What Does NOAA’s Warmest Year Tell Us About Climate Sensitivity to CO2?

A friend of mine who works for the EPA emailed me a link to NASA’s Earth Observatory page pitching 2014 as the warmest year on record, and asked if  “I dismiss their findings.” The following is an edited version of my reply suggesting the Global Average Chimera tells us precious little about the climate’s sensitivity to CO2, and the uncertainty is far greater than the error bars illustrated in Anthony Watts post 2014: The Most Dishonest Year on Record.

I simply asked my friend to consider all the factors involved in Gavin Schmidt’s making of the global average temperature trend, and ask you all to do the same. Then decide for yourselves the scientific value of the graph and if there was any political motivation.

1. Consider the greatest warmth anomalies are over the Arctic Ocean because more heat is ventilating through thinner ice. Before the Arctic Oscillation removed thick insulating sea ice, air temperatures were declining. Read Kahl, J., et al., (1993) Absence of evidence for greenhouse warming over the Arctic Ocean in the past 40 years. Nature 361, 335 – 337.

 

NOAA's 2014 Temperature Anomalies

  

After subfreezing winds removed thick ice, then air temperatures rose. Read Rigor, I.G., J.M. Wallace, and R.L. Colony (2002), Response of Sea Ice to the Arctic Oscillation, J. Climate, v. 15, no. 18, pp. 2648 – 2668. They concluded, “it can be inferred that at least part of the warming that has been observed is due to the heat released during the increased production of new ice, and the increased flux of heat to the atmosphere through the larger area of thin ice.”

CO2 advocates suggest CO2 leads to “Arctic amplification” arguing dark open oceans absorb more heat. But the latest estimates show the upper 700 meter of the Arctic Ocean are cooling (see illustration below), which again supports the notion ventilating heat raised air temperatures. Read Wunsch, C., and P. Heimbach, (2014) Bidecadal Thermal Changes in the Abyssal Ocean, J. Phys. Oceanogr., http://dx.doi.org/10.1175/JPO-D-13-096.1.

So how much of the global warming trend is due to heat ventilating from a cooling Arctic ocean???

Change in top 700 meters of Ocean Heat Content between 1993 and 2011

2. Consider that NOAA’s graph is based on homogenized data. Researchers analyzing homogenization methods reported “results cast some doubts in the use of homogenization procedures and tend to indicate that the global temperature increase during the last century is between 0.4°C and 0.7°C, where these two values are the estimates derived from raw and adjusted data, respectively.”

Read Steirou, E., and Koutsoyiannis, D. (2012) Investigationof methods for hydroclimatic data homogenization. Geophysical Research Abstracts, vol. 14, EGU2012-956-1.

So how much of the recent warming trend is due to the virtual reality of homogenized data???

3. Consider the results from Menne. M., (2009) The U.S. HistoricalClimatology Network Monthly Temperature Data, version 2. The Bulletin for the American Meteorological Society, in which they argued their temperature adjustments provided a better understanding of the underlying climate trend. Notice the “adjusted” anomalies in their graph below removes/minimizes observed cooling trends. More importantly ask why does Menne (2009) report a cooling trend for the eastern USA from 1895to 2007, but NASA shows a graph (below Menne’s) with a slight warming trend for all of the USA from 1950-2014?  Does that discrepancy indicate more homogenization, or that they cherry-picked a cooler period to start their warming trend? 

How homogenization warmed the USA

NOAA's 1950-2014 global warming trend 

4. Consider that most of the warming in North America as illustrated by Menne 2009 (above) happened in the montane regions of the American west. Now consider the paper Oyler (2015) Artificial amplification of warming trends across the mountains of thewestern United States, in which they conclude, “Here we critically evaluate this network’s temperature observations and show that extreme warming observed at higher elevations is the result of systematic artifacts and not climatic conditions. With artifacts removed, the network’s 1991–2012 minimum temperature trend decreases from +1.16°C/decade to +0.106°C/decade.

So how much of the recent warming trend is due to these systematic  artifacts???

5. Consider that NOAA’s graph is based on adjusted data and the fact that NOAA now homogenizes temperature data every month and climate trends change from month to month, and year to year. As an example, below is a graph I created from the US Historical Climate Network Cuyamaca weather station in southern California; a station that never altered its location or instrumentation. In 2011 the raw data temperature trend does not differ much from the homogenized trends (Maximum Adj.)

US Historical Network raw and homogenized maximum temperatures at Cuyamaca

Just 2 years later, the 2011 homogenized century warming trend (in black ) increased by more than 2°F the 2015 trend (in red.) I have archived several other similar examples of this USHCN datamanipulation. Then ask your self which is more real? The more cyclical changes observed in non-homogenized data or the rising trend created by homogenized virtual reality?

Cuyamaca's rapidly warming trend created by homogenization

6. Consider that climate change along western North America has been completely explained by the Pacific Decadal Oscillation and the associated cycles of ventilation and absorption of heat. Read: Johnstone and Mantua (2014) Atmospheric controls on northeast Pacific temperature variability and change, 1900–2012.  Such research suggests non-homogenized data may better represent climate reality.

Knowing that the upper 10 feet of the oceans contain more heat than the entire atmosphere ask yourself if decadal warming trends are simply artifacts of the redistribution of heat.

7. Consider that increasingly temperature data is now collected at airports. A 2010 paper by Imhoff, “Remote sensing of the urban heat island effectacross biomes in the continental USA”, published in Remote Sensing of Environment 114 (2010) 504–513 concluded that “We find that ecological context significantly influences the amplitude of summer daytime urban–rural temperature differences, and the largest (8 °C average) is observed for cities built in biomes dominated by temperate broadleaf and mixed forest. For all cities combined, Impervious Surface Area is the primary driver for increase in temperature explaining 70% of the total variance. On a yearly average, urban areas are substantially warmer than the non-urban fringe by 2.9 °C, except for urban areas in biomes with arid and semiarid climates.”

So how much of this recent warming trend can be attributed to increases in Impervious Surface Area in and around weather stations in rural, suburban and urban settings?

8. Consider that direct satellite observations show lost vegetation has a warming effect, and transitions from forest to shrub land, or grassland to urban area raise skin surface temperatures by 10 to 30°F.  Satellite data reveals the canopies of the world’s forests averaged about 86°F, and in the shade beneath the canopy, temperatures are much lower. Grassland temperatures are much higher, ranging from 95 to 122°F, while the average temperatures of barren ground and deserts can reach 140°F. Read Mildrexler, D., et al. (2011) A global comparison betweenstation air temperatures and MODIS land surface temperatures reveals thecooling role of forests. J. Geophys. Res., 116, G03025, doi:10.1029/2010JG001486.

Ask yourself, “how much of the warming trend is due to population effects that remove vegetation??” How much is due to citizens of poorer nations removing trees and shrubs for fuel for cooking and heating or slash and burn agriculture?

9. Consider that neither of the satellite data sets suggest 2014 was the warmest ever recorded.

Global temperature trend from satellite data 

10. Consider that none of the tree ring data shows a warming that exceeds that 1940s as exemplified by Scandinavian tree ring data (from Esper, J. et al. (2012) Variability and extremes of northern Scandinavian summer temperatures over the past two millennia. Global and Planetary Change 88–89 (2012) 1–9.)

Tree ring and Scandinavian instrumental data show show warmest decade in the 1930s

Consider international tree ring experts have concluded, “No current tree ring based reconstruction of extratropical Northern Hemisphere temperatures that extends into the 1990s captures the full range of late 20th century warming observed in the instrumental record.”  Read Wilson R., et al., (2007) Matter of divergence: tracking recent warming at hemispheric scalesusing tree-ring data. Journal of Geophysical Research–A, 112, D17103, doi: 10.1029/2006JD008318.

In summary, after acknowledging the other factors contributing to local temperature change, and after recognizing that data homogenization has lowered the peak warming of the 30s through the 50s in many original data sets by as much as 2 to 3°F, (a peak warming also observed in many proxy data sets less tainted by urbanization effects), ask yourself, does NOAA’s graph and record 2014 temperatures really tell us anything about climate sensitivity or heat accumulation from rising CO2? Or does it tell us more about climate politics and data manipulation?

NOAA's Global Temperature Trend: What does it tell us about the causes?

On Migrating Moose and Migrating Temperature Trends

 

Bull Moose

The biggest threat to the integrity of environmental science is bad science, exaggeration and fear mongering. The recent hype about declining moose populations is just one more example of global warming advocates hijacking and denigrating ecological science. All organisms act locally, yet global warming advocates quickly characterize any local wildlife declines as the dastardly work of global warming.

In northeastern Minnesota, moose populations reached an historic high abundance of 8,840 in 2006, and then rapidly declined to 4,230 in 2012. Most recently in a 2013-2014 survey, estimates dropped to 2,760 moose. Cause for alarm? Perhaps. But moose are a species known to naturally exhibit booms and busts when their habitat can no longer sustain a rapidly growing population. Instead of deeper discussions on the ecological complexities, but reminiscent of the fearful headlines that “childrenwill no longer know what snow is”, the NationalWildlife Federation (NWF) bellowed “People never forget seeing their first moose. But due in part to the effects of climate change, it could well be their last. Moose are being hurt by overheating, disease and tick infestation – all tied to warming temperatures.”  And to magically save the moose, the NWF encourages you to sign their petition to the EPA to curb CO2 emissions, and for just $20 to $50 you can adopt a moose from the NWF. Presumably the $50 moose is in its prime and carries fewer ticks.

The Audubon Society similarly published MysteriousMoose Die-Offs Could be Linked to Global Warming and climate scientists like MichaelMann, who has hitched his scientific status to “dire predictions”, wrongly connect declining moose populations to rising CO2. There are so many reasons to be revolted by their fear mongering and its denigration of ecological science, it’s hard to know where to start. For instance, the greatest spike in moose mortality happens in March at the end of severe winters. Milder winters can be beneficial. While alarmists blame moose deaths on “global warming”, the rapid decline in northeastern Minnesota has happened in a region experiencing bouts of record breaking low temperatures. Nearby International Falls, MN broke its record January low of -37°F set in 2010, by dropping to -41°F in 2014, which followed December’s record setting 8 days with a temperature of less than -30°F. Averaging local temperatures is likely as useless as referring to the global temperature.

Furthermore moose die‑offs are not global. In adjacent habitat of southernOntario, moose are stable or increasing. Estimates of moose populations have traditionally been based on hunters’ harvest and in Scandinavia, the annual harvest was less than 10,000 in the early 1900s. After a century of global warming the moose population reached an all time high with annual harvests increasing20­‑fold to 200,000. Similarly in the 1900s, moose from British Columbia expanded into Alaska and multiplied as the climate warmed. In New England, moose were once more abundant than deer when the Pilgrims arrived. But due to deforestation for farmland and overhunting, moose have been absent from Massachusetts and Vermont for 200 years. In 1901 less than 20 moose were believed to inhabit New Hampshire. But in contrast to fearful globalwarming  theory and species range, since 1980 moose have migrated south from New Hampshire into Massachusetts and Connecticut, despite temperatures that average 4 to 6° F warmer.

Scandinavian biologists suspect the moose population may begin to decline, but their reasons illustrate the complex ecology. Increasing moose densities strain food supplies resulting in lower body mass, lower reproductive success, and lower resiliency. Moose thrive on vegetation common in regenerating forests that have been cleared by insect outbreaks, fires or logging. Scandinavian logging increased during the 20^th century but has now peaked and will decline, and so moose will lose habitat as closed forest canopies reclaim the landscape . Except in eastern Finland, depredation by wolves has been minimal, but wolf populations are now rebounding.

The beststudied moose population exists just east of Minnesota’s northeast border on Isle Royale in Lake Superior and illustrates the boom and bust nature of moose populations. As moose populations globally expanded in the 1900s, they soon colonized Isle Royale around 1912 and rapidly grew to over 3000 by early 1930s. Rapid population growth diminished food supplies and a starvation crash happened in1934. Extensive forest fires in 1936 increased their preferred vegetation and feasting on young vegetation in a regenerating forest, the population rebounded until it peaked , followed by increasing winter starvation. To add another factor governing moose population, in the 1940s wolves colonized Isle Royale.

Wolves attacking a moose

Virtually every college ecology text discusses the predator-prey interactions illustrated by the wolves and moose of Isle Royale. As observed elsewherein the Great Lakes region, moose populations remained low until they began increasing in the 1950s. As seen in the diagram below, moose populations rose but also ebbed and flowed inversely with wolf populations. In contrast to suggestions that global warming is killing moose, during the rapid warming from the 80s to 90s, Isle Royale moose doubled their population, approaching a peak not observed since the 1930s, then suddenly crashing to just 500 in 1997. Moose have slowly rebounded since 2007 and are now at levels 50% higher than the 1950s.

Relationship between moose and wolf populations at Isle Royale

In response to the dramatic decline of moose in northeastern Minnesota, over 100 moose were equipped with radio-collars that could alert biologists to the moose’s impending death, allowing biologiststo account for the deaths of 35 calves and 19 adults.

- 16 calves (46%) were killed by wolves

- 13 calves (37%) calves died due to mother abandonment. Eleven were caused when the mothers abandoned the calve during the act of attaching the collars,  2 were abandoned later.

- 4 calves (11%)  were eaten by bears

- One calf drowned and 1 calf died of unknown causes.

- Of the 19 adults, 10 (53%) were killed directly or indirectly by wolves.

Oddly given those results, biologist received a new $750,000 grant to study the effects of “global warming” on declining moose. I suspect it is politically more convenient to blame declining moose on global warming rather than to blame natural boom and busts, rebounding wolf populations, or researcher induced casualties.

Similar fear mongering blamed global warming for recent declines in New Hampshire’s moose. On a PBSNewshour, the interviewer interspersed interviews with researchers and Eric Orff of the National Wildlife Federation who insinuated that it’s all about climate change. Like the debunked claims of Parmesanthat global warming is killing animals in the south, Orff highlighted dwindling moose populations on the southern end of their range, concluding, “we need to put this earth on a diet of carbs, carbon, and bring back winter.” But New Hampshire’s average temperature has little meaning. Moose can respond to temperature changes by moving to different microclimates. Between a gravel road, open shrub lands, ponds, and closed canopies of deep evergreen forest, temperatures will vary by 20° to 40°F. A mosaic of habitats is more critical than a 1° degree change in average temperature.

In addition, Orff failed to mention that moose have been migrating from New Hampshire southward and thriving where climates averaged 4°F to 6°F warmer and winters are much milder. Orff also failed to inform the public about normal population boom and busts. New Hampshire’s moose population stagnated at fewer than 15 individuals since the mid 1800s and did not begin to rebound until the 1970s. As the climate warmed numbers exploded, by 1988 growing to 1600, and then 7500 by the late 1990s. That increase resulted in more moose‑car collisions and a public clamor for increased moose hunts.  Perhaps because the public would be less likely to “adopt a moose” that needed to be hunted, Orff failed to   mention that according to Fishand Game about half of New Hampshire’s recent population drop from 7500 to 4000 moose was due to a public safety management decision to increase hunting.

New Hampshire’s remaining decline has been blamed on moose ticks, which some suggest have increased due to milder winters. Perhaps. But moose also survive better during milder winters. On average moose are covered with 30,000 ticks and each tick can lay a thousand eggs. When moose populations explode so do the ticks. Unprecedented tick abundance coincides with unprecedented moose populations. Besides biologists have observed such parasite‑driven booms and busts for over a century.

Growing up in Massachusetts, moose were unheard of so far south. We travelled north to Baxter State Park in Maine to canoe the streams with hopes of seeing moose. Moose are indeed sensitive to warmer temperatures, so why would moose migrate southward to a warmer region that was also experiencing rapid “global warming”. Homogenized data suggested a rapid warming trend but as an ecologist, I knew homogenized temperatures are worthless for wildlife studies because the process eliminates natural temperature variations and alters the actual mean temperatures. However I also understood that trends determined from raw data can suffer from changes in instrumentation and/or changes in location.

I first looked at minimum temperatures for the only 2 USHCN weather stations in western Massachusetts where moose populations had been thriving since the 1980s. Both stations exhibited peak warming around the 1950s in the raw data, but after homogenization, that peak was lowered. For Amherst the peak dropped by 4°F. Onto the graphs downloaded on January 10 from the USHCN, I superimposed changes in instrumentation (designated by the vertical red lines) and changes in location (designated by the blue arrows). But those changes did not logically or intuitively explain the newly fabricated warming trend or the cooling of the 1950s peak. (raw data on left, homogenized on right)

Raw (left) and Homogenized (right) Minimum Temperatures for Great Barrington and Amherst MA

So I looked for a USHCN station with no such changes. Only one Massachusetts station, West Medway (below), had not moved and did not change thermometers. I assumed it would serve as the best standard with which to constrain any trend adjustments at other stations.  Yet West Medway was also homogenized (below on right) creating the same virtual warming trend. More importantly, West Medway’s raw minimum temperature trend had the same basic curve as the 2 western stations. 

The homogenization process for both NOAA and BEST creates a “regional expectation” based on similarities among neighboring stations, which in turn guides their temperature adjustments. But if USHCN stations are deemed to be of the highest quality with the fewest gaps and relocations, what data (likely much less reliable) was being used to re-create West Medway’s warming trend. If West Medway’s raw data shared similar trends with nearby stations, wouldn’t Medway’s trend be a reliable “regional expectation”? More troubling, the homogenization process undeservedly altered observed temperature peaks. Like Amherst, homogenization lowered West Medway’s 1950’s peak by 3 to 4°F, a lowering that was also applied to many other stations such as the Reading station (raw data left, homogenized right).

 

Raw and Homogenized Minimum Temperatures at West Medway, MA

Raw and Homogenized Minimum Temperatures at Reading MA

So I was curious how the raw data from West Medway’s nearby USHCN stations compared and affected the regional expectation. The Blue Hill Observatory (below left) sits just 28 miles east of West Medway and is a historical landmark that has not moved. Its trend agrees with West Medway, peaking in around 1950 and then cooling until 1980. However after 1980, due to changes in instrumentation, it is not clear how much of the exaggerated rising trend is due to climatic factors (natural or CO2) or the result of a warming bias caused by new instruments.

Taunton (below right) is located 29 miles southeast of West Medway.  It too exhibits a peak around 1950 and a cooling trend to 1980. However once again the cause of the subsequent warming trend is obscured by the change in the measuring system. However there was one nearby station, Walpole that maintained the same equipment.

Raw Minimum Temperatures at Blue Hill Observatory and Taunton MA

Walpole (below) is situated just 12 miles east of West Medway and just west of the Blue Hill observatory. But Walpole exhibited a warming trend more similar to Massachusetts’ homogenized trend. Of which of those stations should anchor a “regional expectation”? Walpole’s raw data had an odd curve not shared by most of the other stations. Although all stations experienced a warming spike during the 1972-74 El Nino/La Nina event, that peak was typically a degree lower than the 1950s. However Walpole reported an unusually higher 70s peak suggesting that after 1950 the weather station had moved to a warmer microclimate. But NOAA’s metadata did not specifically mention any relocation. Thinking I had missed that information, I rechecked. Although a relocation was not specifically mentioned, the GPS coordinates revealed a significant move in 1973. Yet comparing the raw data (below left) to the homogenized data (below right red), Since 1915 Walpoles raw data remained un-homogenized. Did the warming bias from the 1980s instrumental changes, create a confirmation bias for Walpole?

Raw Minimum Temperatures(left) and Quality Controlled vs Homogenized TMIN at Walpole

What I assume is a most reasonable method to quality control for a location change,  I compared the differences between West Medway (the only unaltered site) and Walpole’s minimum temperatures before and after Walpole’s location change. Between 1905-1973 Walpole averaged 0.3596 +/- 1.07°F warmer than Plymouth. Walpole could vary between 2° cooler one year to 4.6° warmer another. This great variability is natural and expected. Depending on how far east winter storm tracks travel up the east coast, the battle line between cold arctic air masses to the west and warm Atlantic air to the east causes significant temperature changes. Depending on the depth and extent of the cold air mass, the overriding warm Atlantic air can cause different parts of the state to simultaneously experience rain, freezing ice, sleet and snow.

After the station moved, between 1974-2004 Walpole temperatures averaged 2.89 +/- 1.29F warmer than Medway, but with similar year-to-year variability ranging between 1.5 cooler one year to 4.5 warmer another. It is impossible to adjust for such local variations. But to extract a climate trend, it is reasonable to subtract the difference in mean temperatures before and after the relocation. So I subtracted 2.53 F (2.89-0.35) from all Wapole temperatures after 1973 to create my “quality controlled” trend (blue) and plotted that against the USHCN homogenized trend (red in graph above right). Unsurprisingly Walpole’s “quality controlled” data and West Medway’s raw data exhibit very similar trends with peaks and valleys coinciding with the Atlantic Multidecadal Oscillation (AMO).

So I more carefully checked the data from Plymouth about 52 miles to the southeast. Unfortunately data from the Plymouth weather station does not extend back to the landing of the Pilgrims, which marked the beginning of the end for moose in Massachusetts. But after adjusting for Plymouth’s 2 obvious location changes, in 1966 and 1990 (blue arrows), Plymouth’s “quality controlled” data revealed a trend very similar to West Medway and a “regional expectation” related to the AMO. Most interesting, once Plymouth’s location change was accounted for there was no instrumental warm bias. As discussed by Davey and Pielke, a warming bias if often associated with MMTS temperature instruments, because new instruments and a new location happened simultaneously. Insignificant location changes could cause a warming bias when weather stations were moved closer to a building and subjected to a warmer micro-climate.

Raw Minimum Temperatures(left) and Quality Controlled vs Homogenized TMIN at Plymouth MA

It is highly likely that due to its effect on storm tracks and competing air masses, the AMO can explain most of the east coast’s temperature trends in a manner similar to how the Pacific Decadal Oscillation controls the USA’s west coast trends as published by Johnstone 2014.  Unfortunately this relationship has been obscured by a highly questionable homogenization process.

To be clear, I am not suggesting a conspiracy of data manipulation by climate scientists. I am arguing that the homogenization process is ill-conceived and erroneously applied. Many local dynamics are overlooked  by a one-size fits all digital make over. Monthly homogenization can amplify those mistakes and has changed trends from one year to the next (as discussed for Death Valley). Homogenization has failed to adjust for documented location changes, yet created adjustments to untainted data where none were needed. Before we conclude that global warming is killing moose and creating unusually warmer winters, we need examine more closely local dynamics and their relationship to landscape changes and natural ocean cycles much more closely. Understanding local micro-climates are more important that a nebulous global climate. While it may be wise to think globally all organisms react locally, as do all weather stations.

NOAA's Atlantic Multidecadal Oscillation Index