A recent op-ed  in the Wall Street Journal dredges up debunked conclusions drawn from a cherry-picked set of temperature measurements to try to call into question the reality and potential severity of climate change.
In a nutshell, authors Richard McNider and John Christy argue that warming in the upper atmosphere since 1979 is less than models had predicted and, therefore, models can’t be trusted and climate change shouldn’t be a concern.
In fact, virtually all climate data and research show that the Earth is warming. And it will continue to do so if we keep pumping greenhouse gases into the atmosphere. And this warming will bring an increased risk of more frequent and intense heat waves, higher sea levels, and more severe droughts, wildfires, and downpours.
To get at the facts, we can draw on recent climate assessments, including the State of the Climate report  compiled by National Oceanic and Atmospheric Administration (NOAA), the Intergovernmental Panel on Climate Change (IPCC ) Working Group I report, and the National Research Council’s America’s Climate Choices , plus other recent research (Thorne et al., 2011 , Santer et al., 2013 ).
Based on this research, here are three things to keep in mind:
1) The Earth is warming.
The op-ed authors focus on upper air measurements alone, arguing these are the best indicator for climate change. But this narrow view omits a wealth of valid observational data. We know that the surface over the land and ocean has been warming, the lower atmosphere has been warming, and the heat stored in the ocean has increased (see Figure 1). The upper atmosphere data shown by McNider and Christy include a portion of the stratosphere  (a separate layer of the atmosphere above the troposphere) that is cooling in response to the accumulation of greenhouse gases. To put it simply, greenhouse gases trap heat, especially in the low-level troposphere. This means less heat from the Earth reaches the stratosphere, resulting in cooling at this higher layer. By including a portion of this layer, McNider and Christy are averaging two different climate-change-induced effects to conclude that nothing at all is happening.
2) Models are successful at simulating many important aspects of the climate.
Models get the “big picture” of climate correct (i.e., the ups and downs of global temperature over many years, as well as the climate’s response to big events, like volcanic eruptions). This applies to both the surface temperature (see Figure 2) and the upper atmosphere temperature (Thorne et al., 2011 ).
Although McNider and Christy question the extent to which greenhouse gases contribute to warming, the pattern of warming due to greenhouse gases found in models bears a strong resemblance to the warming observed in nature (see the “fingerprinting” work by Santer et al., 2013 ).
3) Climate change is a risk management issue, not a binary choice between action and inaction.
In many public policy areas, we make decisions about the future in the face of incomplete or imperfect knowledge. For example, we make decisions about our military resources, even though we don’t know exactly where the next conflict might occur. Or we make decisions about infrastructure with limited knowledge of when, where, or how severe the next earthquake will be.
In the face of these other public policy issues, we use the best information at our disposal about future conditions to inform the decisions we must make today. Climate models (as a group) have projected warmer temperatures than has been experienced over the recent decade or so. But this discrepancy does not invalidate models’ utility in understanding how greenhouse gases can warm the planet, as confirmed by observations and fundamental scientific understanding (see Figure 2).
In the end, models should not be seen as magic black boxes that predict the future, but as illustrative tools for a range of potential futures. And what’s clear is that none of these futures looks like the past.
Aside from models, there are many other ways to identify risks, such as examining changes that have occurred to date and pinpointing critical vulnerabilities. Rising sea levels, increasing frequency and severity of heat waves, more intense precipitation events, and changes in ecosystems are climate change impacts now. Extreme weather events demonstrate the potential vulnerabilities of our communities. There are steps we can take to reduce our emissions and build resilience to help manage these risks, even if models are incomplete or imperfect.
The bottom line: The Earth has warmed during the 20th and early 21st centuries, with greenhouse gas emissions from human activities playing an important role. The ability of computer models to simulate some of the most prominent features of the climate system provides high confidence that the rising concentrations of atmospheric greenhouse gases will continue to warm the planet. In the absence of policies to curtail these emissions, the warming is likely to be substantial, bringing an increased risk of many unwanted impacts.
Figure 1. A subset of Essential Climate Variables from the 2012 State of the Climate report . The report is compiled by NOAA but includes contributions from hundreds of scientists. The two left panels show warming at the surface and in the lower troposphere from 1960 to 2012. The right panel shows the increase in the heat content of the upper layers of the ocean from 1993 to 2012. The different lines correspond to different data sets depicting these variables.
Figure 2. A comparison of observed and simulated global surface temperatures, from the latest IPCC  report (AR5, Working Group 1, Chapter 9). The thick black lines (solid and dashed) correspond to different data sets of observations. The thin colored lines correspond to climate model simulations. The thick red line represents the average of the climate models. The multi-model mean and the observations track each other relatively well on a decade-to-decade basis during the 20th century, and both exhibit similar responses to large volcanic eruptions (noted by green, dotted vertical lines). The differences between observations and models are larger on the year-to-year timescale, but the observations are generally within the “envelope” represented by the numerous models.