Comparison & Integration Shell
(COINS)

An integrative modelling tool for carbon accounting and natural resource management

CRC for Greenhouse Accounting
Ecosystem Dynamics Group RSBS, ANU

Introduction | Services Provided by COINS | Key Features | Availability

Key Features

Spatial scale

The basic spatial unit within COINS is the grid-cell (Fig. 1a). A model can be run within a single grid-cell, which is conceptually equivalent to running at a single ‘point’. Several single grid-cells can also be run simultaneously (Fig. 1b). Alternatively, models can be embedded within a geo-referenced map, effectively turning the COINS shell into a GIS simulator (Figure 1c, d). Conceptually, there are therefore two spatial modes within COINS, point and spatial. The dimensions of the cells are open to the user to define, therefore making the spatial scale of analysis completely flexible.

Fig. 1. The COINS software is completely flexible in the spatial scale of analysis, and can implement models running at a single ‘point’ (a); a collection of points (b); through to full spatial mode where the model(s) run across multiple cells within a GIS-type framework (c), (d).

When running in spatial mode, there are a number of ways in which models can be represented and combined (Fig. 2).

• The same model and parameterisation can be run across the whole spatial domain (Fig. 2a).
• Different versions (or instances) of the same model can be run at different locations within the same spatial domain. For example a model may be parameterised separately for ‘grassland’, ‘forest’ etc., and these instances of the model are assigned spatial locations (Fig. 2b).
• Different models can be run at different locations, for example a grass model in one region, and a tree model in another (Fig. 2c).
• Within-cell variability can be incorporated through allowing a proportional representation of different models (or different instances of the same model) within single grid cells (Fig. 2c,d). In this case models are assigned to a proportion of the total area of the grid-cell.

Fig. 2. Examples of how different models, and different parameterisations of the same model, can be combined within the COINS environment.

Temporal scale.

There are three temporal scales at which models can be combined; daily, monthly and yearly. The temporal resolution is implemented via three nested loops, and models are called only when appropriate. For example, within the same simulation, model A could be running at a daily timestep, and model B at monthly. The simulation proceeds at a weekly timestep, but model B is called only at monthly intervals.

Databases

Underlying the COINS shell, and accessible to all models, are a range of databases for running models of terrestrial carbon dynamics in Australian ecosystems. These include continental surfaces of long-term monthly average climate variables (e.g. minimum temperature, maximum temperature, precipitation and evaporation), and a wide range of other environmental data such as land-cover classifications, soils data etc. CRC members can view available datasets at http://www.greenhouse.crc.org.au/members/datasets.cfm

The extensive CRC for Greenhouse Accounting historical climate database is also available to COINS simulations, and comprises the following monthly interpolated climate variables (January 1900-December 2001); minimum temperature, maximum temperature, precipitation and evaporation (January 1970-December 2001). Development of this climate database is an ongoing project.

Alternatively, users can provide their own data specific to their own application. All spatial datasets are imported into COINS as arrays of 32-bit floating point numbers, using the ARCINFO *.flt format (with an associated *.hdr header file). For point-scale simulations, users can build their own database in their choice of spreadsheet or database program. COINS currently reads this information as a tab-delimited text file.

Runtime visualization of model outputs.

Any combination of output variables can be viewed as a simulation progresses, depending upon the type of simulation selected (point vs. spatial), and the parameters and variables defined in the model. These outputs include scalar quantities, vectors, matrices (e.g. maps) and XY-scatter plots (including plots of output variables vs. time) (Fig.3). Tables of observations can also be imported into COINS, and as a simulation progresses these observations can be graphically compared against the model outputs, allowing ‘visual’ run-time validation of models to be assessed.

Fig. 3. Three different ways of viewing run-time output in COINS. (a) Maps, (b) Scalar quantities, (c) timelines.

Scheduling events in COINS

An event scheduler allows complex simulations to be built, specifying events such as disturbance, fertilizer addition etc. The scheduler also allows replicate model runs to be performed, for example performing factorial experiments via the sequential modification of input parameters. The scheduler has its own command syntax. Scheduler commands can also be entered one-line-at-a-time through the command-line prompt (Fig 4.)

Fig. 4. The scheduler in COINS allows complex simulation runs to be defined, such as factorial experiments, and the scheduling of specific events to occur at specific times. A schedule can be saved as a batch file for later processing.

Vector and Matrix manipulations

COINS contains a small utility which allows a number of vector and matrix-based operations to take place. Example include filling matrices with various sorts of random and regular patterns, rotate, edit and crop maps, and saving maps in different formats.

Data handling

In any GIS-type application there are significant issues regarding the storage and retrieval of large amounts of information. For example, the monthly historical precipitation database alone comprises approximately 3Gb of disk space. COINS manages large amounts of data through a consistent naming convention, and by creating sub-set copies of the required input datasets for a given simulation from the primary data source.

Map viewer

For closer interrogation of maps COINS has a ‘duel resolution’ GIS map viewer, with a scaleable background image, and an independently scaleable and moveable ‘magnifying glass’ (Fig. 5). The Map viewer allows ‘zooming in’ of particular areas, and also contains a data-drill for interrogating the current model outputs for any given cell, and for viewing the contents of that cell as timelines as the simulation proceeds.

Fig. 5. The coins map viewer allows maps to be interrogated simultaneously at two spatial scales, through the moveable magnifying glass (a). A right-mouse-click on any grid-cell initializes the data drill, and allows all model outputs for that cell to be viewed as either text or graphs as the simulation proceeds.

Monte-Carlo capability

Uncertainty analyses are conducted in COINS via Monte-Carlo simulation. An interactive dialog box allows the user to select parameter values from any one of 22 continuous or 6 discrete probability distribution functions, allowing sensitivity of model outputs to input parameters to be readily assessed, and to provide estimates of uncertainty around model results. The Monte-Carlo interface is similar to the commercially available @Risk software, with graphical representations of the distributions, and the ability to specify correlation structures among random parameters (Fig. 6).

Fig. 6. The COINS Monte-Carlo option allows input parameters to be selected at random from any one of 22 continuous or 6 discrete random distributions. Replicate runs of the model are then performed, allowing sensitivity of model outputs to uncertainty in model inputs to be quantified.

Model initialization wizard, and the *.cis file

A new simulation can be set-up via a model initialization wizard, which sequentially prompts the user for all the information required to define the model or models to be used in the simulation, the input data required, the outputs to be graphed, the masks to be used, etc. The wizard automatically creates the main COINS *.cis run file. This is a ASCII text file, which contains all of the information required to load and run a simulation. One created, it can be edited off-line and re-opened in coins for ‘batch-mode’ type operation. The initialization or ‘initial conditions’ files are kept separate from the model run file, to allow different initialization conditions to be specified and saved. The *.cis file also maintains links with any scheduled events that may have been specified.

Datalogging

As a simulation proceeds a range of options are available for automatically logging model results for post-simulation analyses. This includes maps in GIS format, time-lines of output variables, and XY plots.

Data workbench

A data workbench is also provided to allow limited analysis of model outputs, including pairwise comparison of maps on a cell-by-cell basis, and visualization of the sensitivity of model outputs to variations in model inputs from Monte-Carlo analyses

Inter-cell communication

Models which require inter-cell communication, such as cellular automata, are also compatible with the COINS environment. A simple examples is shown in Fig 7. Potential applications include NRM problems that demand the representation of lateral fluxes of material or events, such as hydrological flows or fire spread, or ecological models with spatially spreading processes such as seed dispersal or animal movement.

Fig. 7. Models which require inter-cell communication, such as cellular automata (CA), can be readily incorporated into COINS. The example above shows the COINS implementation of Conway’s classic ‘game of life’. Other more complex simulations, such as models of hydrological flux or fire-spread, can be similarly implemented.

Availability

A working version of COINS is available, but it is still very much in beta-testing mode. Any interest in using the COINS Shell should be lodged with the CRC for Greenhouse Accounting.

Contacts:

Ian Davies: Ian.Davies@greenhouse.crc.org.au
Stephen Roxburgh: Stephen.Roxburgh@greenhouse.crc.org.au

Services provided by COINS environment<< Previous