Guest essay by Eric Worrall
Climate scientists finally seem to be learning from the acute embarrassment of colleagues who rashly trumpet the predictions of their climate models, as if they mean something.
Northeast warming more rapidly than most of US
New England is likely to experience significantly greater warming over the next decade, and beyond, than the rest of the planet, according to new findings by climate scientists at the University of Massachusetts Amherst.
The region’s temperatures are projected to rise by an average of 3.6 degrees Fahrenheit above pre-industrial levels by 2025, according to the study, published this week in PLOS One, a journal published by the Public Library of Science.
The scientists found that the Northeast is warming more rapidly than any other part of the country except Alaska — and that the 3.6 degree Fahrenheit rise in the region is likely to come two decades before the rest of the world gets to that point.
“I tell my students that they’re going to be able to tell their children, ‘I remember when it used to snow in Boston,’” said Ray Bradley, an author of the study and director of the Climate System Research Center at the University of Massachusetts. “We’ll have occasional snow, but we won’t have weeks and weeks of snow on the ground.”
The press release may seem assertive, but delve into the referenced study, and you find a little more caution about the claimed reliability of the predictions. That abstract starts reasonably confidently;
Consequences of Global Warming of 1.5 °C and 2 °C for Regional Temperature and Precipitation Changes in the Contiguous United States
Ambarish V. Karmalkar, Raymond S. Bradley
Published: January 11, 2017
The differential warming of land and ocean leads to many continental regions in the Northern Hemisphere warming at rates higher than the global mean temperature. Adaptation and conservation efforts will, therefore, benefit from understanding regional consequences of limiting the global mean temperature increase to well below 2°C above pre-industrial levels, a limit agreed upon at the United Nations Climate Summit in Paris in December 2015. Here, we analyze climate model simulations from the Coupled Model Intercomparison Project Phase 5 (CMIP5) to determine the timing and magnitude of regional temperature and precipitation changes across the contiguous United States (US) for global warming of 1.5 and 2°C and highlight consensus and uncertainties in model projections and their implications for making decisions. The regional warming rates differ considerably across the contiguous US, but all regions are projected to reach 2°C about 10-20 years before the global mean temperature. Although there is uncertainty in the timing of exactly when the 1.5 and 2°C thresholds will be crossed regionally, over 80% of the models project at least 2°C warming by 2050 for all regions for the high emissions scenario. This threshold-based approach also highlights regional variations in the rate of warming across the US. The fastest warming region in the contiguous US is the Northeast, which is projected to warm by 3°C when global warming reaches 2°C. The signal-to-noise ratio calculations indicate that the regional warming estimates remain outside the envelope of uncertainty throughout the twenty-first century, making them potentially useful to planners. The regional precipitation projections for global warming of 1.5°C and 2°C are uncertain, but the eastern US is projected to experience wetter winters and the Great Plains and the Northwest US are projected to experience drier summers in the future. The impact of different scenarios on regional precipitation projections is negligible throughout the twenty-first century compared to uncertainties associated with internal variability and model diversity.
The content of the full study seems much less certain, with lots of qualification of the confident sounding assertions in the abstract and the press release.
… Under RCP8.5, all models indicate 1.5°C warming over CONUS before 2040 and 2°C warming before 2060, but the TCTs for RCP4.5 are distributed throughout the century. Notably, the ensemble mean temperature projections over CONUS reach 2°C warming by early to mid 2030s and 1.5°C warming by early 2020s despite large differences in TCTs for individual models. The observed warming over CONUS by 2014 relative to the baseline was roughly 1.0°C but a number of models suggest it to be over 1.5°C. Such disagreements could result from the mismatch in observed and simulated variability on decadal timescales
The lower bounds of temperature projections in Fig 3 typically represent how low climate sensitivity models respond to RCP4.5 whereas the higher bounds are determined by the responses of high climate sensitivity models to RCP8.5. Since the low sensitivity models indicate very little warming globally as well as regionally throughout the century, the lower bounds of projections are comparable across regions.
The Northeast is projected to cross the 2°C threshold about 15 years earlier than the South and about two decades before GMAT. For instance, about 80% of the projections indicate 2°C global warming by 2060 whereas the same percentage of projections cross the 2°C threshold in the Northeast by 2040.
All regions in the contiguous United States are projected to cross the 2°C warming threshold about 10-20 years earlier than the global mean annual temperature. While there is a large spread in TCTs across all regions, 75% to 90% of the models reach 2°C warming by 2050 for every region in the US. We believe that our estimates of TCTs based on 5-year means of annual mean temperatures may be conservative since we require that all the subsequent 5-year means beyond the identified threshold crossing time exceed the selected temperature threshold. The unpredictable nature of internal climate variability could advance or delay TCTs by a few years to a couple of decades regionally as demonstrated using the initials conditions ensemble. But this uncertainty in regional TCTs is smaller than the spread arising from using different models and two different scenarios. The consequence of large climate variability at regional scales, however, suggests that it may prove difficult to distinguish between the consequences of global warming of 1.5°C and 2°C for regional changes.
A cooling trend in the southeast US, the so called “warming hole”, in the second half of the twentieth century was not captured by CMIP5 models.
On the other hand, the eastern US is projected to reach the 2°C target in the 2020s regardless of the scenarios.
The CMIP5 multi-model ensemble used in this study samples the structural diversity in model formulation, but was not designed for systematic exploration of uncertainties, and therefore may not span the full range of outcomes in climate projections . Additionally, the use of one realization for every model is inadequate to capture the effect of internal variability that plays a significant role in driving regional changes from years to decades.
If future generations of models reduce contributions from internal variability and model uncertainty substantially, studying precipitation response to different scenarios may become important. This, however, is unlikely given that the internal climate variations over CONUS remain large and highly unpredictable over the next 20-50 years.
Read more: Same link as above
Climate science – where any observation is acceptable, any uncertainty can be accommodated, and no theory is falsifiable.