USHCN – sunshine hours

USHCN Infilling – A rebuttal to Nick Stokes

Published on July 9, 2014July 9, 2014 by sunshinehours12 Comments

Nick Stoke and Zeke Hausfather don’t like my graphs showing a difference between Estimated (infilled) and non-Estimated (not infilled) USHCN data

Nick said:

“The difference between average estimated and average non-estimated, doesn’t reflect estimation. It just reflects changes in the kind of stations that were being estimated.

For some reason, they were more likely to be warmer. I don’t know why, but they were. It’s just the wrong way to do it.”

I’m not sure I believe in coincidences.

I believe Nick and Zeke both created an anomaly baseline for each month for each station. Nick used 1900 to 2013. Zeke said he used 1961-1990.

Neither of them added trend lines.

I believe both used all the data (infilled and non-infilled). (As of my posting this they had not confirmed). I disagree with that. Infilled data should not be used in the baseline.

As an example of what infilling does using the difference by station from the 1895-2013 baseline here is Dec 1998 to 2013 Tmax.

In this case, the real data had a downwards trend of -0.82C/decade and the infilled data was -0.64C/decade which is .18C/decade higher not because infilled data was hotter, but because it was less cold.

From 1895 to 2013 the infilled trend is consistently .02C/decade to .03C/decade higher than the non-infilled trend.

The red dots are higher on the 1980 and later side of the graph while they are lower on the pre-1960 part of the graph.

Cooling the past, warming the present.

Misleading Information About USHCN At Judith Curry’s Blog

Published on July 7, 2014July 8, 2014 by sunshinehours14 Comments

Zeke has an post over at Judith Curry’s blog where he makes a claim that “infill has no effect on CONUS-wide trends” (among other claims).

I disagree. Here are the 12 months comparing Estimated and Non-Estimated and then all the data. Blue = Final data (non-Estimated) Red = Final Data Estimated Green = All Final Data.

Infilling does change the trend. For example, Estimating changed the trend in Dec from .06C per decade to .07C per decade because the Estimated data had a trend of .15C/decade.

USHCN Tmax – Hottest July Histogram – Raw vs Adjusted

Published on July 3, 2014July 3, 2014 by sunshinehours13 Comments

Yesterday I did a post looking at every station and finding out which year had the hottest July. Today I am showing you what effect the adjustments have.

I am using USHCN Final Tmax monthly data and comparing it to the raw data . From the file v2.5.0.20140627. (This is not necessarily all the data NOAA uses)

The format is a little different, but it is designed to allow easy comparison. Red is the raw data. Green is Final and muddy green is where raw and final overlap.

As you can see, the number of years near the present have a lot green. Which means the adjustments created more record hot years.

The 1910s to 1980s have red at the top which indicates raw data. Translation: More record July’s occurred in the past using raw data.

For the Coldest July data, the results are the opposite. Adjusting creates more colder July’s in the past and takes away cold July’s from the present.

USHCN Tmax – Coldest and Hottest July Histogram Will Surprise You!

Published on July 2, 2014July 2, 2014 by sunshinehours12 Comments

The other day I noted that NOAA had July 1936 back on top. Now don’t take this post as an endorsement of the adjustments and infilling, but I thought I would check which years did stations set their hottest July and coldest July.

I am using USHCN Final Tmax monhtly data . From the file v2.5.0.20140627. (This is not necessarily all the data NOAA uses)

I don’t think the histogram of the hottest July will actually be a surprise. 216 stations had their hottest July in 1936.

But the coldest July histogram will surprise you. 1992 and 1993 tied with 148 stations each having the coldest July. A small subset of the stations at bottom.

I just picked the top 15 stations with the coldest July in 1992 and the hottest July in 1936. There are 92. The temperatures are in Celsius.

STATE	Station	NAME	Coldest_Year	Coldest_July	Hottest_Year	Hottest_July
IA	USH00130133	ALGONA 3 W	1992	23.22	1936	35.98
IA	USH00131402	CHARLES CITY	1992	24.18	1936	34.5
IA	USH00132724	ESTHERVILLE 2 N	1992	23.06	1936	35.82
IA	USH00132977	FOREST CITY 2 NNE	1992	23.69	1936	35.72
IA	USH00132999	FORT DODGE 5NNW	1992	24.73	1936	36.1
IA	USH00134063	INDIANOLA 2W	1992	26.57	1936	37.13
IA	USH00134894	LOGAN	1992	26.56	1936	39.88
IA	USH00135952	NEW HAMPTON	1992	23.38	1936	35.31
IA	USH00137147	ROCK RAPIDS	1992	23.96	1936	37.48
IA	USH00137161	ROCKWELL CITY	1992	24.81	1936	36.69
IA	USH00137979	STORM LAKE 2 E	1992	23.85	1936	36.54
IA	USH00138296	TOLEDO 3N	1992	24.82	1936	35.34
IL	USH00115326	MARENGO	1992	24.66	1936	33.89
IL	USH00118916	WALNUT	1992	24.94	1936	36.2
IN	USH00122149	DELPHI 2 N	1992	26.5	1936	36.38

Here are a sample of 15 that have neither.

STATE	Station	NAME	Coldest_Year	Coldest_July	Hottest_Year	Hottest_July
AL	USH00011084	BREWTON 3 SSE	1989	30.77	1902	36.71
AL	USH00012813	FAIRHOPE 2 NE	1928	30.05	2000	34.93
AL	USH00013160	GAINESVILLE LOCK	1916	30.25	1952	37.26
AL	USH00013511	GREENSBORO	1967	28.49	1901	35.09
AL	USH00015749	MUSCLE SHOALS AP	1967	29.59	1930	36.57
AL	USH00017157	SAINT BERNARD	1967	27.33	1952	35.2
AL	USH00017304	SCOTTSBORO	1906	29.21	1980	35.86
AL	USH00017366	SELMA	1916	28.1	2000	36.65
AL	USH00018178	THOMASVILLE	1916	29.87	1930	36.35
AL	USH00018380	TUSCALOOSA ACFD	1889	29.5	1930	37.77
AL	USH00018469	VALLEY HEAD	1967	28.31	1930	36.48
AR	USH00031596	CONWAY	1989	27.83	1954	38.96
AR	USH00031632	CORNING	1905	29.14	1930	38.45
AR	USH00032356	EUREKA SPRINGS 3 WNW	1950	27.51	1980	38.65
AR	USH00032444	FAYETTEVILLE EXP STN	1950	27.86	1954	38.03

USHCN 2.5 – Kansas Mapped – July 1936 and 2012

Published on June 29, 2014June 29, 2014 by sunshinehours13 Comments

UPDATE: I am adding the following means:

1936 raw / tobs / final mean = 101.47 / 100.83 / 100.32

2012 raw / tobs/ final mean = 99.57 / 99.66 / 99.48

Original Start of post:

I’ve been doing posts about USCHN and Estimated data. I wanted to visualize.

This is the USHCN July 1936 and 2012 TMAX data for Kansas.

Form the most part, in July 1936 the red data (Final) is colder than the raw temperatures (black) and in 2012 it is the other way around.

The data is in F (originally in C so there may be very slight conversion issues).

Black temperatures are raw, blue are TOBS and red are Final adjusted. Stations with just a red have no raw data. The data is just ‘E’stimated.

Click for full size.

USHCN 2.5 – Estimated Data Is Warming Data – USA 1945 to 1980

Published on June 29, 2014 by sunshinehours1Leave a comment

I’ve been posting on USHCN and the effect of “estimating” or “Infilling” on the Final data.

Earlier today I showed that Estimating/Infilling made the 1980-2014 trend steeper upwards.

Now I will show the opposite: When the trend is down, infilling makes the trend steeper downwards.

I repeat … this is the Final data after all the other adjustments. About 15% of the data is Estimated from neighboring stations.

I will post all the monthly graphs … but just discuss the first – January.

The data is 1945 to 1980. So it covers a period of cooling.

The trend of REAL data is –0.58C/decade. That’s the 39,766 values referenced in the legend.

Then they add in about 15% Estimated data with a trend of -0.79C/decade. That’s the 3987 values.

The net result is a new trend of -0.61C/decade.

Presto. Magic. A -0.58C trend is now a -0.61C trend.

Not a big change. But it is always there. But the red line is almost always well above the others. Sometimes by 2C.

(Click on graphs for larger).

USHCN 2.5 – Estimated Data Is Warming Data – USA 1980 – 2014

Published on June 29, 2014 by sunshinehours12 Comments

I’ve been posting on USHCN and the effect of “estimating” or “Infilling” on the Final data. I did Arizona earlier.

I repeat … this is the Final data after all the other adjustments. About 15% of the data is Estimated from neighboring stations.

I will post all the monthly graphs … but just discuss the first – January.

The data is 1980 to 2014. So it has the 1980-1990 warming trend in it.

The trend of REAL data is 0.23C/decade. That’s the 35,854 values referenced in the legend.

Then they add in about 15% Estimated data with a trend of +0.66C/decade. That’s the 4516 values.

The net result is a new trend of +0.33C/decade.

Presto. Magic. A .23C trend is now a .33C trend. (Click on graphs for larger).

USHCN 2.5 – OMG … The Old Data Changes Every Day (Mapped)

Published on June 28, 2014June 28, 2014 by sunshinehours12 Comments

Update: I added a few other states

I’ve been doing posts about USCHN and Estimated data. I wanted to visualize.

This is the USHCN May 2014 TMAX data for Illinois.

The data is in F (originally in C so there may be very slight conversion issues).

Black temperatures are raw, blue are TOBS and red are Final adjusted. Stations with just a red have no raw data. The data is just ‘E’stimated.

Click for full size.

A few other states:

USHCN 2.5 – OMG … The Old Data Changes Every Day Updated

Published on June 28, 2014June 28, 2014 by sunshinehours12 Comments

UPDATE: I got a request for a histogram by Decade for the differences. I have attached them at the bottom.

A few days ago I did a post about the USHCN data changing every day. I focused on one month of one year,

So this is a stats update. From Jun 21 2014 to Jun 22 2014 the TAVG data had 22,996 temperature values changed (out of 138136)

The changes were spread out through the years. And the change was predominantly to warm the data.

This is a histogram of the Year.

This is a histogram of the change. (The right side of 0 says the data got warmer)

< a href=”https://sunshinehours.files.wordpress.com/2014/06/1950s-difference-ushcn-tavg-v2-5-0-20140622-compared-to-v2-5-0-20140621.png”> 1950s Difference USHCN TAVG v2.5.0.20140622 compared to v2.5.0.20140621

USHCN Only 51 Stations Have A Full Set Of Monthly Data For 1961-1990

Published on June 25, 2014June 25, 2014 by sunshinehours12 Comments

UPDATE: Published. Not in moderation.

Over at WUWT I have a couple of comments in moderation in the argument about US climate data. This one of them (slightly clarified).

Zeke is quoted: “The way that NCDC, GISS, Hadley, myself, Nick Stokes, Chad, Tamino, Jeff Id/Roman M, and even Anthony Watts (in Fall et al) all calculate temperatures is by taking station data, translating it into anomalies by subtracting the long-term average for each month from each station (e.g. the 1961-1990 mean)”

To create a 1961-1990 baseline, you would have 360 monthly values.

The USHCN monthly data has error flags. The E flag means the data for that month is Estimated. There is not enough daily data to created a monthly value.

There are ONLY 51 stations that had 360 values without an E flag from 1961-1990.

That means only 51 out of 1218 stations have relatively complete data to use as a baseline.

WY MORAN 5 WNW USH00486440 is one of the 51

WY NEWCASTLE USH00486660 is one that failed. 61 months of the 360 had an E flag. (Admittedly my comment had a typo over at WUWT).

And I just looked at the E flag. There lots of other flags.

Here is the other comment:

Anthony, you should double check Zeke’s work.

Using USHCN Final Tavg dated v2.5.0.20140622

July 2012 – 880 Stations have data without the E for Estimated flag.

There are 1218 stations.

27% of the July 2012 Stations are missing data.

July of 1895 has 472 station reporting Real (non-Estimated) data

61%. of the July 1895 stations are missing data.

Now remember, I am only look at the monthly records. Monthly records avoid the E flag if there are enough daily data. It doesn’t mean there is data for every day.