HadISDH 2012p Trends compared to HadISDH 2011

HadISDH 2012p Trends compared to HadISDH 2011

HadISDH.landq.1.0.0.2012p has now been created. There are differences in both the gridbox trends and large-scale average trends that need to be investigated. Both 2011 and 2012p have been run using the new SLP conversion algorithm and the MOHC in house PHA algorithm. The only things that differ are:

            - the extra 12 months of data for most stations

            - some historical changes to the archive resulting in addition and loss   of some years of data for some stations (discussed in previous post)

            - we now have 51 new stations that weren't in 2011 because they did not have enough data before

026720 042700 060710 060960 125500 201070 233830 249880 255380 255510 257030 299620 309250 312630 360340 369820 377350 402650 412160 415300 420270 442120 442310 442320 442820 505480 602850 647000 701740 702710 712220 719450 722119 722348 722925 723825 724936 725376 725705 765710 783670 789620 841170 862970 895320 896110 910660 916970 919450 961090 984440

            - we have now lost 57 stations that were in 2011 but no longer have enough data 

023360 026220 040560 040890 042300 062470 133890 135780 222920 246880 254000 256210 296760 359690 383530 384030 435550 442650 443580 444540 471920 484770 485010 623370 624320 680320 683280 702070  702315 710750 711974 714963 719170 722786 724288  724935 725865 725946 726375 726487 726548  726555 726796 727890 784850 785830 789540 800940 801120 804130 828250 829150 915200 944300  962210  984270 986420

Differences in large-scale trends fit by using the median of the pairwise slopes and showing the 5th and 95th percentiles of those slopes as a measure of whether the trend is significantly different from a zero trend:

HadISDH 2012p

GLOB 1:     0.089 (0.080 to 0.098) g/kg/decade
NHEM 1:    0.086 (0.075 to 0.098) g/kg/decade
TROP 1:      0.133 (0.119 to 0.148) g/kg/decade
SHEM 1:     0.013 (-0.005 to 0.031) g/kg/decade

HadISDH 2011
GLOB 2:     0.095 (0.086 to 0.105) g/kg/decade
NHEM 2:    0.092 (0.081 to 0.103) g/kg/decade
TROP 2:      0.143 (0.128 to 0.157) g/kg/decade
SHEM 2:     0.013 (-0.007 to 0.032) g/kg/decade

So all trends are within 0.01 g/kg/decade and there is no change to the general conclusions of significant moistening for the Globe, Northern Hemisphere and Tropics. A slight moistening is shown for the Southern Hemisphere but this is not significantly different from zero.

Differences at the gridbox level:

Difference in decadal trends between HadISDH 2012p and HadISDH 2011.

 90.4% of gridboxes have trends of the same direction. HadISDH 2012p trends appear slightly lower/drier.

Conclusions:

There will be changes at the gridbox level with the year on year updates that may not be desirable for some users. HadISDH is designed as a monitoring product for large-scales, where changes are more robust to year on year changes in the data. We advise users of the smaller scale aspects to carefully assess the data before use, especially in the case of annual updates.