"Terrific"???
It's simply loaded with disinformation, innuendo and false comparisons.
Let's focus on the 'no warming for the past 15 years'. This is a myth that the result of 'cherry picked' dates, and can be shown to be completely untrue when using a scientifically valid approach to looking at the data.
This website has several global temperature datasets that you can enter in your own timeframes, filtering ranges, etc to observe the data for yourself:
http://www.woodfortrees.org/plot/
This site lets you interactively create your plots, and then you can link those data plots directly, which is really pretty cool.
Looking at the HADCRUT4 global temperature series, the way you can actually see trends is by running an averaging filter to smooth out the noise, which often hides true trends.
Here is HADCRUT4 global mean without ANY running filter:
The amount of seasonal and annual variation creates a rather big mess and it becomes difficult for the human eye/brain to see trends.
To make trend-spotting easier, you can run a 10-year (decadal) average (just boxcar 120 months of data together), but this can often create misleading inversions at certain points along the plot. Instead, you run 2 or 3 sub-decadal averaging filters that 'sum up' to the 120 months (e.g. 29, 39, 52 months averages) in sequence, and you eliminate the 'local inversions' which often skew the information over short time frames. (FWIW, Judith Curry has a website and page which details this method. Yes, the Judith Curry who is the AGW 'skeptic').
Here is that link, where the 'local errors' that smoothers create are described:
http://judithcurry.com/2013/11/22/data-corruption-by-running-mean-smoothers/
Here is what the HADCRUT4 data looks like with the decadal averaging (each data point represents one month of data, so 120 points = 10 years):
If we 'zoom in' from 1980 on, you can see what the past decade really looks like:
The green line uses the 'simple' decadal average, which just uses the single running filter - this creates the misleading inversions that Curry explains are artifacts of that method, and are not a good averaging method. The red line uses sequential running mean filters that 'add up' to a decadal filter, but to a better job of showing the true local trends. The red line uses the method recommended by Curry's site.
It is quite easy to see from this plot that we have, AT MOST, a 5-year hiatus in warming (red line). The green line (using the method Curry does NOT recommend) is even worse (because this method generates local inversions and biases that are 'not real', but an artifact of the filtering)
The reason the graph only runs to 2010 is because the data point for that decadal average (2005-2015) is plotted at the 2010 data point. Since we don't have data points out to 2020 yet, we will not know what the 2015 decadal average data point looks like until 2020. But it is quite easy to see that there is nothing CLOSE to any '15 year' or '18 year' hiatus. A '15 year hiatus' means that the data point on the running mean for 1995 is 'equal to' the data point at 2010. It is not remotely close.
Look at the running graph from the 1880's on; there is simply nothing close to a 'recent hiatus' in the data yet; at least nothing that is a deviation any bigger than anywhere else along the historical record. You can find a 'hiatus' just about as long from 1989 to 1995....
Go visit the site and look at the data for yourself. Make your own plots, using running means.
If you want a decadal averaging filter, use mean ranges of 29, 39, 52 in sequence. If you want a 'true' 17 year running mean, use 49, 66, 88.
Here is the GISTEMP dT global mean dataset, using the same decadal filter:
Again, no 'hiatus'. You can look at the BEST dataset, MSS, UAH etc and you cannot identify clear hiatus trends in any global datasets (note that some datasets are 'land only' and thus only represent 30% of the full temperature dataset for all of the Earth).
There are countless other fallacies and errors in that article, too many to debunk in one post.
