I have been posting on ways of dealing with cells in the HADCRUT 4 surface temperature anomaly grid which have no data. This follows a new much-discussed paper by Kevin Cowtan and Robert Way in QJRoyMetSoc. They used satellite data and kriging to extend the data, particularly into polar regions. The result that created particular interest concerned trends in a recent period, 1997-2012, which has been characterised as a "pause", partly because HADCRUT showed little trend. C&W found that the trend was significantly greater with this extra care.
I have been discussing this, in posts here and here. In the second post, I showed that simply infilling missing cells each month with the average for the latitude band, rather than implicitly by the global average (the effect of omission) also gave a substantially greater trend over this period.
Interest has been expressed in the seasonal nature of this change. The Arctic amplification is expected to be predominantly a winter effect (Serreze). So here are plots of the various infills (None, my lat av, C&W and also UAH for comparison) over the four seasons, and annually. I have taken William Connolley's suggestion of plotting against sin latitude to avoid overemphasising the polar bands (Mercator-style).
Here is the plot. It's an active plot - you can choose your season. The "HAD4 Lat" is my infill with latitude band averages, with parameter r=0.1, as described here. The other datasets were introduced and plotted here.
Annual DJF MAM JJA SON |
The background stripes indicate the latitude bands; N pole is on the right. There's a faint red vertical marking multiples of 30°. In the centre is a little figure showing the global totals, on the same scale. DJF etc indicate the seasons by initial of the months.
Here is a table of the global trends, 1997-2012:
| Season | HADCRUT 4 | HAD 4 Lat Av | UAH | C&W Hybrid |
| DJF | -0.0489 | 0.0097 | 0.0276 | 0.0536 |
| MAM | 0.066 | 0.1019 | 0.0639 | 0.1276 |
| JJA | 0.0892 | 0.0966 | 0.12 | 0.1187 |
| SON | 0.1081 | 0.1277 | 0.1575 | 0.1688 |
| Annual | 0.0539 | 0.0844 | 0.0936 | 0.1187 |
An interesting aspect is that, while the Arctic does have a marked maximum in NH winter, that is a minimum season of of global trends, and quite markedly so. There is a substantially negative trend in the adjacent N latitudes, from about 40-60°. It is interesting to speculate on whether these are related. The global trend was also negative for this season in the original HADCRUT, but the infilling, particularly with the C&W hybrid, had the greatest effect.
The Antarctic region does not have a marked seasonality in the trend, and nor does any other (they don't have big trends at all).
Posts
Interesting, thanks.
ReplyDeleteA source of bias in UAH could have been the change in sea ice cover. That could have someone created problems for the C&W hybrid. However, if the adjustments of C&W of the HadCRUT were mainly in winter, and in winter the ice cover did not change much, that potential problem becomes less likely. Here might still be potential problems with the characteristics of the sea ice. I have no idea whether age, thickness or surface structure is important.
Interesting point, we have actually discussed Sea ice a bit here:
ReplyDeletehttp://www-users.york.ac.uk/~kdc3/papers/coverage2013/update.131020.pdf
However the use of the UAH data or not makes rather little difference in the Arctic, because the area is small.
ReplyDelete(Actually it's a bit more subtle - the kriging range reduces when you switch to hybrid, but the sill reduces faster, so the uncertainties are reduced but some of the predictive power is shifted from the extrapolation to the satellite data.)
Kevin
Is the HadCRUT4 Lat DJF trend wrong in the table? Doesn't seem to fit with the latitudinal plot.
ReplyDeleteRegarding linkage between high-mid latitude cooling and Arctic warming in Winter I've checked time series in reanalyses and there doesn't seem to be a huge degree of correlation or anti-correlation in the timing of their respective anomalous periods. In Arctic Winter there is a large anomalous increase in 2005, and temperatures have stayed around that heightened level since then. 2005 Winter in the high-mid latitudes was on the cold side but around average really, nothing special. The prolonged spell of cold Winters begins in 2008.
Doesn't mean they aren't related - it would seem strange if they weren't in some ways - but I would suggest they're distant cousins rather than siblings. Perhaps the Arctic Winter warming has triggered sea ice reductions which have, in turn, led to circulation changes which have caused a spell of cold Winters.
Triggered by Judith Curry's latest posts would you be able to plot Arctic time series for Cowtan & Way for comparison with reanalyses?
Paul,
ReplyDeleteI'm not sure where you are seeing a disagreement with DJF. The table values are supposed to correspond to the rectangles in the middle, and it seems to me in that case they do.
Thanks for the info on timing of winter changes - your summary sounds right. On plotting and reanalyses, I'll check that out. I probably should plot the Lat Av time series too, rather than just trends.
Just seemed odd that C&W Hybrid and your HadCRUT4 Lat track with each other in almost all bands yet there is a relatively large trend difference. I guess it suggests a chunk of C&W's extra annual trend comes from outside the Arctic.
DeletePlease check GISS's land temp 1200 km product for the 90-60N sector against CRUTEM4's for the same region, from 1997 till today (baseline 1981-2010 for both). You will find that the linear trend of the latter one is ever so slightly steeper than the trend of the former one. Strangely, GISS doesn't attain a higher trend 1997-2013 than CRU for the near-global curve (without Antarctica), 90N-60S, even when they include 'all' of the Arctic land areas through their smoothing and CRU don't.
ReplyDeleteThen look at the SSTa product of UKMO, the HadSST3. It has a HUGE warm bias north of 60N compared to other well-established SSTa datasets:
http://i1172.photobucket.com/albums/r565/Keyell/90-60NSSTax_zps98c921ec.gif
Why is this very large, very obvious and downright extraordinary warm bias not addressed in the study? What on earth happened in 2000/2001?!
If anything, there is considerably LESS heat accumulated in the Arctic over the last 15-16 years than what the HadCRUt4 is showing ...
Kristian
The data from different global land temperature groups can be tricky to cross-compare, partly due to incompatible treatment of coastal masking. Assuming you're using Climate Explorer to generate these time series, try setting to only consider land points when outputting GISS 1200 and you should see a larger trend. This setting applies a coastal mask to prevent generally slower warming coastal and island regions from being over-weighted by interpolation into ocean grid cells. There might be issues with CRUTEM4 along similar lines (although less so) due to their large 5ยบ grid size, but it's possible they've accounted for this in the record construction(?)
DeleteOn a general point keep in mind that what you're plotting with these zonal (e.g. 60N-90N) averages can be misleading when translating to influence on the global record. As Nick has stated, where grid cells are empty the global or zonal average of covered cells is effectively used to infill. As an extreme illustrative example let's say there is only one covered cell within 60-90N, which features very strong warming. When creating a 60-90N average the missing cells would take their values from this strong warming cell, so the zonal average would also feature strong warming. However, when creating a global average those same missing cells will take their values from the global average which might be completely different. In that case despite indicating a strong warming Arctic zonal average our lone Arctic cell has very little influence at the global scale. The purpose of this paper is to determine what the replacement for those missing Arctic cells should be, instead of simply using the global average as an infill - it's not about the coverage which already exists in HadCRUT4.
The difference between HadSST3 and other records in the Arctic is interesting though. Note that it's not just the trend but also the variance which is larger. HadSST3 variance is closer to that seen in the source ICOADs data so I would suggest the interpolation (both spatial and temporal) used in other records is tending to (artificially) smooth it. That doesn't obviously effect the trend but I think there is a potential time-varying bias:
Data collection in the Arctic is spotty, but it's also seasonal and seemingly opportunistic. As sea ice cover has declined there will have been greater scope for collection of data. Some Arctic regions are now free of sea ice perennially so Winter data can be collected where it hasn't been previously, and the large Summer decline means there are many regions now open for measurement which weren't previously. This time-varying coverage could potentially cause a bias in the trend. I've found the Summer HadSST3 60N-90N trend is largest, which would fit with the possibility of such a bias, though it's not definitive. Due to the interplolation employed in other records there is much less variation in coverage, hence no bias.
Cowtan & Way's method would actually fix such a bias because it ensures all grid cells are covered in every month.