A web application from:


With the support of:

GCM compareR

GCM compareR is a web application developed to assist ecologists, conservationists and policy makers at understanding climate change scenarios and differences between Global Circulation Models (GCMs), and at assisting the triage of subsets of models in an objective and informed manner. GCM compareR is written in R and uses the web app development package shiny. The code of this app can be find in the project’s github, https://github.com/marquetlab/GCM_compareR.

The number of GCMs that are accessible to researchers and practitioners has grown large. Concretely, meteorological research centers worldwide have contributed more than 35 different GCMs for four distinct climate change scenarios as part of the Coupled Model Intercomparison Project Phase 5 (CMIP5; (Taylor, Stouffer, and Meehl 2012)). All these models have shown good performance and skill in predicting historical climatic data, but present differences among them as a result of different sources of uncertainty (including model formulation, resolution and sensitivity to initial conditions, climate noise; (Flato et al. 2013)). GCMs could be ranked by their skill at specific geographic areas, but models that most accurately predict historic data are not necessarily the most useful for making future climate projections (Knutti 2008).

In practice, best practices when conducting any evaluation advice for using multi-model approaches where differences in GCMs projections are adequantely understood and assessed as uncertainty (Pierce et al. 2009, Flato et al. (2013)). Also, and even though the ideal case would use all available GCMs, researchers are often forced to work with a few selected models for computational restrictions (Barsugli et al. 2013). However, the choice of some GCMs and not other has the potential to influence results (Synes and Osborne 2011), and thus it should be made following informed and replicable procedures (P. Mote et al. 2011, Snover et al. (2013), Vano et al. (2015)).

GCM compareR has been design to serve the purpose of informing about differences and similarities between GCMs and climate change scenarios, and of assisting the triage of models that best suit every used needs.


Use of the App

GCM compareR contains tabs that might be used from left to right to define a comparison scenario, retrieve results and generate a report with them.

  • The Intro tab includes all the information needed to use the app. Move the Workflow section to find full details about to use the app, and go to About to find information about who developed it.
  • In Scenario you will be able to make all choices: select the GCMs you would like to compare, pick a climate change scenario (year of projection, RCP…) and set the geographic extent of your analysis. Use the Analyse button on this tab to trigger the start of the analyses.
  • The tabs Explore selected GCMs, Variation from present and Variation among futures will display the results after the calculation is completed. Finally, Report will download a report with all the figures produced and some explanatory text.

Lastest news

  • Release of GCM compareR (Sep 26, 2018)

Citation

Please, if you use GCM compareR as part of your research, cite the app as:

Fajardo, J, Corcoran, D., Roehrdanz, P, Hannah, P, Marquet, P (2018) GCM compareR: A web application to assess differences and assist in the selection of global circulation models for climate change research (in prep).


Contact us

Please, email derek.corcoran.barrios@gmail.com with any question of create an issue on github.


References

Barsugli, Joseph J, Galina Guentchev, Radley M Horton, Andrew Wood, Linda O Omearns, Xin-Zhong Liang, Julie A Winkler, et al. 2013. “The practitioner’ s dilemma : How to assess the credibility of downscaled climate projections.” Eos 94 (46): 424–25.

Flato, G., J. Marotzke, B. Abiodun, P. Braconnot, S.C. Chou, W. Collins, P. Cox, et al. 2013. “Evaluation of Climate Models.” In Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, edited by T.F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P.M. Midgley. Cambridge, United Kingdom; New York, NY, USA: Cambridge University Press.

Knutti, Reto. 2008. “Should we believe model predictions of future climate change?” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 366 (1885): 4647–64. doi:10.1098/rsta.2008.0169.

Mote, Philip, Levi Brekke, Philip B. Duffy, and Ed Maurer. 2011. “Guidelines for constructing climate scenarios.” Eos 92 (31): 257–64. doi:10.1029/2011EO310001.

Pierce, David W, Tim P Barnett, Benjamin D Santer, and Peter J Gleckler. 2009. “Selecting global climate models for regional climate change studies.” Proceedings of the National Academy of Sciences 106 (21): 8441–6. doi:10.1073/pnas.0900094106.

Snover, Amy K., Nathan J. Mantua, Jeremy S. Littell, Michael A. Alexander, Michelle M. Mcclure, and Janet Nye. 2013. “Choosing and Using Climate-Change Scenarios for Ecological-Impact Assessments and Conservation Decisions.” Conservation Biology 27 (6): 1147–57. doi:10.1111/cobi.12163.

Synes, Nicholas W., and Patrick E. Osborne. 2011. “Choice of predictor variables as a source of uncertainty in continental-scale species distribution modelling under climate change.” Global Ecology and Biogeography 20 (6): 904–14. doi:10.1111/j.1466-8238.2010.00635.x.

Taylor, Karl E., Ronald J. Stouffer, and Gerald A. Meehl. 2012. “An overview of CMIP5 and the experiment design.” Bulletin of the American Meteorological Society 93 (4): 485–98. doi:10.1175/BAMS-D-11-00094.1.

Vano, Julie A., John B. Kim, David E. Rupp, and Philip W. Mote. 2015. “Selecting climate change scenarios using impact-relevant sensitivities.” Geophysical Research Letters 42 (13): 5516–25.


Workflow

GCM compareR web App is designed in tabs to separate the processes of defining a comparison scenario and retrieve the results. This tabs are generally used from left to right. The following is a description of the most common use of the app:

1. Definition of scenario

The tab Scenario is used to define the comparison to perform. The tab is composed by a side panel, where the options are selected, and a main panel including a map where the area of interest is delimited.

To start, use the options in the side panel to define the comparsion of GCMs:

a. Global Circulation Models (GCMs). Check the climatic models of interest and uncheck those that should not be included in the comparison. Use the tab GCM details above the map to toggle a table with details for any climate model. Not all models are available for all combination of scenarios (GCMs and year).
b. Climate Change Scenario. Select a future year and a Representative Concentration Pathway (RCP) to define the scenario. RCPs are standardized scenarios of future climate defined by varying emissions resulting from land use, population growth and technology (Moss et al. 2010).

c. Type of comparison. Two posibilities are currently available: (a) compare one bioclimatic variable against another bioclimatic variable, and (b) compare two groups of bioclimatic variables. We recommend comparing temperature against precipitation variables (i.e. the default selection compares mean annual temperature agains annual precipitation). In the special case of comparing groups of variables, it is important to combine in the same group only variables that are modulated in a similar fashion (e.g. for all variables in the group higher is more and smaller is less). Notice that projections of these variables will be scaled and combined; not following this advice will make the interpretation of the results hard to interpret and likely confusing. When two single variables are compared, two types of results are produced: with the actual values of the variables and with scaled values. When groups of variables are compared, only the result of scaled values is presented, as this is a requirement for model combination.
d. Study area. Select one of the five options provided here and the interactive map to define your area of interest. Only pixels within the selected area will be compared.

  • Define a squarred area: this can be done through the first (Select drawing a rectangle over the map) and the last options (Enter bounding-box coordinates). The draw the rectangle over the map, click first the black square icon on the upper left corner of the map, and drag later a rectangle on the map. The other option will draw the rectangle automatically after you insert minimum and maximum longitude and latitude values (this option may be prefered for replicability).
  • Define the study area by entering countries, biomes or ecoregions names. To use the option, select one of the other three options (Select by country, Select by biome, Select by ecoregion) and enter the name(s) of the item(s) in the box that will appear. More than one name may be selected. Once you enter one or more names, the map will refresh and display the selection in red (this might take a few seconds depending on your connection). This type of selections can be refined by using drawn rectangles. For instance, if you want to study the Mediterranean forests, woodlands and shrubs of Chile in South America, you can enclose this part of the biome in a rectangle right after the biome name has been selected and the map has turned red.

Once you are done with selecting the study area and options, press the ANALYZE green button below the map to proceed with the calculations. Once you have pressed the button, the following tabs will become selectable and you will see a progress bar informing the execution of the analyses. You may always come back to this tab to modify your defined scenario.

2. Check the results

The results are presented in next three tabs.

- Explore selected GCMs. This tab displays the climatic models select with a common color scale that permits their comparison. Each selected bioclimatic variable is shown as a different group of maps. The first map in each group shows the “baseline” variable (current climatic conditions), and the second shows the average of the values for each pixel across all GCMs.

- Variation from present. This tab includes the results of comparing all models with the current climate. It is useful to learn about the spread of GCMs in the context of their differences with the current conditions for a particular escenario in the area selected. The spread of mean differences can be found under the subtab Spread of GCMs, and maps showing the spatial pattern of these differences are in the subtab Maps of differences. The scatterplot and table can be shown displaying the projected values for the variables (e.g. total annual precipitation projection for each model) or their difference in respect to current climate (e.g. projected increase or decrease in annual precipitation).

- Variation among futures. This tab includes the results of comparing all models against the average (or ensemble) projection in the future scenario. It is useful to learn about the spread of GCMs with focus on how different are they from the average projected values for a particular escenario in the area selected. The spread of mean differences can be found under the subtab Spread of GCMs, and maps showing the spatial pattern of these differences are in the subtab Maps of differences. This comparison may be done using unscaled variable values or scaled (the only option when the comparison involves more than a couple of variables).

All three tabs include visualization options to add the borders of countries, biomes or ecoregions to the maps. These maps can be downloaded by right clicking on the image and selecting the option to save. Scatterplots and tables can also be downloaded by using each particular download buttons (tables are not part of the reports, so do not forget downloading them from here if you want to use them later).

3. Download a report with all results

This last tab downloads a report including all produced and visualized results (except for tables, that might be download separatedly in .csv format. Beware that the results that are not visualized will not be included in the report (Shiny apps work by making analyses only when their visualization is required).

You will find further assistance within each tab.

References

Moss, Richard H., Jae A. Edmonds, Kathy A. Hibbard, Martin R. Manning, Steven K. Rose, Detlef P. Van Vuuren, Timothy R. Carter, et al. 2010. “The next generation of scenarios for climate change research and assessment.” Nature 463 (7282). Nature Publishing Group: 747–56. doi:10.1038/nature08823.


Development

GCM compareR has been developed by Javier Fajardo, Derek Corcoran, Patrick Roehrdanz, Lee Hannah and Pablo Marquet in Marquet Lab in Pontificia Universidad Católica de Chile, in Santiago de Chile. It was built as part of the Spatial Planning for Protected Areas in Response to Climate Change initiative (SPARC) project, a GEF initiative leaded by Conservation International (CI), and with the support of Instituto de Ecología y Biodiversidad (IEB) in Chile.


Authors

Javier Fajardo
Derek Corcoran
Patrick Roehrdanz
Lee Hannah
Pablo Marquet


Climatic data

This application uses downscaled climate data published by the Research Program on Climate Change, Agriculture and Food Security (CCAFS). All the raster data used by GCM compareR is available from their package (Chamberlain 2017).


References

Chamberlain, Scott. 2017. Ccafs: Client for ’Ccafs’ ’Gcm’ Data. https://CRAN.R-project.org/package=ccafs.


Released versions

Version 1.0.0 (Oct 18, 2018)

  • Release of GCM compareR v1.0.0

Version 0.9.2 (Sep 26, 2018)

  • Update in text in all panels from the Intro tab.

Version 0.9.1 (Aug 21, 2018)

  • Update in tab names.

Version 0.9 (Aug 21, 2018)

  • Corrected an error where result tabs were disabled after a second use of the ANALYZE! button.
  • Corrected an error that caused the report to include figures from old executions.
  • Short instruccion for rectangle selection added below the map in the study area tab.
  • Warnings added about overlap between rectangle selection and countries/biomes/ecoregions.
  • Tab added with information for GCMs (developing institutions, full name, website…).
  • Intro is selected as default tab when starting.

Version 0.8 (before Aug 21, 2018)

(This was the first version online shared between authors when sharing the manuscript draft)

SELECT A SCENARIO


Global Circulation Models (GCMs)

Climate Change Scenario

Type of Comparison

Note: Only some scatter plot results are available with this option (the differences of scaled values, in 'Variation among futures'). All maps will still be produced.


Study Area

Warning: invalid ranges may cause the app to crash. Do not enter xmax or ymax smaller than xmin and ymin (and viceversa). Latitude valid values range from -90 to 60, and longitude from -180 to 180.


Press button when ready
Press again after changes
Instructions for selecting on the map:
- Click on the square icon to drag a selection.
- To analyze only a section of the countries/biomes/ecoregions selected, draw a selection over them after entering their names on the left panel.
- Be careful when using rectangle selections and country/biome/ecoregion: if there is no overlap, the app will crash. The same will happen if your rectangle does not cover any land. Make a new selection by draging again if necessary.
This table include the details for all GCMs available in GCM compareR. Notice that not all GCM are available for all scenarios

SELECTED OPTIONS:

Selected GCMs

These maps show all selected GCMs for a climate change scenario with a common variable and color scale. The first two maps for each bioclimatic layer are the baseline scenario (current climatic conditions) and the ensemble of the mean values among all GCMs.

Visualization options


SELECTED OPTIONS:

Variation from present

These results focus on differences between GCMs projections and the baseline (current climate). They can be used to identify which GCMs forecast larger or smaller changes in climate (e.g. units of increase in mean annual temperature) and to diagnose in which direction are produced those changes (e.g. some models might project an average reduction in annual precipitation and others an increase)


Scatterplot

To display the spread between GCMs, the averaged projected value for two selected bioclimatic variables is plotted for each GCM together with the average projection ensemble and the baseline. Axis values can indicate the projected values or the difference with current climatic condition values


Maps

These maps show the spatial pattern of changes projected by GCMs from current climate (BASELINE). The average ensemble projection across GCMs is also shown (ENSEMBLE)

Visualization options



Use the buttons on the upper-right corner of the figure to interact or zoom and download (a different version)






Use the buttons on the upper-right corner of the figure to interact or zoom and download (a different version)





Table with difference data

These results are not available when the comparison involves more than a pair a bioclimatic variables. For this comparison, values in variables need to be scaled before they are combined, and for this reason the only scatterplot produced is the one for scaled differences to the average ensemble. Find it in the tab 'Variation among futures' choosing scaled differences on the side panel.




SELECTED OPTIONS:

Variation among futures

Results displayed here focus on differences within the universe of selected GCMs. As a first step, an average ensemble climate projection is calculated by averaging the value in each downscaled GCM on a pixel basis. Then, the value in each model is compared to the ensemble, to determine whether its projection is greater or smaller than the average.


Scatterplot

The scatterplot is centered on the average ensemble, and each GCM is plotted using their mean difference to the ensemble values.

The circle indicates 2 standard deviations


Maps

These maps show the spatial distribution of each GCM deviations from the mean ensemble for each bioclimatic variable.


Visualization options


Use the buttons on the upper-right corner of the figure to interact or zoom and download (a different version)





Table with difference data

Use the buttons on the upper-right corner of the figure to interact or zoom and download (a different version)





Table with difference data

These results are not available when the comparison involves more than a pair a bioclimatic variables. For this comparison, values in variables need to be scaled before they are combined, and for this reason the only scatterplot produced is the one for scaled differences to the average ensemble. Find it in the tab 'Variation among futures' choosing scaled differences on the side panel.




Report of the session


In this tab, you can generate a report as a pdf file containing all the results produced during the session.

This report is intended at 3 things:

- Assure the repeatability of the analysis, as it will contain all the setting that you used to obtain the results

- Provide you with a full summary of all the results that were obtained during the session

- Complement the results with information relevant to their interpretation

Make sure to generate each one of the results in all the pages available in the app for them to appear in your report.


Generate report