Breaking News

New analysis shows Lamar Smith’s accusations on climate data are wrong

Scott K. Johnson

A research vessel drops off a float that will make measurements as it drifts.
When the National Oceanic and Atmospheric Administration (NOAA) updated its global temperature dataset to include revisions to the underlying databases, US House Science Committee Chair Lamar Smith (R-TX) somehow knew that the scientists had done it wrong. In fact, he accused them of having “altered temperature data to get politically correct results.”
How could he know that? Well, Rep. Smith rejects the conclusions of climate science—like the fact that the Earth’s climate is warming. Despite the slight up and down wiggles of the estimated annual temperatures, the warming of the last couple decades increased ever so slightly in NOAA's new analysis. To Smith, the result was a red flag. Suddenly he wanted to see the researchers’ e-mails and echoed the accusations of contrarian blogs about scientists’ supposedly nefarious adjustments to sea surface temperature measurements.
The update to NOAA's dataset was really pretty subtle.
The update to NOAA's dataset was really pretty subtle.
Rather than invoking scientific conspiracies, issues like this should be settled by analyzing the data. A new study, led by University of California Berkeley’s Zeke Hausfather, does just that—and Rep. Smith won’t like these results, either.
Producing a dataset of global surface temperature is more complicated than calculating the average of a few numbers—especially when it comes to the temperature of the seawater that covers most of the planet’s surface. Since the oceans are sparsely monitored, researchers need all the measurements they can get their hands on, which include instruments on buoys and measurements made in a variety of ways on ships.
Not only do different measurement techniques have the potential to give slightly different temperature readings, but practices change over time. At one point, wooden buckets were thrown over the side and hauled up. Then it was canvas buckets, which allowed a little more evaporative cooling on the way up. Then it was mainly sensors at the engine’s coolant water intake. And in recent years, the number of scientific buoys and autonomous floats has exploded. Getting all the data collected over the years linked up on an apples-to-apples basis takes careful analysis.
NOAA had been relying on version 3b of a marine temperature dataset called ERSST. A 2015 paper described a version 4 update, which included important adjustments to ship-based measurements made since 1940. The adjustment corrected for the offset between ship-based and buoy-based measurements in more recent years.
In order to see how well these adjustments linked up different types of measurements over the last 20 years, Hausfather and his colleagues took advantage of some modern datasets with fewer complications. There have been enough buoy and satellite measurements to provide decent coverage back to 1997, and the fleet of autonomous floats in the Argo network can go back to about 2005. Taken separately, each dataset can be used as an independent test of NOAA’s effort to stitch together multiple types of measurements.
In addition, we can ask whether, say, satellite measurements confirm that version 4 of that ERSST dataset (which doesn’t include satellite measurements) really is more accurate than version 3b was.
The comparison has to be made carefully because each dataset includes measurements from different locations that don’t add up to perfect global coverage. So among other tests, the researchers broke the map into grid cells and used only those cells where there were measurements in both datasets being compared.
As it happens, the Argo float data, the buoy data, and the satellite data each hew closer to the updated dataset that NOAA used. The older version (3b) gives a global average that is too cool in recent years, growing to an offset of about 0.06 degrees Celsius. The researchers repeat this same analysis for two more major sea surface datasets that are used by the UK Met Office and the Japanese Meteorological Agency for their global temperature records. Both of those datasets also drift cooler than the comparison data, but less so than NOAA’s old dataset.
The difference between four major sea surface temperature datasets (versions 4 and 3b of NOAA's data, the UK Met Office's HADSST3, and the Japanese Meteorological Agency's COBE-SST) and buoy measurements (red) or satellite measurements (blue).
The difference between four major sea surface temperature datasets (versions 4 and 3b of NOAA's data, the UK Met Office's HADSST3, and the Japanese Meteorological Agency's COBE-SST) and buoy measurements (red) or satellite measurements (blue).
That cooling drift might be caused by a shift in the mix of ship-based measurement techniques. In addition to finding the average offset between buoy and ship measurements, the updated dataset NOAA used also weighted buoy measurements more strongly in its averaging, which would minimize any problems in the ship measurements.
While this study is certainly useful for researchers, it’s important to note that we’re actually talking about very small differences between global temperature datasets—too small to have real implications for our understanding of global warming. But it does confirm that NOAA and the scientists who revised the sea surface temperature dataset were just doing good science. The data don’t lie.
Open Access at Science Advances, 2016. DOI: 10.1126/sciadv.1601207  (About DOIs).

No comments