Worldwide climate modeling centers participating in the 5th Climate Model Intercomparison Program (CMIP5) are providing climate information for the ongoing Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC). The output from the CMIP5 models is typically provided on grids of ~1 to 3 degrees in latitude and longitude (roughly 80 to 230 km at 45° latitude). (The Global Climate Change (GCCV) viewer visualizes the global model data sets on a country-by-country basis.) To derive higher resolution data for regional climate change assessments, NASA has statistically downscaled maximum and minimum air temperature and precipitation from 33 of the CMIP5 models to produce the NEX-DCP30 data on a very fine 800-m grid (Figure 1) over the continental United States (Thrasher et al., AGU Eos Trans, Volume 94, Number 37, 2013, doi:10.1002/2013EO370002).
The full NEX-DCP30 dataset includes 33 climate models and their respective downscaled data for historical (1950-2005) and 21st century simulations under four Representative Concentration Pathways (RCP) emission scenarios developed for AR5. (Further details regarding the science behind developing and applying the RCPs are given by Moss et al., Nature, Volume 463, 2010, doi:10.1038/nature08823) Our application, the NEX-DCP30 Viewer, includes historical and future (2006-2099) climate for RCP4.5 (one of the possible trajectories for greenhouse gas (GHG) emissions in which atmospheric GHG concentrations continue to rise but are capped so as not to exceed a radiative equivalent of 4.5 Wm-2 in 2100) and RCP8.5 (the most aggressive emissions scenario in which GHGs continue to rise unchecked through the end of the century leading to an equivalent radiative forcing of 8.5 Wm-2). We include 30 of the 33 models in the viewer that have the RCP4.5 and RCP8.5 data; the remaining two scenarios, RCP2.6 and RCP6, are available in the NEX-DCP30 data set.
The NEX-DCP30 Viewer allows the user to visualize projected climate change for any county in the continental United States. To create a manageable number of permutations for the viewer, we have averaged the NEX-DCP30 data into 25-year climatologies that span the 21st century. The viewer provides a number of useful tools for characterizing climate change such as: climographs (plots of monthly averages), histograms that show the distribution or spread of the model simulations, monthly time series spanning 1950-2099, and tables that summarize changes in the quantiles of temperature and precipitation (e.g., extremes). The application also provides access to comprehensive, 3-page PDF summary reports and CSV files for the entire US, each state and each county. We do not provide access to the primary NEX-DCP30 data set. The data can be downloaded in NetCDF format from either the Earth System Grid Federation data portal (http://cmip-pcmdi.llnl.gov/cmip5/data_portal.html) or, in the near future, from the US Geological Survey’s Geo Data Portal (http://cida.usgs.gov/climate/gdp/).
Overview of the NEX-DCP30 Viewer
Understanding climate model data from many models in time and space is a challenge. We have attempted to design a viewer that strikes a balance between visualizing and summarizing climate information and the complexity of site navigation. Many of the features of the viewer can be discovered and learned by experimenting; however, for reference we provide the following tutorial that explains the details of the viewer.
The main window:
The main window of the NEX-DCP30 Viewer (Figure 2) displays maps of change in maximum temperature, minimum temperature or precipitation between a selected future simulation and the historical simulation and two accompanying plots. As indicated by the red arrow above the color bar, the changes shown are in the annual average maximum temperatures for the period 2050-2074 versus 1980-2004 in the RCP8.5 simulations. The maps provide a general impression of the spatial variability of climate change across the continental United States. Hovering the cursor over a state produces a popup window of the average value for that state. The climograph (Figure 2, lower left) compares monthly averages and standard deviations (vertical bars), which are a measure of variability, for the present and future periods. The histogram (Figure 2, lower right) displays the distribution of change for all model simulations included in particular experiment and is a quick way to visualize the spread of the simulated climate change anomalies over the selected geographic area.
Controls and map navigation:
The dropdown menus across the top of the application (outlined in red in Figure 3) are used to select either annual or monthly means, the average of all 30 model simulations (Mean Model) or a selected model (list of models) and variable of interest.
Clicking on a state zooms the map to that state and displays a map and charts similar to those of the US example, but with values for the selected state (heavy line in Figure 4). Likewise, clicking on a county displays the map and plots for that county (heavy line, Figure 4). The button menu at the top left of the map always displays the currently selected state/county, which is shown on the map in with a heavy bold outline. By clicking the "Continental United States" button, the application will zoom out back to the national level.
At the state level, changes in temperature do not vary greatly, especially Mean Model values that are averaged annually over 30 models. Changes in precipitation, however, can display substantial spatial variability, especially in mountainous areas such as Oregon (Figure 6) and for only one selected model and month. Figure 6 illustrates NEX-DCP30 downscaled changes in precipitation for December as simulated by version 4 of the Community Climate System Model which has been developed by the National Center for Atmospheric Research in Boulder, CO. The influence of land-sea contrasts along the coast and inland mountain ranges are clearly evident in the precipitation pattern. As indicated by the plot of monthly values in the lower left, averaged over Oregon, there is not a great deal of change in the future. This is further corroborated by the histogram to the right, which shows most of the CMIP5 models simulate little or no change relative to the average rainfall for December.
Climograph and Histogram tabs:
The climograph plot in the bottom left of the application (Figure 7) compares the climatology of the historical (1980-2004, blue line) and the future simulations (here, 2050-2074, red line) for Benton County, Oregon using the Mean Model of the RCP8.5 experiment. The vertical error bars indicate the standard deviation (which is a measure of variability) for each month over the respective time periods. In the example, the maximum temperature for 2050-2074 is consistently warmer in all months than that of the historical period, displays comparable monthly variability, and, because the error bars do not overlap, suggests that the changes are statistically significant. Hovering the cursor over a particular month (in this case, April) displays values for the mean and standard deviation of the historical and future simulation. The maximum temperature for April is simulated to warm by about 2°C in Benton County, Oregon in 2050-2074 under the RCP8.5 emission scenario. Clicking on the chart lines will changes the selected month displayed in the map above. No observed data are plotted because the NEX-DCP30 data has been bias corrected (see Section 3.2 Methods) or adjusted so that the historical simulation (blue line) is very similar to the observed values.
The histogram plot in the bottom right of the application (Figure 8) displays the distribution of climate change simulated by all the models for the selected geographic area, experiment and time period. Bins for temperature and precipitation change are indicated on the horizontal axis and the percent of the 30 models falling within each bin is indicated on the vertical axis. The histogram gives a sense of the range and distribution of climate change simulated by the models. Hovering the cursor over the histogram bars produces a window that summarizes the distribution and which models fall within each bin (Figure 9). Continuing with the example of Benton County, Oregon, the average change for all models is 3.4 °C and 30% (9/30) of the models simulate a warming of 3.0 to 3.5 °C in July maximum temperature. However, there are clearly a few outlying models that simulate a warming less than 1 °C and greater than 5.5 °C. Clicking on a histogram bar cycles through all the models in that bin and changes the map and climatology plot to the selected model.
Time Series tab:
The Time Series tab allows the user to visualize the long-term (1950-2099) trends of projected temperature and precipitation through the end of the 21st century. It is also a way to see how the RCP4.5 and RCP8.5 emission scenarios produce different (or similar) changes. The user can use the radio buttons located on the lower right of the window to plot either the actual temperatures or the changes in temperature. In the case of Benton County, Oregon actual July maximum temperature (Figure 10), the RCP4.5 and RCP8.5 scenarios the warming trends more or less track each other until around 2040 when they begin to diverge as a result of stabilizing GHGs in RCP4.5 simulations and continued increases in GHGs in the RCP8.5 simulations. The time series of relative change ( Figure 11) show about 2.4 °C difference between the RCPs at the end of the century. As in the other plots, in either time series plot hovering the mouse over the graphs will produce popup windows showing the time and values of the selected data points. Both ways of looking at the data can be useful: actual values can show, for example, what year temperature crosses some threshold (e.g., the freezing point in winter or maximum summer temperature related to growing degree days) and relative changes can show when projected temperatures have warmed by more than 2 °C relative to the 1980-2004 average. We note that, although the historical simulations (1950-2005) are bias corrected (see Section 3.2 Methods), an individual year is not expected to represent the actual year in these simulations.
Data Table tab:
The Data Table tab provides an easy way to explore the values of change in temperature and precipitation for a given model and selected geographic area. The column headers can be clicked on to sort into either ascending or descending order (Figure 12). The flags to the left of the model name indicate the global model’s country of origin. Sorting the models by the magnitude of change, for example, is a convenient way to explore the range and spatial pattern of climate change. Clicking on a row will select a model and display the change in the map above. In Figure 13, the ACCESS1-0 model has been selected and the 2050-2074 change in maximum temperature is mapped.
Percentile Table tab:
The Percentile Table sorts all the monthly data from 1980-2099 into 25-year bins and percentiles for the RCP4.5 and RCP8.5 emission scenarios (Figure 14). These tables provide a way to explore how climate change alters not just the average but also the extreme values across the scenarios. In the case of temperature shown in Figure 14, the 10th percentile represents the coldest temperatures, the 50th percentile represents median temperatures and the 90th percentile represents the warmest temperatures. In this example, the 10th percentile (coolest maximum temperatures) for Benton County, Oregon warm by 2.1 °C in 2075-2099 (relative to 1980-2004) in RCP4.5, whereas the 90th percentile (warmest maximum temperatures) change by 2.9 °C, indicating slightly more warming in maximum July temperature extremes. Greater warming in the extremes is evident in the RCP8.5 simulations in which the 10th percentile value changes by 3.5 °C and the 90th percentile value changes by 5.1 °C.
Model Info tab:
The Model Info tab displays the full name of the modeling center and country of origin for each global model that was downscaled in the NEX-DCP30 data set (Figure 15).
Scenario, Time Period, Units tab:
The Scenario and Time Period tab (Figure 16) allows the user to select either the RCP4.5 scenario or the RCP8.5 and the time period of interest:
- 2025-2049 versus 1980-2004
- 2050-2074 versus 1980-2004
- 2075-2099 versus 1980-2004
Download Summary tab:
The Download Summary tab provides access to both PDF summaries and the data used in the time series graphs. We have created 2.5 million three-page PDF summaries of all of the charts and tables included in the viewer tabs that can be downloaded using the Download Summary tab (Figure 17). The summary reports are available for each geographic area (US, states and counties), for every month and climate model in both metric and English units. All three variables (maximum temperature, minimum temperature and precipitation) for both RCP 4.5 and RCP 8.5 emission scenarios are included in the reports to facilitate comparisons between scenarios and among models. An example of a summary for Mean Model July maximum temperature for Benton County, Oregon is illustrated in Figure 18.
- Time series for the annual average or a given month (July in this example) are plotted from 1950-2099 in both absolute values (left axis) and changes relative to the 1980-2004 average (right axis).
- The quantile table included in the summary PDF report expands on that of the Percentile Table tab by including the 1st, 5th , 95th and 99th percentiles. These percentiles represent the most extreme values within the 25-year monthly time series.
- The climographs shown in the summary PDF include plots for both RCP4.5 and RCP8.5 and climatologies and standard deviations for each 25-year time period.
- Histograms for both RCP4.5 and RCP8.5 and all three time periods are displayed in the summary report PDFs. These enhanced plots help visualize how the distribution of climate change anomalies change through time and across both emission scenarios. In the case of July max temperature for Benton County, Oregon the distribution of anomalies in RCP4.5 is tightly grouped around ~3 °C in the 2025-2049 period but widens to a larger range by the end of the century. The RCP8.5 distribution is both wider and exhibits stronger warming in 2050-2074 and 2075-2099 2075-2099 than the RCP4.5 scenario. The bin that contains the selected model is displayed in red, which helps indicate where the selected model is within the overall distribution of models.
The monthly average data used in the 1950-2099 time series plots for the selected geographic area, model and averaging period are available for users wishing to do additional analyses and exploration. Clicking on the Download Time Series buttons (Figure 18) will download files in comma separated variable (CSV) format that can be opened in spreadsheet or other programs (Figure 19). Metadata is included to describe the file contents and the monthly temperature and precipitation values for the two scenarios are registered in time by the model year and month. Note that the data are the raw averages and not the differences between the scenarios and the historical period.
Citation to the web application:
Alder, J. R. and S. W. Hostetler 2013. NEX-DCP30 Climate Downscaling Viewer. US Geological Survey http://www.usgs.gov/climate_landuse/clu_rd/nex-dcp30.asp doi:10.5066/F7W9575T