Graphs like this show how the trends of temperature differ between the coldest days (or nights) of the year, the hottest ones, and all those ranked in between.
This first post on this topic is a “sampler” of Manilla data that I will present. It compares my first 9-year period March 1999 to February 2008 with the 9-year period September 2003 to August 2012, four and a half years later.
All the days (or nights) of the year are arranged from the coldest on the left to the hottest on the right. Columns show by how much the day or night of that rank has trended warmer or cooler during the nine years. (See also Notes below.)
In the earlier period (blue), most winter days and a few mid-summer days cooled at 0.1 to 0.2 degrees per year. Days in spring and autumn, and cooler days in summer warmed at less than 0.1 degrees per year.
In the later period (red), all days of the year cooled, but there was a gradient from no cooling in midwinter to extremely rapid cooling (more than 0.3 degrees per year) in midsummer.
In the earlier period (blue), nights in the warmer half of the year, and in midwinter warmed at about 0.1 degrees per year. There was no warming either in midsummer or in the warmer part of winter.
In the later period (red), it was now in the cooler half of the year that nights warmed at about 0.1 degrees per year. Nights in the warmer part of summer cooled more and more rapidly as they approached midsummer, where the cooling rate was 0.25 degrees per year.
[The 50-year average warming of this part of australia is 0.015 degrees per year. That is, less than two tick-marks on the y-axis.]
This graph and its commentary appeared as a post in “weatherzone” forums on 25/10/12:
Earlier similar posts in date order were:
1. To August 2008 and August 2009 (9/9/09)
2. To August 2009 and August 2010 (22/9/10)
3. To August 2010 and August 2011 (15/9/11)
Note on graphical technique.
This kind of graph appeared in this paper:
Paul C. Knappenberger, Patrick J. Michaels, and Robert E. Davis (2001), “Nature of Observed temperature changes across the United States during the 20th century”, Climate Research 17, 45-53. “Chip” Knappenberger believes that it was original.
Note on graph details.
The number of days in each year is reduced to 364 by deleting the 29th of February and the 31st of August. This allows plotting of 91 columns of days averaged in groups of four. Text boxes noting temperatures and seasons are approximate only.
2 thoughts on “Ranked Hot and Cold Days”
Still trying to get my head round that method of graphing!
The key to the method is the re-arrangement of temperature data (max or min) for all the days of each year in rank order from the coldest to the hottest. Then it becomes possible to see the temperature trend of the hottest day, for example. Also for the coldest, and for any other day in the rank order, such as the 20th hottest.
I have annotated the graphs with seasons and mean temperature values, but I have tried to make it clear that these labels are not definitive, as the individual columns are. They simply give the viewer some orientation.
Preparing such graphs from available data could avoid a lot of loose talk about whether heat waves are getting worse, for example.
Right now, it has been said that cats in Australia are having more kittens due to warmer winters, particularly in 2012. Such graphs could confirm that such a specific climate change has occurred and where. (And do the cats respond to warm days or warm nights?)