MOTIVATION.
The way in which the climate changes in response to increases in anthropogenic greenhouse gases is one of the foremost questions for the scientific community, policymakers, and the general public. A key approach for examining climate, especially how it will change in the future, uses complex computer models that include atmosphere, ocean, sea ice, and land components. Some models also simulate additional facets of the Earth system, including marine chemistry and biology. Model simulations indicate that temperatures have warmed over the past century, and will continue to rise into the future due to greenhouse gas forcing (IPCC 2014). However, the very large number of model simulations, the sheer volume of data they have generated, and output that might not be directly relevant for many applications can make it extremely difficult for potential users to access, view, and evaluate the data.
While useful web tools exist for viewing model-simulated climate change, including the “Climate Reanalyzer,” “Climate Wizard,” “National Climate Change Viewer,” “KNMI Climate Change Atlas,” and “Climate Variability and Diagnostics Package,” the Climate Change Web Portal offers some unique capabilities, including examination of model bias, intermodel variability, changes in variance, and physical and biogeochemical model output from the ocean.
The Climate Change Web Portal (www.esrl.noaa.gov/psd/ipcc/) was developed by the NOAA/ESRL Physical Sciences Division to access and display the large volumes of climate and Earth system model output from the Coupled Model Inter-comparison Project Version 5 (CMIP5; Taylor et al. 2012, van Vuuren et al. 2011) that informed the recently released Intergovernmental Panel on Climate Change (IPCC) report. The portal has two components that encompass 1) land and rivers or 2) oceans and marine ecosystems. Recent changes in federal agency directives and programmatic mandates require federal managers to consider climate change in water resources and environmental planning. As a result, resource managers are now required to make judgments regarding which aspects of climate projection information are applicable to a given decision, including decisions to modify system operations, invest in new or improved infrastructure, and establish long-term management objectives. The web portal provides scientists, resource managers, and stakeholders with a framework to evaluate and interpret the models by comparing them to observations (land/rivers portion) during the historic record and view how they project climate change in the future. To this end, federal water and fisheries managers have already used this tool in decision-making processes. The goal of this manuscript is to introduce the reader to the capabilities of the web portal.
METHODS AND EXAMPLES.
By preprocessing the model output and utilizing a number of software tools, the web portal allows users to quickly display maps and time series via a series of menu options. As a first step, output from the CMIP5 models, which have different horizontal resolutions, are interpolated to a 1° lat-lon grid to allow for intermodel comparisons. Statistics for different climate metrics are then computed on the common grid. A combination of software languages, including Javascript, Python, and NCAR’s Command Language (NCL), are used to access the NetCDF files to generate an image in real time. From the portal, a set of menus allows the user to choose 1) an individual model or the model ensemble mean; 2) an experiment (i.e., past or future greenhouse gas forcing); 3) fields to display, such as precipitation and ocean temperature at 100-m depth; 4) statistics, such as the mean, median, and 90 percentile (%), for the land component and standardized anomalies for the ocean component; 5) annual mean or three-month seasons; 6) time periods in the twentieth and twenty-first centuries; and 7) a predefined or user-defined region. Once the menu choices are selected, either four maps or two time series are displayed.
We illustrate the features of the system via examples of the land/river and ocean components of the portal. The first example (Fig. 1) shows the 90th percentile of the surface air temperature (SAT; °C) during June-July-August (JJA) for the years 1911–2005 (the SAT of the 10th warmest summer in each grid square over the 95-year period) over North America from 1) observations (University of Delaware Terrestrial Air Temperature); and 2) the ensemble mean of the CMIP5 models; 3) the difference between the two, indicating the model bias; and 4) the difference between the 90% SAT in the RCP 8.5 experiment during the twenty-first century (2006–2100) minus the values in the historical period (1911–2005), indicating the climate change signal. The ensemble model mean generally matches the observed pattern of very warm summer seasons, where the 90% SAT exceeds 25°C over the southwest United States and the southern Great Plains, with values less than 20°C over the Rocky Mountains and northwest United States. However, on average the models are too warm by approximately 0.5°–2°C over most of the Great Plains but slightly cooler than observed over the southeast United States. The bias has a complex pattern over Mexico and the western United States due in part to the smoothed representation of mountains in climate models. SAT extremes in JJA are more likely over the entire domain in the twenty-first century relative to the twentieth, especially away from the coasts, where the change in the 90% value exceeds 5°C between 35°N and 55°N.
Snapshot from the Land and Rivers section of the Climate Change Web Portal depicting the 90th percentile of JJA seasonal mean near-surface air temperature (SAT, °C) for the years 1911–2005 from (upper left) observations (University of Delaware Terrestrial Air Temperature); and (upper right) the ensemble mean of the CMIP 5 models; (lower left) the difference between the two, indicating the model bias; and (lower right) the difference between the 90% SAT in the RCP 8.5 experiment during the twenty-first century (2006–2100) minus the values in the historical period (1911–2005).
Citation: Bulletin of the American Meteorological Society 97, 4; 10.1175/BAMS-D-15-00035.1
The web portal can also be used to examine time varying changes. For example, the 30-year running mean of observed and simulated precipitation (mm) over the entire year for the New England watershed or Hydrologic Unit Code (HUC, a hierarchical representation of river basins) is presented in Fig. 2. In general, the models simulate more precipitation over New England during the twentieth century than observed (GPCC version 5), although the observed values are within the full range of the CMIP5 models (left panel). The right panel shows the observed and simulated precipitation values with their respective means over the 1901–2005 period removed (“anomaly”). Both observations and the models indicate an increase in precipitation for New England. While the spread in the precipitation increases among the models toward the end of the twenty-first century, all model simulations indicate an increase in precipitation by 2100. Enhanced precipitation, which is especially prominent in winter (not shown), could lead to increased flooding when the snow melts in late winter/early spring.
30-year running mean precipitation time series for area average precipitation (mm year−1) in the New England watershed (HUC) for (left) mean values and (right) anomaly values obtained by removing the 1901–2005 climatology from both the observations and the individual model simulations. GPCC observations are in black, the CMIP5 ensemble mean is in red, and gray shading represents the entire CMIP5 model range (light gray), the 10th–90th percentile range (darker gray), and the 25th–75th percentile range (darkest gray).
Citation: Bulletin of the American Meteorological Society 97, 4; 10.1175/BAMS-D-15-00035.1
Due to the absence of adequate observations for some ocean fields, the plots for the ocean component of the web portal are based solely on the climate model output. The annual and ensemble mean 0–700-m heat content (J m−2) in the North Pacific Ocean is shown in Fig. 3, including 1) the mean during the historical period (1956–2005); 2) the mean climate change signal given by the heat content in 2006–2055 minus 1956–2005; 3) the year-to-year variability as indicated by the standard deviation during the historical period; and 4) the ratio of the interannual variance in the future relative to the historical period. The mean heat content is relatively high in the subtropics and low in high latitudes with a tight gradient in between at 40°N, especially in the western side of the basin. The heat content is indicative of the wind-driven, upper-ocean circulation with subtropical and subpolar gyres and the Kuroshio/Oyashio Extension current along the tight gradient between them. The latter is a region of enhanced interannual variability relative to the rest of the North Pacific Ocean. The difference between periods indicates that the heat content of the entire North Pacific increases in the first half of the twenty-first century. However, the increase is not uniform, but is concentrated along 40°N in the western Pacific, suggesting either a northward shift of the Kuroshio/Oyashio current extension and/or an increase in the surface heat flux into the ocean or an increase convergence of heat near the front (Wu et al. 2012). Finally, the interannual heat content variability decreases during 2006–2055 relative to 1956–2005 over most of the North Pacific except at 45°N, just north of the front during the twentieth century.
Snapshot from the Ocean and Marine Ecosystems section of the Climate Change Web Portal depicting the CMIP5 ensemble mean ocean heat content integrated over the top 700 m (J m−2) for (upper left) mean during the historical period (1956–2005); (upper right) mean climate change signal from the RCP8.5 scenarios: 2006–2055 minus the 1956–2005 period in the historical experiments; (lower left) year-to-year variability as indicated by the standard deviation during the historical period; and (lower right) ratio of the interannual variance in the future relative to the historical period; presented as ratio rather than the difference of the variances, as the former is used to test for significance via the F-test.
Citation: Bulletin of the American Meteorological Society 97, 4; 10.1175/BAMS-D-15-00035.1
Annual average sea surface salinity (SSS) fields over the North Atlantic as simulated by NCAR’s Community Climate System model, version 4 (CCSM4; Gent et al. 2011) are shown in Fig. 4. The climatological mean SSS during 1956–2005 exhibits a maximum (>36 psu) in the subtropics and the Mediterranean, with higher values in the western Atlantic and minimum values (<33 psu) over most of the Arctic Ocean. The CCSM4 indicates that SSS will increase in the subtropics and decrease north of 40°N in the twenty-first century relative to the twentieth. The standard deviation of SSS is maximized in the northwest Atlantic near 40°N, at the boundary between the salty subtropical and relatively fresh subpolar gyres, and in the vicinity of the sea ice edge that extends from north of Iceland northeastward to Svalbard. The ratio of the twenty-first century to twentieth century SSS standard deviation is greater than 1 over most of the Atlantic north of 30°N, suggesting that salinity variability will increase over much of the North Atlantic in the future, especially between Iceland and Great Britain.
Snapshot from the “Ocean and Marine Ecosystems” section of the Climate Change Web Portal depicting the CMIP5 ensemble mean sea surface salinity (PSU) for (upper left) mean during the historical period (1956–2005); (upper right) mean climate change signal from the RCP8.5 scenarios: 2050–2099 minus the 1956–2005 period in the historical experiments; (lower left) year-to-year variability as indicated by the standard deviation during the historical period; and (lower right) ratio of the interannual variance in the future relative to the historical period.
Citation: Bulletin of the American Meteorological Society 97, 4; 10.1175/BAMS-D-15-00035.1
Earth system models in the CMIP5 archive simulate aspects of the biogeochemistry in the ocean, including primary production by phytoplankton that grow via the uptake of carbon and other inorganic molecules using energy provided by sunlight. Generally, marine ecosystem models simulate several classes of phytoplankton, although the number of kinds that are represented differ between models. The annual average primary production from all phytoplankton classes over the upper 150 m is shown for the Arctic and subpolar oceans (>50°N) in Fig. 5. In the historical climate (1956–2005), the North Atlantic, North Pacific, and Bering Sea are very productive, while the central Arctic is not. Several factors influence primary productivity, including light and temperature, which are limiting at high latitudes, and nutrients, which limit phytoplankton growth in midlatitudes and the Tropics. The primary productivity during the historical period indicates that conditions are conducive for phytoplankton growth during spring through fall in subpolar regions but ice cover, cold temperatures, and long periods without sunlight limit the annual production in the central Arctic and on both sides of Greenland. Productivity is enhanced north of Europe where warm water from the Atlantic enters the Arctic Ocean. The climate change signal (2050–2099 minus 1956–2005) exhibits reduced primary productivity over the North Atlantic and Gulf of Alaska and increased productivity in the Arctic, the Sea of Okhotsk, and most of the Bering Sea. The largest increase in productivity in the Arctic coincides with the largest decrease in sea ice (not shown), which enables more light to reach the ocean, allowing for more photosynthesis. The decrease in productivity in the North Atlantic and Gulf of Alaska may result from an increase in stratification, due to a freshening and warming near the surface (see Capotondi et al. 2012), which reduces the amount of nutrients mixed into the upper ocean from the deeper ocean.
Snapshot from the “Ocean and Marine Ecosystems” section of the Climate Change Web Portal depicting the CMIP5 ensemble mean net primary productivity of carbon by phytoplankton in the top 150 m (1 e−9 mol m−2 s−1) for (upper left) mean during the historical period (1956–2005); (upper right) mean climate change signal from the RCP8.5 scenarios: 2050–2099 minus the 1956–2005 period in the historical experiments; (lower left) year-to-year variability as indicated by the standard deviation during the historical period; and (lower right) ratio of the interannual variance in the future relative to the historical period.
Citation: Bulletin of the American Meteorological Society 97, 4; 10.1175/BAMS-D-15-00035.1
SUMMARY.
While the Climate Change Web Portal was initially designed for hydrologic and fishery applications, we anticipate that it will be useful to a wide range of users. To that end, we have included additional information, including tutorials and metadata accessible through help links on the portal. In addition, the derived fields used to make the plots can be downloaded as a netCDF file, so users can use their own software package to create plots. The portal is designed so that more variables, experiments, statistics, and features can be added in the future. Currently, there are some capabilities—such as comparing models side-by-side, comparing ocean model output with observations, and comparing the variability of the climate change signal among all the models—that are not possible. We plan to add these features and enhance web portal tutorials in the future. We feel that this tool provides a useful framework for users to assess current and future changes in CMIP5 climate simulations. More details on climate modeling, the IPCC report, the CMIP5 experiments, and observational datasets can be at www.esrl.noaa.gov/psd/ipcc/references.htm.
ACKNOWLEDGMENTS
We thank Levi Brekke, Jeff Arnold, and Roger Griffis for the encouragement and for providing financial support through the Bureau of Reclamation, Army Corps of Engineers, and NOAA’s National Marine Fishery Service through the acknowledge (SMECC) program, respectively.
FOR FURTHER READING
Capotondi, A., M. Alexander, N. Bond, E. Curchitser, and J. Scott, 2012: Enhanced upper-ocean stratification with climate change in the CMIP3 models. J. Geophs Res.-Oceans, 117, C04031, doi:10.1029/2011JC007409.
Gent, P. R., and Coauthors, 2011: The Community Climate System Model version 4. J. Climate, 24, 4973–4991, doi:10.1175/2011JCLI4083.1.
IPCC, 2014: Summary for Policymakers. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, C. B. Field et al., eds. Cambridge University Press, 1–32.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An Overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, doi:10.1175/BAMS-D-11-00094.1.
van Vuuren, D. P., and Coauthors, 2011: The representative concentration pathways: An overview. Climate Change, 109, 5–31, doi:10.1007/s10584-011-0148-z.
Wu, L., and Coauthors, 2012: Enhanced warming over the global subtropical western boundary currents. Nat. Climate Change, 2, 161–166.
