thirdly, Aronnax provides an easy-to-use model that may be included in future modelling hierarchies with minimal effort, thereby enabling new research questions to be addressed.
+Aronnax serves three distinct purposes. Firstly, many of the studies that use a model like Aronnax do not provide the source code, see e.g. [@Davis2014, @Fevrier2007, @Johnson2002a, @Stern1998]. This increases the likelihood that coding errors go undetected, and requires that each research group spend time developing their own idealised model. Aronnax solves these problems by providing an open source, tested model for the community to use. Secondly, Aronnax furthers the goals of scientific reproducibility since a simulation can be entirely specified with a set of input files and a version number. Thirdly, Aronnax provides an easy-to-use model that may be included in future modelling hierarchies with minimal effort, thereby enabling new research questions to be addressed.
 
 There are a number of other publicly available ocean models. Of these the most abundant are general circulation models and quasigeostrophic models. General circulation models such as [NEMO](https://www.nemo-ocean.eu/), [GOLD](https://www.gfdl.noaa.gov/gold-ocean-model/), [MOM6](https://github.com/NOAA-GFDL/MOM6), and [MITgcm](http://mitgcm.org/) solve a less idealised version of the Navier-Stokes equations and can be coupled with sea ice and atmospheric models to create fully coupled climate models. Because the underlying equations are derived with fewer approximations these models can more faithfully simulate a wider range of flow regimes. However, this comes at a price; general circulation models are extremely complex, with numerous free parameters that must be specified, often prior to compiling the source code. It is possible to use most of these models in idealised configurations, but doing so requires a substantial investment of time from the user, and non-trivial computing resources. In comparison, Aronnax is easy to install and cheap to run.