Plots from CMIP3 / CMIP5 climate model R Shiny app

Here are some screenshots of plots comparing CMIP3 and CMIP5 climate models. The plots come from the web application I have been developing in R for comparing and evaluating SNAP’s downscaled CMIP3 and CMIP5 GCM data. The app is still under development and not yet released. The plots I’ve shared here focus exclusively on CMIP5 RCP 6.0. CMIP3 can be assumed absent unless specifically mentioned in a plot caption. When present in comparison to CMIP5, I used the SRES A1B scenario. The plots make use of the five models SNAP has previously evaluated as being most sensible for our purposes in Alaska and the Arctic.

I stuck with basic captions and cropped the original plot titles out of the top of the graphics because dynamically generated plot titling and subtitling in the app based on user inputs has not been fully coded yet. Although the app offers a number of R plotting styles, I have selected just a few here, and have retained use of the basic colorblind-friendly default color palette throughout.

First a few annual and decadal time series plots.

Temperature annual time series

Temperature annual time series

Temperature decadal time series

Temperature decadal time series

Temperature decadal time series

Temperature decadal time series

Next, several plots examining monthly (intra-annual) patterns.

Monthly temperature for three decades

Monthly temperature for three decades

2010 - 2039 monthly temperature

2010 – 2039 monthly temperature

2040 - 2069 monthly temperature

2040 – 2069 monthly temperature

2070 - 2099 monthly temperature

2070 – 2099 monthly temperature

2010 - 2099 monthly temperature, CMIP3 vs. CMIP5 all models

2010 – 2099 monthly temperature, CMIP3 vs. CMIP5 all models

2010 - 2099 monthly precipitation, CMIP3 vs. CMIP5 all models

2010 – 2099 monthly precipitation, CMIP3 vs. CMIP5 all models

On average, do the collection of models from one CMIP phase tend to be drier or wetter than the other group over the projected course of the 21st century?

2010 - 2099 CMIP3 vs. CMIP5 annual precipitation proportions

2010 – 2099 CMIP3 vs. CMIP5 annual precipitation proportions

Scatter plots comparing precipitation and temperature enable us to explore which models might be the most extreme combination of warm and dry, which can be informative for fire behavior modeling using SNAP’s ALFRESCO model. The GCM which is generally the warmest over time may also be relatively wet and not necessarily one that would drive high fire activity in ALFRESCO. We can examine which models fall disproportionately into a given quadrant, and whether this is projected to occur consistently across time or there is a trend in movement toward the extremes we move further into the future.

Precipitation ~ Temperature | CMIP phase + decade, using all ten models for May, June and July

Precipitation ~ Temperature | CMIP phase + decade, using all ten models for May, June and July

2010 - 2099 May Precipitation ~ Temperature | Model

2010 – 2099 May Precipitation ~ Temperature | Model

2010 - 2099 June Precipitation ~ Temperature | Model

2010 – 2099 June Precipitation ~ Temperature | Model

2010 - 2099 July Precipitation ~ Temperature | Model

2010 – 2099 July Precipitation ~ Temperature | Model

Boxplots are just one of several ways to use the app to explore the spread in climate values around some mean level or trend.

Boxplots with points overlay, showing variability in temperature among and between models over 2010 - 2099 for May, June and July

Boxplots with points overlay, showing variability in temperature among and between models over 2010 – 2099 for May, June and July

Boxplots with points overlay, showing variability in precipitation among and between models over 2010 - 2099 for May, June and July

Boxplots with points overlay, showing variability in precipitation among and between models over 2010 – 2099 for May, June and July

Development on the app continues. Bugs are being fixed. Features are being added, enhanced., and stabilized.

This entry was posted by Matt Leonawicz.

4 thoughts on “Plots from CMIP3 / CMIP5 climate model R Shiny app

  1. Stunning plots! Any chance to share the R code used to process the CMIPx data?

    Please keep the plots coming!

    • Thanks! There is code online, but you are not going to like it: http://leonawicz.github.io/SNAPQAQC/ See the “App code” drop down menu. Plot scripts are at the bottom.

      The issue I allude to is that this is by far my most complex app. The plot functions are highly abstracted to handle many arrangements of input data, as well as being highly specific to the data/context of the app. As the app documentation on the above site details, this app is not publicly available, still under development, and far from useful as a template for other apps. But code is available.

      Some code related to SNAP’s downscaling and other processing of CMIPx climate model outputs is also available online, but not everything is available at this time. Some at the site above. Some here: http://leonawicz.github.io/Downscaling/ Note that these pages are in a transient state, some incomplete, still under development.

      • Hi Matt. Sorry for the late response; I didn’t get any notification for you reply.

        Thank you a lot for pointing out the website with the scripts. Even though I am not using Shiny (just processing CMIP5 data in R instead), your lines of code will surely be a valuable source of inspiration.

        I totally understand your struggle. Dealing with this amount of data (i.e. CMIP5) is not a trivial challenge, and extracting meaningful information from them is even more challenging.

        Congratulations again for you great job and keep the graphics coming!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: