USCRN Cooling

What is USCRN?

One of the principal conclusions of the 1997 Conference on the World Climate Research Programme was that the global capacity to observe the Earth’s climate system is inadequate and deteriorating worldwide and “without action to reverse this decline and develop the GCOS [Global Climate Observing System] , the ability to characterize climate change and variations over the next 25 years will be even less than during the past quarter century” (National Research Council [NRC] 1999). In spite of the United States being a leader in climate research, long term U.S. climate stations have faced challenges with instrument and site changes that impact the continuity of observations over time…

NOAA’s response to the NRC concerns is the USCRN, a network of stations deployed across the continental U.S.

Why CRN is Needed

Where is USCRN data? here.

And what has been the trend at USCRN stations over the last 15 years? (4/2007 – 4/2022)

Air Temperature
$ . crn.sh; crn; plot
Linear Regression Trend: -0.027934 °C/yr

That would be a total cooling of 15 years * -0.027934 °C/year =~ 0.42 °C averaged across all 98 stations that have consistent data going back to 4/2007.

Breakdown by station, in units: °C/year

1                   AK_Fairbanks_11_NE +0.04617
2                        AK_Sitka_1_NE +0.01976
3   AK_Utqiagvik_formerly_Barrow_4_ENE +0.18223
4                      AL_Clanton_2_NE -0.01507
5                   AL_Courtland_2_WSW -0.05665
6                     AL_Cullman_3_ENE -0.03464
7                     AL_Fairhope_3_NE +0.00464
8                      AL_Gadsden_19_N -0.04251
9                  AL_Gainesville_2_NE -0.02576
10                 AL_Greensboro_2_WNW -0.04768
11                  AL_Scottsboro_2_NE -0.01470
12                     AL_Selma_13_WNW -0.03024
13                      AL_Selma_6_SSE -0.03856
14                AL_Valley_Head_1_SSW -0.04865
15                 AR_Batesville_8_WNW -0.05950
16                        AZ_Elgin_5_S -0.01692
17                      AZ_Tucson_11_W -0.01442
18                    CA_Merced_23_WSW +0.03619
19                   CA_Redding_12_WNW +0.03952
20             CA_Stovepipe_Wells_1_SW -0.03849
21                     CO_Boulder_14_W -0.02105
22                      CO_Cortez_8_SE +0.01141
23                     CO_Dinosaur_2_E +0.04307
24                  CO_La_Junta_17_WSW -0.05558
25                  CO_Montrose_11_ENE +0.00016
26                       CO_Nunn_7_NNE -0.06019
27             FL_Everglades_City_5_NE +0.07494
28                   FL_Titusville_7_E +0.07542
29                     GA_Newton_11_SW +0.01076
30                       GA_Newton_8_W -0.01395
31               GA_Watkinsville_5_SSE -0.01863
32                  HI_Mauna_Loa_5_NNE +0.02550
33                       ID_Arco_17_SW -0.04111
34                      ID_Murphy_10_W -0.02431
35                   IL_Champaign_9_SW +0.00057
36                   IL_Shabbona_5_NNE -0.05673
37                  KS_Manhattan_6_SSW -0.10209
38                    KS_Oakley_19_SSW -0.13040
39             KY_Bowling_Green_21_NNE -0.06931
40                 KY_Versailles_3_NNW -0.06288
41                  LA_Lafayette_13_SE -0.01139
42                      LA_Monroe_26_N -0.02333
43                  ME_Limestone_4_NNW -0.10207
44                     ME_Old_Town_2_W -0.05897
45                     MI_Chatham_1_SE -0.10539
46                 MN_Goodridge_12_NNW -0.09950
47               MO_Chillicothe_22_ENE -0.04920
48                     MS_Newton_5_ENE -0.06509
49                   MT_St._Mary_1_SSW -0.06547
50                MT_Wolf_Point_29_ENE -0.00727
51                 MT_Wolf_Point_34_NE -0.08048
52                   NC_Asheville_13_S -0.01759
53                  NC_Asheville_8_SSW -0.01700
54                      NC_Durham_11_W -0.04043
55                       ND_Medora_7_E -0.07155
56                  ND_Northgate_5_ESE -0.15485
57                  NE_Harrison_20_SSE -0.04485
58                    NE_Lincoln_11_SW -0.05038
59                    NE_Lincoln_8_ENE -0.01464
60                    NE_Whitman_5_ENE -0.04729
61                       NH_Durham_2_N -0.05508
62                     NH_Durham_2_SSW -0.04165
63                  NM_Las_Cruces_20_N -0.00039
64                  NM_Los_Alamos_13_W +0.04407
65                     NM_Socorro_20_N -0.03662
66                        NV_Baker_5_W -0.00906
67                    NV_Mercury_3_SSW -0.03112
68                      NY_Ithaca_13_E +0.00221
69                    NY_Millbrook_3_W -0.06797
70                     OK_Goodwell_2_E +0.00749
71                   OK_Stillwater_2_W -0.06988
72                 OK_Stillwater_5_WNW -0.02301
73                       ON_Egbert_1_W -0.07154
74                 OR_Corvallis_10_SSW +0.02164
75                  OR_John_Day_35_WNW -0.01361
76                     OR_Riley_10_WSW -0.00229
77                    RI_Kingston_1_NW -0.03927
78                     RI_Kingston_1_W -0.03514
79                   SC_Blackville_3_W +0.00990
80              SC_McClellanville_7_NE +0.01364
81                   SD_Buffalo_13_ESE -0.03807
82                      SD_Pierre_24_S -0.12013
83               SD_Sioux_Falls_14_NNE -0.06385
84                  TN_Crossville_7_NW -0.04031
85                    TX_Bronte_11_NNE -0.07873
86                  TX_Edinburg_17_NNE +0.00096
87                   TX_Monahans_6_ENE -0.02487
88                    TX_Muleshoe_19_S -0.04374
89                  TX_Palestine_6_WNW -0.07924
90             TX_Panther_Junction_2_N -0.02188
91               VA_Cape_Charles_5_ENE -0.04508
92            VA_Charlottesville_2_SSE -0.03805
93                WA_Darrington_21_NNE +0.04027
94                    WA_Quinault_4_NE +0.05220
95                    WI_Necedah_5_WNW -0.07419
96                    WV_Elkins_21_ENE -0.03204
97                    WY_Lander_11_SSE -0.07578
98                      WY_Moose_1_NNE -0.07829

Only 23 out of 98, that is, ~23.5% show a warming trend.

But there is more!

The nice thing about these stations is that they not only provide air temperature data, but most are also equipped with sensors to measure surface temperature. The results are even more dramatic at the 92 sites with persistent IR surface data:

Surface Temperature
$ . crn.sh; crn; plot
Linear Regression Trend: -0.084043 °C/yr

That would be a total cooling of 15 years * -0.084043 °C/year =~ 1.26 °C averaged across all 92 stations that have consistent surface data going back to 4/2007.

Breakdown by station, in units: °C/year

1                   AK_Fairbanks_11_NE -0.00860
2                        AK_Sitka_1_NE -0.07764
3                     AK_St._Paul_4_NE +0.19834
4   AK_Utqiagvik_formerly_Barrow_4_ENE +0.08741
5                     AL_Fairhope_3_NE -0.07363
6                      AL_Gadsden_19_N +0.03686
7                      AL_Selma_13_WNW +0.09666
8                  AR_Batesville_8_WNW -0.12344
9                         AZ_Elgin_5_S -0.15520
10                      AZ_Tucson_11_W -0.13846
11                    CA_Merced_23_WSW +0.01445
12                   CA_Redding_12_WNW +0.00997
13             CA_Stovepipe_Wells_1_SW -0.15013
14                     CO_Boulder_14_W -0.10934
15                      CO_Cortez_8_SE -0.10363
16                     CO_Dinosaur_2_E -0.09103
17                  CO_La_Junta_17_WSW -0.11102
18                  CO_Montrose_11_ENE -0.16274
19                       CO_Nunn_7_NNE -0.09077
20             FL_Everglades_City_5_NE -0.11284
21                   FL_Titusville_7_E -0.00759
22                     GA_Newton_11_SW -0.00052
23                       GA_Newton_8_W -0.07369
24               GA_Watkinsville_5_SSE -0.12636
25                  HI_Mauna_Loa_5_NNE -0.02257
26                       ID_Arco_17_SW -0.20966
27                      ID_Murphy_10_W -0.16679
28                   IL_Champaign_9_SW +0.03643
29                   IL_Shabbona_5_NNE -0.13466
30                  KS_Manhattan_6_SSW -0.12101
31                    KS_Oakley_19_SSW -0.18244
32             KY_Bowling_Green_21_NNE -0.12606
33                 KY_Versailles_3_NNW -0.06785
34                  LA_Lafayette_13_SE -0.00229
35                      LA_Monroe_26_N -0.13776
36                  ME_Limestone_4_NNW -0.13709
37                     ME_Old_Town_2_W -0.11059
38                     MI_Chatham_1_SE -0.16315
39                 MN_Goodridge_12_NNW -0.10715
40               MO_Chillicothe_22_ENE -0.07844
41                     MS_Newton_5_ENE -0.07688
42                   MT_St._Mary_1_SSW -0.15023
43                MT_Wolf_Point_29_ENE -0.05258
44                 MT_Wolf_Point_34_NE -0.14085
45                   NC_Asheville_13_S -0.12034
46                  NC_Asheville_8_SSW -0.17203
47                      NC_Durham_11_W -0.16045
48                       ND_Medora_7_E -0.05318
49                  ND_Northgate_5_ESE -0.12388
50                  NE_Harrison_20_SSE -0.02705
51                    NE_Lincoln_11_SW -0.09265
52                    NE_Lincoln_8_ENE -0.04629
53                    NE_Whitman_5_ENE -0.10772
54                       NH_Durham_2_N -0.03671
55                     NH_Durham_2_SSW -0.03245
56                  NM_Las_Cruces_20_N -0.06474
57                  NM_Los_Alamos_13_W -0.10742
58                     NM_Socorro_20_N -0.21379
59                        NV_Baker_5_W -0.00542
60                    NV_Mercury_3_SSW -0.06139
61                      NY_Ithaca_13_E +0.00060
62                    NY_Millbrook_3_W -0.09216
63                     OK_Goodwell_2_E -0.06627
64                   OK_Stillwater_2_W -0.15102
65                 OK_Stillwater_5_WNW -0.19385
66                       ON_Egbert_1_W -0.15461
67                 OR_Corvallis_10_SSW +0.02328
68                  OR_John_Day_35_WNW +0.00458
69                     OR_Riley_10_WSW -0.06991
70                     PA_Avondale_2_N -0.08590
71                    RI_Kingston_1_NW -0.03059
72                     RI_Kingston_1_W -0.08983
73                   SC_Blackville_3_W -0.07645
74              SC_McClellanville_7_NE -0.06866
75                   SD_Buffalo_13_ESE -0.10984
76                      SD_Pierre_24_S -0.15606
77               SD_Sioux_Falls_14_NNE -0.11285
78                  TN_Crossville_7_NW -0.11332
79                    TX_Bronte_11_NNE -0.14726
80                  TX_Edinburg_17_NNE -0.03847
81                   TX_Monahans_6_ENE -0.06433
82                    TX_Muleshoe_19_S -0.05690
83                  TX_Palestine_6_WNW -0.00939
84             TX_Panther_Junction_2_N +0.00723
85               VA_Cape_Charles_5_ENE -0.02184
86            VA_Charlottesville_2_SSE -0.07100
87                WA_Darrington_21_NNE -0.08743
88                    WA_Quinault_4_NE -0.19209
89                    WI_Necedah_5_WNW -0.24823
90                    WV_Elkins_21_ENE -0.06492
91                    WY_Lander_11_SSE -0.19794
92                      WY_Moose_1_NNE -0.20405

Only 11 out of 92, that is, ~12% show a warming trend.

Now this is important: I’m not saying that the US has cooled over the last 15 years. All I’m saying is that the newest and best stations suggest this. You decide the significance of this.

Enjoy 🙂 -Zoe

Update 06/07:

If we break up CRN data by years, this is what we get:

$ cat years.csv

0 12.012
1 11.5878
2 11.3553
3 11.889
4 12.9113
5 12.0568
6 11.4294
7 11.9557
8 13.0134
9 13.102
10 12.0355
11 12.2281
12 12.4914
13 12.4444
14 12.541

$ tail -8 year.csv | gmt gmtregress -Fp -o5

-0.00460833333333

Year 0 is 2007/5 to 2008/4 (inclusive), and Year 14 is 2021/5 to 2022/4 (inclusive)

The last 8 years show a definite COOLING trend for air temperature.

$ cat years.csv

0 12.564
1 12.0218
2 11.8927
3 12.6139
4 13.5311
5 12.7203
6 11.7274
7 12.2206
8 13.1326
9 13.2569
10 12.3038
11 12.3714
12 12.4584
13 12.7154
14 12.6693

$ tail -12 years.csv | gmt gmtregress -Fp -o5

-0.0173821678322

The last 12 years show a definite COOLING trend for surface temperature.

Code:

# Zoe Phin, 2022/06/04
# File: crn.sh
# Run: . crn.sh; require; download; crn; plot; sta

require() { sudo apt-get install -y gmt gnuplot; }
download() {
    wget -O crn.zip https://www.ncei.noaa.gov/pub/data/uscrn/products/monthly01/snapshots/CRNM0102202205300730.zip
    unzip crn.zip
}
crn() {
    for sta in $(ls -1 CRN*); do
        [[ $(egrep '200704|202204' $sta | awk '{print $9}' | grep -v 9999 | wc -l) -lt 2 ]] && continue
        tmp=${sta##CRNM0102-}; echo -n "${tmp%%.txt} "
        awk '$2>=200704 && $2<=202204 { printf "%.2f ",$9 }' $sta 
        echo
    done | awk '{ 
        for (m=2;m<=NF;m++) { if ($m!="-9999.00") { T[m-2]+=$m; N[m-2]+=1 } }
    } END { 
        for (m in T) print 2007+4/12+m/12-1/24" "T[m]/N[m] 
    }' > crn.csv
}
plot() { 
    cat crn.csv | gmt gmtregress > plot.mo
    echo "set term png size 740,470
        set key out top horizontal
        set yrange[-4:28]; set xrange[2007:2022.8]
        set ylabel 'Surface Temperature (°C)'
        set xtics 5; set mxtics 5; set mytics 5
        set grid xtics mxtics ytics
        plot 'plot.mo' u 1:2 w lines t 'Monthly' lw 1 lc rgb 'black',\\
             'plot.mo' u 1:3 w lines t 'Trend'   lw 2 lc rgb 'blue'
    " | gnuplot > crn.png
    echo -n "Linear Regression Trend: "
    cat crn.csv | gmt gmtregress -Fp -o5 | awk '{printf "%.6f °C/yr\n",$1}'
}
sta() {
    let n=0
    for sta in $(ls -1 CRN*); do
        [[ $(egrep '200704|202204' $sta | awk '{print $9}' | grep -v 9999 | wc -l) -lt 2 ]] && continue
        tmp=${sta##CRNM0102-}; echo -n "$((++n)) ${tmp%%.txt} "
        awk '$2>=200704 && $2<=202204 && $9!~/9999/ { y=substr($2,0,4); m=substr($2,5,2);
            printf "%4.2f %4s\n",y+m/12-1/24,$9 
        }' $sta | gmt gmtregress -Fp -o5
    done | awk '{
        printf "%-3d %34s %+8.5f\n", $1, $2, $3
    }' | tee sta.csv
    echo -n "Avg: "
    cat sta.csv | awk '{S+=$3}END{print S/NR}'	
}

Published by Zoe Phin

https://phzoe.com

20 thoughts on “USCRN Cooling

      1. Thanks. Good info.
        Wasn’t suggesting that you re-check an average, just making sure I more or less understood the graph.
        I was surprised (from the notes) that they had some “corrected” readings in with the other data.

        Like

  1. Zoe, as usual, an interesting analysis.

    I got the data you used and tried to replicate your data. Unfortunately, I was unable to do so.

    In all cases, the trends I get are much more positive than those that you got.

    I suspect that this is because you’ve taken the trend of the data itself. However, because each dataset starts and ends in April, the data is high at the start and low at the end … as in your Figure 1. It’s a problem with all cyclical data. The apparent trend depends sensitively on where you start and stop.

    In your case, this has led to incorrectly low trends. The only way to get the real trend is to remove the seasonality first.

    Let me invite you to re-calculate the trends after removing the monthly averages (seasonality), and let us know what you get.

    My best regards to you,

    w.

    Liked by 1 person

    1. I anticipated this. It DOES depend on where you start the year, but guess what, nature doesn’t care about ancient Romans’ choice of when to start the year. In fact, it makes most sense to start at NH vernal equinox (Mar ~21). That would be April, as the next full thing.

      The problem with removing seasonality is that you destroy data. The most obvious destruction is volitility, and the distribution of extremes.

      The most important thing is starting and ending at the same time of year. That’s it. Cyclicality then doesn’t matter. But if you don’t like the outcome, you can complain about it. My goal was to see if it warmed over time, not if the years warmed over time. Subtle but big difference.

      Either way, there is cooling in CRN over the last 6-8 years any which way you slice the data.

      Thanks Willis
      -z

      Liked by 1 person

  2. Zoe, you say:

    “The most important thing is starting and ending at the same time of year. That’s it. Cyclicality then doesn’t matter.”

    I’m sorry, but that’s simply not true.

    Here’s an example. This is a section of a trendless sine wave.

    I’m sure you can see the problem. As with your data, the sine wave is high at the start and low at the end, so it looks like the trend is downwards … when in fact the sine wave has no trend at all.

    My best to you as always,

    w.

    Liked by 1 person

    1. What is this? Sin(pi/15)?

      The distorted slope is -0.0039 here … for 4 cycles. It’s -0.00028 for 15 cycles.

      You’re correct. We should aim for zero, but -0.00028 is tiny.

      Now here’s the fun part … you can’t make any linear extrapolation in a chaotic system you don’t understand. What if you’re inside a sin curve you don’t know about?

      – z

      Liked by 1 person

      1. Zoe, if you think the incorrect slope in the figure I posted is “tiny”, you need glasses.

        And even in longer datasets, the difference is quite large. Not only that, but it can convert an actual increase into a decrease or vice versa.

        Look, the simple fact is that YOU DID THE MATH WRONG! And you exposed your misunderstanding by claiming that:

        “The most important thing is starting and ending at the same time of year. That’s it. Cyclicality then doesn’t matter.

        That is simply not true, as my graph above clearly demonstrates. Not true in the slightest.

        You also say:

        “The distorted slope is -0.0039 here … for 4 cycles. It’s -0.00028 for 15 cycles.”

        That’s not true even by eye. Look at the graph. Over the four cycles shown in the graph, the total drop is from about +0.25 to -0.25, which is about -0.5 in four cycles. NOT -0.0039 in four cycles as you claim.

        In fact, with a peak-to-trough amplitude of 2.0, as shown in my graph, the “distorted slope” is -0.116 PER CYCLE, which is -0.46 over four cycles. Note that that actual calculation agrees with my Mark 1 Eyeball estimate.

        So with a peak-to-trough amplitude of 25°C as in your graph, and 15 cycles, the error would be on the order of

        0.116 / 15 * 25 = -0.12°C

        Not a small error at all. A very large error.

        Stop trying to defend the indefensible, and come back when you’ve recalculated the CRN results AFTER removing the cyclical part of the signal. You’re a good scientist with a most keen and inquisitive mind, but your extreme unwillingness to admit when you are wrong is holding you back greatly.

        In friendship,

        w.

        Like

        1. OK, Willis, 0.42C – 0.12C = 0.3C Cooling

          You win. ~1/3rd error accounted the way you wanted without any checking.

          My slopes were calculated per unit, not per cycle. 0->120.

          I don’t want to get rid of cyclicality.

          Here’s a set of half year T’s for 3 years:

          (2,2),(2,2),(3,1)

          You say there is no trend, because the averaged year eliminates the trend. That’s fine. But I consider the last 6 months to be part of a sequential timeline, and therefore there was an overall cooling trend. This is more intuitive to an average person: last 6 months were colder compared to last year and 3 years, so obviously there was cooling.

          So the best thing to do is keep the cyclicality and adjust for error.

          Like

  3. I hate to disagree with Mrs. Smarty Pants,
    and I would never disagree with Willie E.
    I would call the USCRN trend from 2005 through 2021
    to be a flat trend. So far 2022 has not been warm enough
    for those of us who live in Michgan — we don’t need
    official data to know that — and we want our
    global warming back!

    The usual arguments:
    The US 48 states is perhaps only 1.5%
    of Earth’s surface, and this is not a 30+
    year climate trend.

    But, on the other hand, 330 million people
    are affected by this US climate, while not one
    person lives in the global average temperature.

    I recently went to the NOAA website to get
    the latest USCRN chart, and couldn’t find it.
    That tells me the data must be contradicting
    the official glo-baloney warming narrative.
    Or perhaps I need new reading glasses.

    Like

  4. The correct way to address the issue is to first deseasonalize the data. That is, take the 12th difference of the monthly data series. It drives me nuts that climate scientists use an arbitrary 30 window to calculate “anomalies.” Once you have the deseasonalized data, regress this data against a time trend. You will find that almost none of the stations have a statistically significant trend in temperature. I have a subset of stations from the CRN universe that I have been following for years, and none of them have significant changes in temperature. The 7 CA stations are worth a look. Not much temperature change in any of them. If you throw in monthly levels of CO2, there are some straightforward cross-sectional/time series methods to test for the significance of CO2.

    Liked by 1 person

      1. Zoe, By deseasonalizing, I mean subtracting from the current month’s temperature the temperature from the same month a year earlier. Yes, you lose the first 12 months of data in creating the deseasonalized series. Take the differenced series and regress it on a time trend (1,2,3,…..for as many months as you have data.) You want to test for the significance of the regression coef on the time trend variable. So regress Deseasonalized temp = Const + B(time trend series) + e. The B coef is positive and significant according to alarmists. There are much more sophisticated ways to address the issue of trending temperatures. Alarmists claim that temperature data is nonstationary, that is, the mean temperature is increasing. The CRN data shows that this claim is false for the US land temperature since about 2005. Why the climate industry ignores basic time series analysis is beyond me. Using a reference period is just wrong. The issue Willis raised about starting and stopping points disappear once you deseasonalize the data.

        Liked by 1 person

  5. Regions smaller than the entire globe of course do not match the global average temperature. It is more difficult to project future temperatures, the smaller the region is. The United States is a tiny sliver of the world, even when you include Alaska (i.e., continental U.S. rather than just contiguous U.S.). Nobody projects that all regions of the world will warm at the same rate. For explanation of U.S. temperature projections, see https://nca2018.globalchange.gov/chapter/2#key-message-5

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: