Questions to Ask about a Plot before You Try to Read and Interpret It

Before you begin to describe or interpret any plot of data, such as the global temperature and temperature difference plots in Lab Activity #1, there are several types of questions that you should try to answer about the plot:

• Questions about metadata (that is, data or information about the data):
  • What quantity(ies) is (are) being plotted?
    • What units are used to express the data values?
    • Have "raw" data (observations or model output) been analyzed or processed before being plotted (e.g., averaged over time or space)?
  • When were the data recorded?
  • Where were the data recorded?
    
  • (A secondary, but sometimes valuable, question to ask might be, what instruments were used to record the data, or what numerical model calculated the data, if a forecast or analysis model generated them?)
    
    
• Questions about the plot (that is, the method and conventions used to visualize the data):
  • How are the data plotted or visualized?
    • What conventions have been adopted (stated and/or unstated) to implement the plotting/visualization method?
      • On a graph of one quantity plotted as a function of the other, which axis corresponds to the dependent variable and which to the independent variable? What are the vertical and horizontal scales (range of values) on each axis? Are colors assigned to plotted curves or symbols assigned to data points, and if so, what do the colors or symbols represent?
      • On a color-shaded plot or a color-filled contour plot, what colors have been assigned to what (range of) data values? What are the maximum and minimum data values (or range of values) represented?

Plots of Monthly Mean Surface TemperatureData (created using My World GIS software)

For the two temperature plots in Lab Activity #1, some responses (and some actual answers) to these questions would be:

• What data are plotted? Some sort of temperature is plotted. A description of the data available through My World GIS says the following:
  
  "Temperature is measured directly at thousands of weather stations around the world, as well as by ships, airplanes, and satellites. The temperature data set in [My World GIS] was created by combining measurements from all these sources and using a complex computer program to calculate probable temperatures in locations for which there are no measurements available." It adds, "Data for some grid cells are based on inferences and interpolations between observations and might be inaccurate for some cells."
  
  Whatever the source, the temperatures are averaged temporally (that is, over time) over a full month and spatially over 2.5° latitude × 2.5° longitude "cells". They are expressed in Kelvins.
  
  
• When were the data recorded? In January and in July, 1982. We don't know how frequently the temperatures were measured over the course of each day during those two months, but the answer would vary depending on the individual sources of data that went into the overall data set.
  
  
• Where were the data recorded? First, the data are supposed to represent temperature at the earth's surface. Whether that means air temperature close to the earth's surface or the temperature of the surface itself (water, rock, soil, sand, etc.), isn't clear. Second, the data were recorded globally, though with incomplete coverage because there are some parts of the earth where temperatures aren't recorded by one means or another.
  
  
• How were the data plotted? The visualization is a color-shaded plot. A table of colors (20 in this case) is defined, and these colors are assigned to span a specified range of temperatures so that each color in the table represents a particular sub-range of the full range of temperatures. For the monthly-average surface temperatures, a color table consisting of 20 colors, from dark blue at one end of the color table and ranging through progressively lighter shades of blue, then transitioning abruptly to shades of yellow, then orange, and then red. The 20 colors are assigned to a range of temperatures from 220K (–53°C or –63°F) at the dark blue end of the color table to 320K (47°C or 117°F) at the dark red end. Each color represents the same increment of temperature, which is (320K – 220K)/20 = 5K. Since an increment of temperature in Kelvins and in degrees Celsius (degrees centigrade) are equal, each color represents an increment of 5°C, too, which is 5°C × (9°F/5°C) = 9°F.

As a result of the nature of the data plotted (spatially averaged values over 2.5°× 2.5° latitude/longitude cells, using color shading as the visualization method), when you zoom in closely on the plot you see "pixelation", in this case the individual square cells over which the temperatures are averaged, each a single color.

Note, though, that in three dimensions on a globe, while it is true that latitude lines are evenly spaced and parallel, longitude lines are, in contrast, most widely spaced at the equator and converge at the poles. (At the equator, 1° of latitude and longitude span the same distance—namely, 111 km. At progressively higher latitudes, 1° of longitude spans a progressively smaller distance, while 1° of latitude spans the same distance everwhere.)

In the plots below, the map projection separates the longitude lines at the poles and spreads them out until they all become parallel to each other, creating a rectangular map in two dimensions in which each 2.5° latitude × 2.5° longitude grid cell appears as a square and all of the cells are the same size. The distance from north to south across each cell is the same for all cells (about 277 km), but the distance across each cell from east to west varies from a maximum of about 277 km at the equator to about 11 km across the center of the row of grid cells that join at the poles (each of which is centered at 87.75° latitude). Hence, the true size of the grid cells is largest at the equator and smallest at the poles, though you wouldn't guess that looking at the uniformly sized grid cells in the map projection used here.

 

The varying grid cell sizes has implications for how we calculate global averages using these data sets, which we'll learn more about later.

Features of plots of January and July average surface temperature for a selected year (e.g., 1982):

• Temperatures are generally warmest in the low latitudes (which are defined to be between 30°S and 30°N latitude) and decrease progressively across the middle latitudes (between 30° and 60°N and between 30°S and 60°S latitude) to their coldest values in the high latitudes (between 60° and 90°N and between 60° and 90°S). (This is a qualitative description of the broadest, global aspects of the temperature pattern.)
  
  
• There are some large temperature differences between places at the same time of year (e.g., differences > 83°C [~150 °F] in January and > 111°C [~200 °F] in July). (Note that a difference or change in temperature of 1K is the same as a difference or change in temperature of 1°C, which is the same as a difference or change in temperature of 1.8°F = 9/5°F.) (This begins to quantify some aspects of the temperature pattern, which can add value to the qualitative description.)
  
  
• One of a number of secondary aspects of the pattern is that although the hottest temperatures generally lie within the low latitudes, they lie north of the equator in July but tend to lie south of the equator in January (a distinction that is most obvious over land).
  
  

Plots of Monthly-Average Surface Temperature

January 1982	July 1982

Plot of the Difference between Montly-Average Surface Temperatures from Different Months

It is not always easy to describe differences between plots of similar quantities, such as monthly-average surface temperatures in January and July, simply by placing them side by side. One, possibly more effective way to look for differences between them might be to make similar plots for the intervening months and animate the series of monthly plots. Another, often very powerful method is simply to subtract the values on one plot from the values in the corresponding cells on the other, and plot the difference. Lab Activity #1 does this for you, and the answers to some of the questions raised about the individual January and July plots of monthly-average surface temperature, are different:

• What data are plotted? The July minus January difference in monthly-average surface temperatures.
  
  
• When were the data recorded? Same answer as for the individual monthly-average data plots.
  
  
• Where were the data recorded? Same answer as for the individual monthly-average data plots.
  
  
• How were the data plotted? The visualization is a color-shaded plot. A color table containing 60 colors, starting with dark blue and transitioning through progressively lighter shades of blue, to white, and then to light red through progressively darker reds. These colors are assigned to values of July minus January temperature differences ranging from –60K to 60 K (which is the same as –60°C to 60°C, since differences or changes in temperature are the same whether expressed in Kelvins or °C). (In °F, the range of temperature differences would be 9/5 larger: –108°F to 108°F.) Each individual color therefore represents an increment of 2K or 2°C (or 3.6°F).

• July – January (that is, "July minus January") monthly-average temperature differences are opposite in sign in N. and S. Hemispheres (positive differences, shown in red, imply that it is warmer in July than in January; red colors appear almost exclusively in the N. Hem.; negative differences, shown in blue, imply that it is warmer in January than in July, and blue colors appear almost exclusively in the S. Hem.). In 1982, a more or less typical year, in the polar regions in both hemispheres the differences between July and January approached 60°C [108°F].
  
  
• July – January temperature differences are smallest in the low latitudes and largest at higher latitudes (though there seems to be a relatively abrupt increase in the size of the differences near the Antarctic coastline and around the edge of the Arctic Ocean, and these aren't always land/water boundaries).
  
  
• July – January temperature differences are generally small over oceans and are much larger over land. Notable exceptions to this observation: (a) in the low latitudes, where temperature differences are small over land, and (b) at the highest latitudes where they are often large over oceans.
  
  
• July – January temperature differences over land seem larger farther inland than near most coasts, especially the western coasts at midlatitudes.
  
  
• Other, even smaller or more subtle features. For example, large lakes and inland seas (e.g., Great Lakes in the U.S.; Caspian Sea in western Asia) show smaller July – January temperature differences than surrounding, land-based areas.(The resolution of the data is barely high enough to tell, though!)