Tony Heller On US Temperatures

Tony Heller made the following mashup of two graphs intended to show that NASA lies about CONUS temperatures as part of a criminal conspiracy to fabricate global warming. It will be very easy to show that Tony Heller is the one lying. 

It took about 5 minutes for me to document why the bottom graph is an improvement over the top graph, and there's nothing "inexplicable" about it. The information needed to explain the changes comes a page on NASA's website that documents the developments and improvements to the GHCN station dataset over time. NASA isn't hiding anything. A subset of this network was used to form the USHCN dataset that was used by NASA and NOAA until 2014.  Here are some relevant updates introduced since the top 1999 graph:

1. In 1999, USHCN was expanded from 138 to 1062 stations.[1] And NASA's methodology improved to account for station moves that corrected for unrealistic outliers. Areas with missing data were filled in, meaning that the area-weight of nearby stations was expanded for better coverage of the US.[2]
2. In 2000, bias correction for urban heat islands was introduced.[3] 
3. In 2009, improvements to the time of observation bias correction were introduced with UHCN v2.[4]. USHCN was expanded again to 1218 stations.
4. In 2010, Further improvements for UHI adjustments using night light radiance were introduced.[5]
5. In 2012, USHCN was improved to USHCN v2.5.[6]
6. In 2014, USHCN was deprecated in favor of nClimDiv (which increased its station count to about 10,000 stations), though USHCN data is still stored on NOAA's ftp site.

So compared to the bottom graph in 2017, the top graph in 1999 had 138 stations and didn't account for station moves and other forms of bias such as station siting, urbanization and time of observation. In other words, the bottom graph is a better representation of changes in CONUS temperatures, and the data has higher spatial resolution than the top graph. In fact, the raw USHCN and bias-corrected USHCN data are both publicly available, and anyone can download the data and plot the differences. For the following graph I used USHCN  raw and final data compiled by Jim Java on X. The net effect of bias correction in CONUS temperatures does increase CONUS warming trends by a small amount, but we're still "hotter than the dust bowl" now.

Heller believes that because his bottom graph is different from his top graph in a way he doesn't like, NASA must be engaging is some sort of evil criminal plot to fabricate global warming by fabricating data. But we can actually test if this is the case. In 2005, NOAA introduced their USCRN network, and this network contains 114 ideally-located stations in only rural areas that do not require bias correction. USHCN was deprecated in 2014 and replaced by nClimDiv with about 10,000 stations, and this dataset uses bias correction and homogenization procedures identical to USHCN v2.5. We have 20 years of data from both networks that allow us to compare the two. It turns out that the pristine and raw USCRN dataset shows marginally more warming than the homogenized nClimDiv.

Reanalysis datasets are unaffected by many of the biases that are corrected for in homogenized datasets, but the trends for nClimDiv and ERA5 are nearly identical.

The following graph shows how the trends in several CONUS datasets compare with the raw USCRN. The raw USCRN dataset has the greatest warming trend of the four, though the CIs are large enough that they are not significantly different from each other. So the best raw data we have from 2005 on show that bias correction in nClimDiv is not adding any spurious warming relative to the USCRN raw data.

The following chart shows significant acceleration in trends as well.

	Trends thru June 2026 and 2σ Uncertainty
CONUS Dataset	2005-2026	1940-2026	1895-2026
USCRN	0.5 ± 0.2°C/decade
ERA5	0.4 ± 0.2°C/decade	0.18 ± 0.02°C/decade
nClimDiv	0.5 ± 0.2°C/decade	0.17 ± 0.03°C/decade	0.10 ± 0.02°C/decade

NOAA still stores station data for the old, deprecated USHCN dataset on their ftp site, even though NOAA no longer publishes it for public use, since the network is no longer maintained and many stations that were in the network have been closed. But the raw data that Heller is using is still there, and so it's possible to plot both the raw USHCN data and compare it to the current, homogenized nClimDiv. This allows us to see the effect that NOAA's adjustments (and additional stations) has had on the raw data. Bias correction has reduced CONUS temperatures by an average of 0.37°C in the first 30 years of the dataset, 1895-1924 when both raw and homogenized data are set to a 1991-2020 baseline. However, there's also a clear warming trend in the raw data, and current temperatures are warmer than the dust bowl years.

Heller has to focus on CONUS temperatures to accuse NASA and NOAA of criminal fraud to fabricate global warming. If Heller had used GMST data, he wouldn't be able to make his case. The effect of bias correction on global mean surface temperatures goes in the opposite direction to Heller's conspiracy theory. Bias correction to SSTs significantly increases temperatures prior to 1940, and adjustments to SSTs are greater than adjustments to land temperatures, so the net effect of all bias correction procedures on thermometer datasets is to decrease the amount of global warming. In fact, the average difference between NOAA's raw and bias corrected GMST anomalies for 1850-1900 is 0.12°C. At the end of 2025, the globe was about 1.38°C above the 1850-1900 mean, so if we used raw data instead of bias corrected data, we would have already reached 1.5°C warming. The raw data shows more global warming than the bias corrected data that Tony Heller is complaining about.

Raw Data from ClimateBrink

If NASA is trying to fabricate global warming by altering CONUS temperatures, they're doing a very poor job, since the sum total of all bias correction procedures lowers the amount of global warming. This is the nail in the coffin of Tony Heller's conspiracy theory. Raw temperature data shows more warming than the bias corrected and homogenized data that Heller is complaining about.

References:

[1] Easterling, D. R., T. R. Karl, J. H. Lawrimore, and S. A. Del Greco. (1999). United States Historical Climatology Network Daily Temperature, Precipitation, and Snow Data for 1871–1997. ORNL/CDIAC-118, NDP-070.

[2] Hansen, J., R. Ruedy, J. Glascoe, and M. Sato, 1999: GISS analysis of surface temperature change. J. Geophys. Res., 104, 30997-31022, doi:10.1029/1999JD900835.

[3] Hansen, J.E., R. Ruedy, M. Sato, M. Imhoff, W. Lawrence, D. Easterling, T. Peterson, and T. Karl, 2001: A closer look at United States and global surface temperature change. J. Geophys. Res., 106, 23947-23963, doi:10.1029/2001JD000354.

[4] Menne, M. J., C. N. Williams, and R. S. Vose, 2009: The United States Historical Climatology Network monthly temperature data, Version 2. Bulletin of the American Meteorological Society, 90 (7), 993–1007.

[5] Hansen, J., R. Ruedy, M. Sato, and K. Lo, 2010: Global surface temperature change. Rev. Geophys., 48, RG4004, doi:10.1029/2010RG000345.

[6] Menne, Matthew J.; Williams, Claude N. Jr. (2012): United States Historical Climatology Network (USHCN), Version 2.5.x. [Indicate minor version and date of access]. NOAA National Centers for Environmental Information. doi: 10.7289/V56W98B4