adventures in chart-crime: making climate data look scary

if a picture is worth 1000 words, imagine how many lies it can tell

the key to engendering a sense of crisis is to remove any sense of perspective. it’s a classic of the genre. “risk is up 50%!” is this a big deal? without a baseline risk figure, you really cannot say. a 50% rise in your risk of getting cancer would probably lead you to reconsider behavior. a 50% rise in risk of being hit my a meteorite? probably not.

this was used to great effect by the covid crowd with claims like “50% reduction in deaths” from vaccines or from other drugs. leaving aside the frequent inaccuracy, rigging, and even falsification of such data: it’s still incredibly misleading. a 50% drop from what? if your prior death risk was 1 in 10,000, halving it is pretty immaterial and even tiny adverse events rates would swamp benefit and lead to negative expected value. but it was presented in a manner that made this hard to see. and let me tell you, if you think the drug cos and health agencies are good at misrepresentation, well buckle up, because they are absolute pikers compared to the climate crowd…

i’m not going to get into the quality of the data here. i have done so in some other places like THIS and the issues with the temperature datasets both running increasingly hot due to bad siting and calibration from bad terrestrial data and the stubborn refusal of the USHCN climate reference network to confirm the rise certainly lead to fertile grounds for debate there, but for now, let’s focus solely upon presentation and allow a gato to play the penn and teller and show you how the trick is done.

the first common trick to make small changes look large is to use variance measures instead of absolute. (charts from anthony watts’ excellent website)

this is the data as reported by UAH (university of alabama huntville)

it looks like a stark rise. but what you are really seeing is 0.6 degrees of change over 44 years. this is not actual temperature, it’s divergence from a baseline (in this case 1991-2000). the chart looks alarming, but you’re actually speaking about a 4.4 decade change less than those often seen in 15 minute periods in any given place.

NASA plays a similar game on scaling by using absolute temperature but manipulating chart scaling. it makes 1.7 degrees (F) over 140 years look like a huge jump.

but look what happens when you use a zero based scaling:

bit less scary in that perspective, no?

but even the timeframe here is manipulated. this data starts in 1880 just as the world was coming out of “the little ice age” caused by the superset of the dalton and maunder minima of solar activity. and that was the coldest period in the last ~9000 years. reams of data back this up. the medieval warm period was warmer than today, the roman warmer still, and the minoan even hotter. none were anywhere close to the holocene climate optimum.

this is a useful compendium of studies on medieval warm period (MWP) temperatures.

the upshot is that the phenomenon was global, and temps were ~1 degree or so warmer.

this was well established canon in the pre-al gore/michael mann era. they erased this issue. the “climategate” scandal of data tampering was all about disappearing the MWP.

“the world has warmed” is not actually a meaningful statement. you always need to ask “since when?” since this morning? since 1880? since 7000 BCE? since multi-cellular life evolved 500mm years ago?

because that one is fun:

the world is actually very cold right now. ice at one pole is uncommon and ice at both poles is quite rare.

but this chart is also a lie because the data it uses is what is called a “splice” where they take recent data from one series (thermometers) and graft it into a set of proxy data from ice cores, sediment, etc. that “spike” at the end is severely exaggerated .

they do this to make “hockey sticks” which look dramatic. but it’s bad stats handing and you should never trust a change in trend that aligns with/is driven by a change in data source.

the world is currently in a deep ice age (last 40mm years or so) and experiences ~13-15k year warming periods (like the one (called the holocene) we inhabit) between ~150k year “snowball earth” periods where ice will once more bury boston a mile deep and reach to the carolinas. this is driven by orbital mechanics and is not a thing we can stop or alter at current technology levels. the next big climate move for earth will be back to the freezer.

the chart below was once canon for the holocene interglacial.

it looks like this and prior to very recently, this data was widely understood to be a 7500 year downtrend in temps that ran away to the downside during the little ice age, then recovered and resumed the slower downtrend.

but now it gets adulterated by grafting on thermometer data to claim that today is warmer than not only the medieval period, but the actual holocene climate optimum 7500 years ago.

it’s a data splice and an absurdity. a perusal of consistent proxies (as linked in the MWP site) makes this clear as does the fact that viking villages keep emerging from greenland and icelandic glaciers. the vikings grew wine on newfoundland. the romans did so in london.

this idea that temps are higher now than then is a data fabrication.

it’s used to mislead.

when one uses a single proxy sans splices, the story is quite different and while, of course, some regional variance is possible, this pattern holds in many, many places using many, many proxies. (except for bristlecone pines which should be ignored as they respond more to CO2 than temps)

but nowhere is the misleading data worse than sea ice.

this chart has been going around.

i cannot speak to its accuracy as i did not find a way to reproduce it using the website it claims to be from, but regardless, here’s your practical exam:

spot the manipulation:

yup. scaling and variance from some mean. are these big numbers or small? there is no way to know. is this a loss of 1%? 10% 90%? you cannot tell.

amusingly, the same website cited allows for a high perspective visualization:

quite a different story, no? typical annual variation is 15-16mm km2 and 80-85% of ice is lost annually from peak to trough. nothing terribly dramatic just happened and the largest ice extent since the satellite record began in 1979 was in 2014.

so, this is a very short dataset and is just not showing much change. last year’s peak was higher than many years in the 90’s. this low was a series low, but it’s a short series, and the low was only .13 lower than the previous low in a series with 16mm annual variance. that’s not even 1%. so, perhaps not the biggest deal…

compounding the issue, “sea ice extent” is a very squishy measure (and we’re on our 5th set of instruments to measure this data)

• The sea-ice extent is calculated as the areal sum of sea ice covering the ocean where sea-ice concentration (SIC) exceeds a threshold (15% for AMSR-E). SICs are derived from various satellite-borne passive microwave radiometer (PMR) sensors using the algorithm developed and provided by Dr. Comiso of NASA GSFC through a cooperative relationship between NASA and JAXA. The following sensor's data were used;

  •Jan. 1980 ～ Jul. 1987：SMMR

• •Jul. 1987 ～ Jun. 2002：SSM/I

• •Jun. 2002 ～ Oct. 2011：AMSR-E

• •Oct. 2011 ～ Jul. 2012：WindSat

• •Jul. 2012 ～ the present：AMSR2

it’s water with >15% ice coverage. so, technically, 100 square km with 14% coverage is deemed to have less ice than 10 km2 with 15%. this metric can be highly affected by winds, currents, etc and potentially quite a poor measure of actual ice quantity. this “new low” is well within the margin of error on actual ice volume and it’s not clear how meaningful the measure is overall in any event.

so perhaps take some of these claims with a grain of salt. (and possibly some tequila. and some lime for the more adventurous.)

they are mostly chart-crime intended to alarm, not inform.