What?
Now that we have HadISDH.landq Its time to produce a land gridded product for the other humidity variables: vapour pressure (e), dewpoint temperature (Td), wetbulb temperature (Tw) and relative humidity (RH). Temperature can also be produced in the process.
This poses the problem of what to do about homogenisation. The simplest thing to do is to run the PHA, that was used for HadISDH.landq, on each of the variable monthly mean datasets. However, on the station or perhaps even the gridbox level this is likely to lead to occurrences of unphysical changes made across the variables. For example, a moist bias may be detected in RH and reduced. This should either come from a moist bias in Td and/or a cool bias in T. Although PHA is good we know that there are occurrences of false alarms or missed breakpoints and that the location or magnitude of a break may not be accurately estimated. So, when applied to a number of variables in isolation things like an increase in T could occur along side a decrease in Td and an increase in RH which would not be physically possible.
To avoid this, one option would be to take the break locations of the original variables provided by ISD (T and Td) and apply adjustments to e, q, RH and Tw with the size of those adjustments estimated from neighbour networks of that variable.
As discussed in a previous post on spatial correlation structures, neighbour networks are different for each variable. So, this more involved method still may lead to issues. However, this can be post-checked. Ultimately, a comparison of data-products created using both methods may provide some estimate of uncertainty.
Here I look at each variable homogenised using PHA seperately. This provides both insight into the sensitivity of each variable to PHA and also into whether Td is indeed the preferred humidity variable to homogenise.
Summary
Tw has the most number of stations after PHA has been applied. This may make it favourable for break point allocation (3551 for Tw verses 3490 for Td).
Td tends to have larger adjustments than Tw and slightly larger changes between the raw and homogenised datasets. This may imply that Td is more sensitive to inhomogeneity which may make it optimal from an ability to detect inhomogeneity point of view compared to Tw. However, this may also mean that Td suffers more false alarms which would make it less desirable as a breakpoint location indicator.
Td matches q and e to a large extent in the changes between the raw and homogenised dataset whereas Tw matches T.
Tw and Td show greatest moistening in the Northern Hemisphere where as q and e show that moistening is largest in the Tropics. Tw and Td measures will be directly influenced by rising T, which is greatest in the Northern Hemisphere, whereas q and e will not be. RH shows the greatest drying signal in the Northern Hemisphere - where T warming trends are largest.
In all cases, homogenisation has reduced the range of trends.
Homogenisation has had very little effect on global mean trends and little effect on Northern Hemisphere mean or tropical mean trends. The Southern Hemisphere is more susceptible to variation.
In all cases, adjustments are close to symmetric about zero, with a perceptible missing middle. T, Tw, Td, e and q have incurred slightly more negative adjustments on average. RH has incurred slightly more positive adjustments.
There are more inhomogeneities detected in RH than for any other variable. T has the least number of detected inhomogeneities, closely followed by Tw, then Td and then jointly by e and q. Strictly speaking, RH is the only variable that is derived from both T and Td (or Tw) and so may be expected to contain more breakpoints. However, in reality it can now be measured directly without any input from a T instrument. Or, it is derived from Td and T. Or, Td is actually derived from RH and T and therefore Td, Tw, e and q will all contain inhomogeneities from both T and Td/RH.
Correlation decay distances appear to be greatest for Tw, then T, followed by Td, e and q and then RH. Station autocorrelation is similar across all variables - highest in the Tropics. Station standard deviations are reasonably tightly constrained for T and Tw (Figures 13 and 14), increasing towards the poles. RH, e and q shows greatest variables around 30 deg N/S - the extratropics (Figures 15, 16, and 17). Td is almost a cross between Tw/T and e/q/RH (Figure 18). The standard deviation of station RH is indeed large as is the spread across stations - RH has the highest number of breakpoints detected. The station standard deviations for T and Tw are slightly lower than for Td and the spread of T and Tw station standard deviations is much tighter. More breaks are detected in Td than for T and Tw. It may of course be the case that there are more false alarms for RH because it is a highly variable variable?
Temperature (T) - Figures 1 and 2
Number of stations passing PHA: 3691
Stations without sufficient highly correlating neighbours: 3
WMO/WBAN ID, latitude, longitude, elevation, country, name
85469099999 -27.17 -109.43 69 CH ISLA DE PASCUA
91066022701 28.20 -177.38 4 US MIDWAY ISLAND NAS
91925099999 -9.80 -139.03 53 PF ATUONA
Number of stations passing netCDF anomaly calculation criteria - sufficient months present to calculate a climatology over the 1976-2005 period:
3571 - no stations removed due to supersaturation or subzero RH.
Number of stations (and list) failing netCDF anomaly calculation criteria - insufficient months present to calculate a climatology over the 1976-2005 period:
120 insufficient month stations
033470,040650,080480,082320,124880,134650,165220,229390,270080,303790,306370,307120,
307290,382220,400010,400090,400160,400390,400610,415300,415940,425590,433330,444540,
485010,488700,488870,505480,610170,612570,612700,612770,612930,612970,614920,614970,
620560,621610,624170,636410,637910,645010,645510,648600,677790,678530,678810,679690,
722026,722038,722108,722730,722789,722867,722925,722955,723536,723825,723927,723930,
724466,724721,724800,724936,724946,725245,725865,725946,727436,727834,727934,743945,
744900,764050,764230,765560,783250,784820,787170,787670,789580,800970,804070,804130,
804190,804210,804270,812000,820220,820980,829150,835760,841170,844250,845310,860110,
860680,862600,863700,864900,865300,870160,873200,873280,875930,876400,877840,889630,
890660,912210,941860,961630,962210,964810,967490,968390,972600,981350,984250,984300
10 stations with the largest adjustments (deg C - relative to the next most recent homogeneous subperiod):
71496399999 6.82
02672099999 -5.88
71826099999 -5.49
53588099999 4.91
71844099999 4.67
72378399999 -4.42
02326099999 -4.35
59948099999 -4.15
76648099999 -3.76
02336099999 -3.52
Average number of adjustments per station: 1.3
Average size of adjustment: -0.11 deg C
1 Standard Deviation of adjustment size: 0.89 deg C
Slight bias in adjustments lowering periods relative to the most recent period.
Large scale decadal trends are changed as follows:
Globe RAW = 0.24 (0.19 to 0.30) deg C/decade
Globe HOMOG = 0.24 (0.19 to 0.29) deg C/decade
N. Hemisphere RAW = 0.30 (0.22 to 0.37) deg C/decade
N. Hemisphere HOMOG = 0.30 (0.22 to 0.37) deg C/decade
Tropics RAW = 0.19 (0.13 to 0.24) deg C/decade
Tropics HOMOG = 0.16 (0.11 to 0.21) deg C/decade
S. Hemisphere RAW = 0.11 (0.07 to 0.14) deg C/decade
S. Hemisphere HOMOG = 0.11 (0.07 to 0.14) deg C/decade
There is no change except for a small decrease in warming trends over the Tropics.
All homogenised large scale trends show significant warming which is greatest in the Northern Hemisphere and smallest in the Southern Hemisphere.
At the gridbox level almost all gridboxes (97%) do not change trend direction when the data are homogenised. The overall range of trends across the gridboxes reduces when using the homogenised data.
Dewpoint Temperature (Td) - Figures 3 and 4
Number of stations passing PHA: 3691
Stations without sufficient highly correlating neighbours: 3
WMO/WBAN ID, latitude, longitude, elevation, country, name
08501099999 39.45 -31.13 29 PO FLORES (ACORES)
68906099999 -40.35 -9.88 54 ZA GOUGH ISLAND
89571099999 -68.58 77.95 13 AY DAVIS
Number of stations passing netCDF anomaly calculation criteria - sufficient months present to calculate a climatology over the 1976-2005 period:
3467 (this has taken out any stations suffering from RH <0%rh (none) or >100%rh (23) as a result of homogenisation - using RH directly)
Number of stations (and list) failing netCDF anomaly calculation criteria - insufficient months present to calculate a climatology over the 1976-2005 period:
201 insufficient month stations
011210,012280,014670,030270,031380,074280,077490,080480,080530,081300,082320,082720,
111820,112140,121250,131730,134650,140240,161140,165220,172340,172900,229390,270080,
294560,294770,303560,303790,306370,306640,307120,307290,309150,309490,370010,400090,
400160,400610,400950,401000,402650,403400,407450,408090,415940,417640,419780,424520,
433330,442150,442980,443140,443470,444540,471300,471360,485010,485650,488870,505480,
600010,600960,603200,603400,603510,605060,606020,607280,610170,612300,612700,612960,
614010,614210,614920,614970,619960,620560,621310,621610,622590,622710,623320,624350,
637910,648600,672370,678430,678530,678610,679690,680240,683280,685120,722029,722038,
722119,722533,722598,722599,722673,722789,722867,722885,722897,722925,722955,723536,
723783,723825,723930,724035,724094,724097,724336,724466,724721,724735,724765,724800,
724936,724946,725128,725314,725705,725715,725865,725955,726515,726796,727436,727834,
727928,743700,744865,744900,764050,764580,765710,769043,784570,784600,784820,785350,
785470,787080,787110,787170,787670,788580,789580,804070,804100,804130,804190,804210,
804270,804280,804380,804470,812000,829150,835760,841170,845310,845640,860110,860680,
862600,870160,873200,875930,876400,879380,889630,890570,890660,912210,915200,941860,
946340,960350,961630,962210,963150,964810,967490,968390,969960,972600,981350,983290,
984250,984270,984290,984300,984400,984460,985480,986420,987480
23 supersaturated stations
061200,064760,110220,111300,151080,200870,202920,219820,230320,319600,547760,577760,
584370,586660,700260,700860,703080,718936,724837,725975,745160,857990,870970
10 Stations with the largest adjustments (deg C - relative to the next most recent homogeneous subperiod):
62103099999 -6.77
55664099999 -6.43
02672099999 -6.11
71496399999 5.98
62103099999 5.94
71826099999 -5.62
44373099999 5.52
44284099999 -5.46
02326099999 -5.33
41240099999 -5.30
Average number of adjustments per station: 2
Average size of adjustment: -0.05 deg C
1 Standard Deviation of adjustment size: 1.19 deg C
Slight bias in adjustments lowering records relative to the most recent period.
Large scale decadal trends are changed as follows:
Globe RAW = 0.16 (0.09 to 0.22) deg C/decade
Globe HOMOG = 0.17 (0.12 to 0.23) deg C/decade
N. Hemisphere RAW = 0.23 (0.14 to 0.31) deg C/decade
N. Hemisphere HOMOG = 0.24 (0.15 to 0.31) deg C/decade
Tropics RAW = 0.14 (0.10 to 0.20) deg C/decade
Tropics HOMOG = 0.14 (0.11 to 0.19) deg C/decade
S. Hemisphere RAW = -0.11 (-0.15 to -0.04) deg C/decade
S. Hemisphere HOMOG = -0.01 (-0.06 to 0.04) deg C/decade
There are tiny increases in the moistening trend for Globe and Northern Hemisphere, no change for Tropics and a change from significantly decreasing (drying) to no significant change for the Southern Hemisphere.
Homogenised large scale trends for the Globe, Northern Hemisphere and Tropics show significant moistening which is greatest in the Northern Hemisphere. There is no discernible trend in the Southern Hemisphere.
At the gridbox level the majority of gridboxes (87%) do not change trend direction when the data are homogenised. The overall range of trends across the gridboxes reduces when using the homogenised data.
Wetbulb Temperature (Tw) - Figures 5 & 6
Number of stations passing PHA: 3691
Stations without sufficient highly correlating neighbours:
3
WMO/WBAN ID, latitude, longitude, elevation, country, name
08501099999 39.45 -31.13 29 PO FLORES (ACORES)
68906099999 -40.35 -9.88 54 ZA GOUGH ISLAND
85469099999 -27.17 -109.43 69 CH ISLA DE PASCUA
Number of stations passing netCDF anomaly calculation criteria - sufficient months present to calculate a climatology over the 1976-2005 period:
3526 (this has taken out any stations suffering from RH <0%rh (none) or >100%rh (25) as a result of homogenisation - using RH directly)
Number of stations (and list) failing netCDF anomaly calculation criteria - insufficient months present to calculate a climatology over the 1976-2005 period:
140 insufficient month stations
034925,040650,040890,080480,082320,083830,134650,165220,229390,270080,303790,306370,
307120,307290,376820,400010,400160,400610,401550,402720,412400,415940,417640,425590,
433330,485010,505480,597920,600010,600600,602850,603400,603510,610170,612570,612700,
612770,612970,614010,614920,614970,620560,621610,623870,637910,646500,648600,672970,
677790,678430,678530,683280,722038,722119,722598,722786,722897,722903,722976,722977,
723069,723825,723927,723930,724097,724466,724666,724721,724800,724835,724946,725128,
725715,725865,725946,726548,727436,727834,727928,727934,744900,764050,764230,765710,
769043,783880,784820,785140,785350,785430,787110,789580,789880,800090,804070,804100,
804130,804190,804210,804270,804380,804470,812000,812020,820980,828250,835760,841170,
844010,845310,860110,860680,862600,870160,873200,874160,875930,876400,879340,879380,
889630,890660,911860,912210,940350,960350,961090,962210,963150,964810,967490,968050,
968390,969960,972600,983290,984270,984290,984460,987480
25 supersaturated months
061200,064760,110220,110280,111300,151080,200870,202920,219820,230320,319600,470610,
547760,577760,584370,586660,700260,700860,703080,718936,724837,725975,745160,857990,
984400
10 stations with the largest adjustments (deg C - relative to the next most recent homogeneous subperiod):
71496399999 6.53
02672099999 -5.64
71826099999 -5.12
02326099999 -4.26
53588099999 4.05
71844099999 3.68
02336099999 -3.51
02286099999 -3.39
72378399999 -3.26
16710099999 3.05
Average number of adjustments per station: 1.5
Average size of adjustment: -0.07 deg C
1 Standard Deviation of adjustment size: 0.69 deg C
Slight bias in adjustments lowering records relative to the most recent period.
Large scale decadal trends are changed as follows:
Globe RAW = 0.19 (0.14 to 0.23) deg C/decade
Globe HOMOG = 0.20 (0.15 to 0.24) deg C/decade
N. Hemisphere RAW = 0.24 (0.17 to 0.31) deg C/decade
N. Hemisphere HOMOG = 0.25 (0.18 to 0.32) deg C/decade
Tropics RAW = 0.15 (0.10 to 0.20) deg C/decade
Tropics HOMOG = 0.15 (0.11 to 0.19) deg C/decade
S. Hemisphere RAW = 0.03 (0.01 to 0.06) deg C/decade
S. Hemisphere HOMOG = 0.03 (0.00 to 0.06) deg C/decade
There are tiny increases in the moistening trend for the Globe and Northern Hemisphere and no change for Tropics or Southern Hemisphere.
All homogenised large scale trends show moistening (debatable for the Southern Hemisphere) which is largest in the Northern Hemisphere and smallest in the Southern Hemispere.
At the gridbox level the vast majority of gridboxes (93%) do not change trend direction when the data are homogenised. The overall range of trends across the gridboxes reduces when using the homogenised data.
Vapour Pressure (e) - Figures 7 & 8
Number of stations passing PHA: 3691
Stations without sufficient highly correlating neighbours:
3
WMO/WBAN ID, latitude, longitude, elevation, country, name
08501099999 39.45 -31.13 29 PO FLORES (ACORES)
68906099999 -40.35 -9.88 54 ZA GOUGH ISLAND
89571099999 -68.58 77.95 13 AY DAVIS
Number of stations passing netCDF anomaly calculation criteria - sufficient months present to calculate a climatology over the 1976-2005 period:
3488 (this has taken out any stations suffering from RH <0%rh (none) or >100%rh (26) as a result of homogenisation - using RH directly)
Number of stations (and list) failing netCDF anomaly calculation criteria - insufficient months present to calculate a climatology over the 1976-2005 period:
177 insufficient month stations
012280,026220,031380,040890,063300,074280,077490,080480,082320,083830,111820,121250,
125500,131730,135780,140240,165220,166840,172900,201070,222920,229390,270080,294560,
299620,303560,303790,306370,306640,307120,307290,362080,370010,377350,383530,400160,
400610,402720,412400,412880,415940,417640,433330,442150,442560,443470,443730,444540,
484770,485010,485650,600010,600960,605060,610170,612300,612570,612700,612720,612960,
612970,614010,614920,614970,620560,621610,622590,623060,624350,637910,648600,672370,
672970,678530,679690,681100,681480,683280,685120,702986,711430,722038,722108,722429,
722598,722673,722867,722903,723536,723783,723930,724094,724097,724288,724800,724935,
724936,725865,726375,726515,726548,726796,727436,727458,727890,727928,744865,744900,
763503,764050,764230,765710,767755,769043,783670,784820,785140,785350,785830,787080,
787110,787670,789580,789880,800090,800970,801120,804070,804100,804130,804190,804210,
804270,804380,804470,812000,820220,829150,835760,844250,845310,845640,860110,860680,
862600,864300,870160,875930,876400,879340,879380,889630,890660,912210,915200,916970,
916990,960350,961630,962210,963150,964650,964810,967490,968390,969350,969960,971800,
972600,983290,984250,984280,984290,984300,984460,985480,987480
26 supersaturated stations
061200,064760,110220,110280,111300,151080,200870,202920,219820,230320,319600,470610,
547760,577760,584370,586660,700260,700860,703080,718936,724837,725975,745160,857990,
870970,984400
10 stations with the largest adjustments (deg C - relative to the next most recent homogeneous subperiod):
48456099999 -5.66
40340099999 -5.36
82281099999 -5.09
41268099999 -4.71
91691099999 -4.52
82281099999 -4.52
43117099999 -4.45
78990099999 4.38
78388099999 -4.29
76577099999 -4.29
Average number of adjustments per station: 2
Average size of adjustment: -0.03hPa
1 Standard Deviation of adjustment size: 0.87 hPa
Slight bias in adjustments lowering records relative to the most recent period.
Large scale decadal trends are changed as follows:
Globe RAW = 0.14 (0.10 to 0.17) deg C/decade
Globe HOMOG = 0.13 (0.10 to 0.17) deg C/decade
N. Hemisphere RAW = 0.13 (0.09 to 0.16) deg C/decade
N. Hemisphere HOMOG = 0.13 (0.09 to 0.16) deg C/decade
Tropics RAW = 0.23 (0.17 to 0.31) deg C/decade
Tropics HOMOG = 0.19 (0.14 to 0.26) deg C/decade
S. Hemisphere RAW = -0.05 (-0.09 to 0.00) deg C/decade
S. Hemisphere HOMOG = -0.00 (-0.04 to 0.04) deg C/decade
There are tiny decreases in the moistening trend for the Globe and a larger decrease for the Tropics. There is no change in the moistening trend for the Northern Hemisphere and no real change for the Southern Hemisphere in terms of detection of a significant trend although the rate has increased from negative to zero.
Homogenised large scale trends show moistening for the Globe, Northern Hemisphere and especially the Tropics. There is no discernible trend in the Southern Hemisphere.
At the gridbox level the majority of gridboxes (89%) do not change trend direction when the data are homogenised. The overall range of trends across the gridboxes reduces when using the homogenised data.
Relative Humidity (RH) just for a reminder - Figures 9 & 10
Number of stations passing PHA: 3683
Stations without sufficient highly correlating neighbours:
11
WMO/WBAN ID, latitude, longitude, elevation, country, name
68104099999 -22.88 14.43 0 NM WALVIS BAY (PELICAN
68994099999 -46.88 37.87 21 ZA MARION ISLAND
84782099999 -18.05 -70.27 458 PR TACNA
85469099999 -27.17 -109.43 69 CH ISLA DE PASCUA
85470099999 -27.30 -70.42 290 CH COPIAPO
85488099999 -29.92 -71.20 146 CH LA SERENA
85930099999 -52.40 -75.10 52 CH FARO EVANGELISTAS
89022099999 -75.50 -26.65 30 AY HALLEY
91610099999 1.35 172.92 4 KB TARAWA
91643080705 -8.53 179.22 2 TV FUNAFUTI NF
91943099999 -14.48 -145.03 3 PF TAKAROA
Number of stations passing netCDF anomaly calculation criteria - sufficient months present to calculate a climatology over the 1976-2005 period:
3412 (this has taken out any stations suffering from RH <0%rh (none) or >100%rh (26) as a result of homogenisation - using RH directly)
Number of stations (and list) failing netCDF anomaly calculation criteria - insufficient months present to calculate a climatology over the 1976-2005 period:
245 insufficient month stations
010470,011060,011150,011940,012120,
012280,020550,026220,026300,031540,032080,032620,033140,033470,034925,038740,038800,
039670,060960,062470,064960,074280,080480,082320,082720,083830,105140,111380,111820,
111900,112140,121250,122150,122300,122850,124550,125500,126700,131730,133480,133690,
134650,140240,161140,165220,166840,170700,170840,222820,222920,229390,270080,277190,
280440,284650,292530,294180,294560,294770,295760,295870,303560,303790,305260,306370,
306640,307120,307290,309150,312630,353020,376820,382220,400010,400090,400160,400390,
400610,402650,403400,407060,407450,408090,415300,415940,419780,420270,425590,433330,
442150,442390,442560,443140,443470,471620,471920,484770,485650,505480,547510,597920,
600600,600960,605060,606020,606110,607280,610170,612300,612700,612960,614010,614920,
614970,620560,621610,636410,637910,648600,677810,678430,678530,678610,679690,683280,
684610,702315,702320,702350,702610,702757,711430,712220,716100,718260,718420,718450,
722029,722038,722119,722345,722348,722429,722533,722598,722599,722673,722867,722897,
722903,722925,722955,723069,723119,723307,723345,723536,723555,723815,723825,723930,
724035,724067,724094,724097,724288,724466,724506,724721,724800,724936,725128,725245,
725314,725396,725524,725715,725955,726375,726515,726796,727344,727436,727458,727834,
727928,727934,744865,744900,745090,762555,762863,763503,763820,764050,764230,764915,
765560,766540,767755,780620,784600,784820,785350,785830,787110,787170,787410,789580,
802100,804070,804100,804190,804270,812000,844250,860110,860680,862330,862600,862970,
864300,865300,870160,870780,873200,875930,876400,876480,878520,889630,890560,890590,
890660,895710,912210,915820,964810,969350,972600,983290,984280,984460,985480,987410
26 supersaturated RH stations
061200,064760,110220,110280,111300,151080,200870,202920,219820,230320,319600,470610,
547760,577760,584370,586660,700260,700860,703080,718936,724837,725975,745160,857990,
870970,984400
10 Stations with the largest adjustments (%rh - relative to the next most recent homogeneous subperiod):
04231099999 26.3100
36982099999 -25.7900
44284099999 -22.9400
44277099999 -22.5800
72654899999 22.2500
44287099999 -20.6400
44213099999 -18.8500
36982099999 17.9600
38545099999 -17.9300
04260099999 -17.2500
Average number of adjustments per station: 2.6
Average size of adjustment: 0.1 %rh
1 Standard Deviation of adjustment size: 4.60%rh
Slight bias in adjustments raising records relative to the most recent period.
Large scale decadal trends are changed as follows:
Globe RAW = -0.19 (-0.28 to -0.07) deg C/decade
Globe HOMOG = -0.13 (-0.22 to -0.02) deg C/decade
N. Hemisphere RAW = -0.11 (-0.28 to 0.04) deg C/decade
N. Hemisphere HOMOG = -0.21 (-0.34 to -0.07) deg C/decade
Tropics RAW = -0.17 (-0.25 to -0.10) deg C/decade
Tropics HOMOG = 0.03 (-0.07 to 0.12) deg C/decade
S. Hemisphere RAW = -0.64 (-0.87 to -0.42) deg C/decade
S. Hemisphere HOMOG = -0.14 (-0.31 to 0.03) deg C/decade
Considerable changes are observed. There is a small decrease in the drying trend for the Globe and a much larger decrease in drying trends for the Tropics and Southern Hemisphere such that they are no longer significant. There is an increase in the drying trend for the Northern Hemisphere however which now becomes significant.
Homogenised large scale trends for the Globe and especially the Northern Hemisphere show significant drying. Trends in the Southern Hemisphere and Tropics are not significant although recent (post 2000) drying is suggested in the Southern Hemisphere.
At the gridbox level many gridboxes (32%) change trend direction when the data are homogenised. The overall range of trends across the gridboxes reduces when using the homogenised data.
Specific Humidity (q) just for a reminder- Figures 11 & 12
Number of stations passing PHA: 3691
Stations without sufficient highly correlating neighbours:
3
WMO/WBAN ID, latitude, longitude, elevation, country, name
08501099999 39.45 -31.13 29 PO FLORES (ACORES)
68906099999 -40.35 -9.88 54 ZA GOUGH ISLAND
89571099999 -68.58 77.95 13 AY DAVIS
Number of stations passing netCDF anomaly calculation criteria - sufficient months present to calculate a climatology over the 1976-2005 period:
3456 (this has taken out any stations suffering from RH <0%rh (51) but not >100%rh as a result of homogenisation - using q<0 and q>qsat(unhomogenise) - not ideal for >100%rh as qsat may have been adjusted too had it been homogenised - change to T)
Number of stations (and list) failing netCDF anomaly calculation criteria - insufficient months present to calculate a climatology over the 1976-2005 period:
184 insufficient month stations
012280,024040,026220,031380,040560,040890,062470,063300,074280,077490,080480.082320,
111820,121250,131730,133890,135780,140240,165220,166840,172900,222920,229390,270080,
303560,303790,306370,307120,307290,369740,370010,376820,383530,400160,400610,402720,
408090,412400,415940,417640,433330,435550,442150,442560,443470,444540,471920,484770,
485010,485650,600010,600960,605060,610170,612300,612700,612720,612970,614010,614920,
614970,620560,621610,622590,622710,623370,624320,624350,637910,648600,672370,678530,
679690,680320,683280,685120,702070,702986,711974,722038,722108,722429,722598,722673,
722786,722867,723536,723783,723930,724094,724097,724288,724721,724765,724800,724935,
725715,725865,725946,726375,726515,726548,726796,727436,727458,727834,727890,727928,
744865,744900,763820,764050,764230,765390,765770,769043,784820,784850,785140,785350,
785830,787080,787110,787670,788580,789540,789580,789880,800940,800970,801120,804070,
804100,804130,804190,804210,804270,804380,804470,812000,820220,828250,829150,835760,
844250,845310,845640,860110,860680,862600,870160,875930,876400,879340,879380,889630,
890660,912210,915200,916990,944300,960350,961630,962210,963150,964650,964810,967490,
968390,969350,969960,971800,972600,983290,984250,984270,984280,984290,984300,984400,
984460,985480,986420,987480
51 subzero RH stations
022860,023360,042020,042300,042310,246880,249620,254000,256210,296760,307580,307810,
308790,309610,310540,314740,359690,384030,442130,442140,442180,442370,442390,442410,
442650,442770,442840,442870,442880,443050,443360,443410,443520,443540,443580,443730,
552990,554720,555780,702315,702320,710750,714800,714963,718260,719170,719330,719430,
726487,726555,740820
10 stations with the largest adjustments (g/kg - relative to the next most recent homogeneous subperiod):
48456099999 -3.57
40340099999 -3.43
82281099999 -3.20
41268099999 -3.15
78367011706 3.14
43117099999 -3.09
91691099999 -2.89
60765099999 -2.79
78388099999 -2.69
61421099999 2.67
Average number of adjustments per station: 2
Average size of adjustment: -0.02 g/kg
1 Standard Deviation of adjustment size: 0.55 g/kg
Slight bias in adjustments lowering records relative to the most recent period.
Large scale decadal trends are changed as follows:
Globe RAW = 0.09 (0.07 to 0.11) deg C/decade
Globe HOMOG = 0.09 (0.06 to 0.11) deg C/decade
N. Hemisphere RAW = 0.08 (0.06 to 0.10) deg C/decade
N. Hemisphere HOMOG = 0.08 (0.06 to 0.11) deg C/decade
Tropics RAW = 0.14 (0.11 to 0.19) deg C/decade
Tropics HOMOG = 0.13 (0.10 to 0.17) deg C/decade
S. Hemisphere RAW = -0.03 (-0.05 to 0.01) deg C/decade
S. Hemisphere HOMOG = 0.01 (-0.02 to 0.04) deg C/decade
There are no changes to the moistening Northern Hemisphere or Globe, a tiny decrease in the moistening trend for the Tropics and no change to the undetectable trend in the Southern Hemisphere.
Homogenised large scale trends for the Globe, Northern Hemisphere and especially the Tropics show significant moistening. There is no discernible trend in the Southern Hemisphere.
At the gridbox level the majority of gridboxes (87%) change trend direction when the data are homogenised. The overall range of trends across the gridboxes reduces when using the homogenised data.
Things to do:
Fit of guassian distribution to detected breakpoint distribution will need optimising at some point for accurate uncertainty estimation.
Now apply adjustments to RH, e, q and Tw based on combined T and Td breakpoints.
Figures
 |
Figure 1 Summary of adjustments made during PHA processing to homogenise HadISD monthly mean TEMPERATURE. a) Adjustment magnitude by latitude. b) Maximum adjustment incurred at each station. Adjustment magnitude here is measured relative to the next most recent homogeneous subperiod as opposed to present day.
|
 |
| Figure 2 Summary of the difference between gridbox and area average decadal trends from the raw and homogenised HadISD monthly mean TEMPERATURE. a) Ratio of raw to homogenised decadal trends. b) Scatter plot showing the agreement between raw and homogenised decadal trends for each gridbox with percentage of gridboxes in each quadrant (warming/warming, cooling/cooling, warming/cooling, cooling/warming). c) Distribution of gridbox decadal trends. d) Annual average time series for large scale cosine(latitude) weighted area averages and decadal trends with 5th to 95th confidence intervals. Trends are fitted using the median of the pairwise slopes with confidence intervals estimated by the 5th and 95th percentiles of the pairwise slopes. A trend is considered to be significantly positive/negative if both the 5th and 95th percentiles are above/below zero. |
 |
| Figure 3 Summary of adjustments made during PHA processing to homogenise HadISD monthly mean DEWPOINT TEMPERATURE. a) Adjustment magnitude by latitude. b) Maximum adjustment incurred at each station. Adjustment magnitude here is measured relative to the next most recent homogeneous subperiod as opposed to present day. |
 |
| Figure 4 Summary of the difference between gridbox and area average decadal trends from the raw and homogenised HadISD monthly mean DEWPOINT TEMPERATURE. a) Ratio of raw to homogenised decadal trends. b) Scatter plot showing the agreement between raw and homogenised decadal trends for each gridbox with percentage of gridboxes in each quadrant (moistening/moistening, drying/drying, moistening/drying, drying/moistening). c) Distribution of gridbox decadal trends. d) Annual average time series for large scale cosine(latitude) weighted area averages and decadal trends with 5th to 95th confidence intervals. Trends are fitted using the median of the pairwise slopes with confidence intervals estimated by the 5th and 95th percentiles of the pairwise slopes. A trend is considered to be significantly positive/negative if both the 5th and 95th percentiles are above/below zero. |
 |
| Figure 5 Summary of adjustments made during PHA processing to homogenise HadISD monthly mean WETBULB TEMPERATURE. a) Adjustment magnitude by latitude. b) Maximum adjustment incurred at each station. Adjustment magnitude here is measured relative to the next most recent homogeneous subperiod as opposed to present day. |
 |
| Figure 6 Summary of the difference between gridbox and area average
decadal trends from the raw and homogenised HadISD monthly mean WETBULB TEMPERATURE. a) Ratio of raw to homogenised decadal trends. b) Scatter
plot showing the agreement between raw and homogenised decadal trends
for each gridbox with percentage of gridboxes in each quadrant
(moistening/moistening, drying/drying, moistening/drying, drying/moistening). c)
Distribution of gridbox decadal trends. d) Annual average time series
for large scale cosine(latitude) weighted area averages and decadal
trends with 5th to 95th confidence intervals. Trends are fitted using
the median of the pairwise slopes with confidence intervals estimated by
the 5th and 95th percentiles of the pairwise slopes. A trend is
considered to be significantly positive/negative if both the 5th and
95th percentiles are above/below zero. |
 |
| Figure 7 Summary of adjustments made during PHA processing to homogenise HadISD monthly mean VAPOUR PRESSURE. a) Adjustment magnitude by latitude. b) Maximum adjustment incurred at each station. Adjustment magnitude here is measured relative to the next most recent homogeneous subperiod as opposed to present day. |
 |
| Figure 8 Summary of the difference between gridbox and area average
decadal trends from the raw and homogenised HadISD monthly mean VAPOUR PRESSURE . a) Ratio of raw to homogenised decadal trends. b) Scatter
plot showing the agreement between raw and homogenised decadal trends
for each gridbox with percentage of gridboxes in each quadrant
(moistening/moistening, drying/drying, moistening/drying, drying/moistening). c)
Distribution of gridbox decadal trends. d) Annual average time series
for large scale cosine(latitude) weighted area averages and decadal
trends with 5th to 95th confidence intervals. Trends are fitted using
the median of the pairwise slopes with confidence intervals estimated by
the 5th and 95th percentiles of the pairwise slopes. A trend is
considered to be significantly positive/negative if both the 5th and
95th percentiles are above/below zero. |
 |
| Figure 9 Summary of adjustments made during PHA processing to homogenise HadISD monthly mean RELATIVE HUMIDITY. a) Adjustment magnitude by latitude. b) Maximum adjustment incurred at each station. Adjustment magnitude here is measured relative to the next most recent homogeneous subperiod as opposed to present day. |
 |
| Figure 10 Summary of the difference between gridbox and area average
decadal trends from the raw and homogenised HadISD monthly mean RELATIVE HUMIDITY. a) Ratio of raw to homogenised decadal trends. b) Scatter
plot showing the agreement between raw and homogenised decadal trends
for each gridbox with percentage of gridboxes in each quadrant
(moistening/moistening, drying/drying, moistening/drying, drying/moistening). c)
Distribution of gridbox decadal trends. d) Annual average time series
for large scale cosine(latitude) weighted area averages and decadal
trends with 5th to 95th confidence intervals. Trends are fitted using
the median of the pairwise slopes with confidence intervals estimated by
the 5th and 95th percentiles of the pairwise slopes. A trend is
considered to be significantly positive/negative if both the 5th and
95th percentiles are above/below zero. |
 |
| Figure 11 Summary of adjustments made during PHA processing to homogenise HadISD monthly mean SPECIFIC HUMIDITY. a) Adjustment magnitude by latitude. b) Maximum adjustment incurred at each station. Adjustment magnitude here is measured relative to the next most recent homogeneous subperiod as opposed to present day. |
 |
| Figure 12 Summary of the difference between gridbox and area average
decadal trends from the raw and homogenised HadISD monthly mean SPECIFIC HUMIDITY. a) Ratio of raw to homogenised decadal trends. b) Scatter
plot showing the agreement between raw and homogenised decadal trends
for each gridbox with percentage of gridboxes in each quadrant
(moistening/moistening, drying/drying, moistening/drying, drying/moistening). c)
Distribution of gridbox decadal trends. d) Annual average time series
for large scale cosine(latitude) weighted area averages and decadal
trends with 5th to 95th confidence intervals. Trends are fitted using
the median of the pairwise slopes with confidence intervals estimated by
the 5th and 95th percentiles of the pairwise slopes. A trend is
considered to be significantly positive/negative if both the 5th and
95th percentiles are above/below zero. |
 |
| Figure 13 Station standard deviations and autocorrelations for monthly mean TEMPERATURE by latitude. |
 |
| Figure 14 Station standard deviations and autocorrelations for monthly mean WETBULB TEMPERATURE by latitude. |
 |
| Figure 15 Station standard deviations and autocorrelations for monthly mean RELATIVE HUMIDITY by latitude. |
 |
| Figure 16 Station standard deviations and autocorrelations for monthly mean SPECIFIC HUMIDITY by latitude. |
 |
| Figure 17 Station standard deviations and autocorrelations for monthly mean VAPOUR PRESSURE by latitude. |
 |
| Figure 18 Station standard deviations and autocorrelations for monthly mean DEWPOINT TEMPERATURE by latitude. |
No comments:
Post a Comment
Note: only a member of this blog may post a comment.