How to paint part of DateListPlot?

Question

My graph is about fires (burned area) over 12 years presented in one Date List Log Plot. Because most of fires are in summer (as expected) I would like to highlight that with different background colors. How can I paint seasons in Date List Plot in different colors? I want to put different background color for winter, spring... Thank you!

Answer 1

I would suggest using Prolog and graphics rectangles like this

fakedata = 
  Transpose[{DatePlus[{1977, 1}, {#, "Month"}] & /@ 
     Range[0, 33*12 - 2], 
    FoldList[#1 + #2 &, 59.5, 
     RandomVariate[NormalDistribution[0.002, 0.5], 33*12 - 2]]}];

You don't need AbsoluteTime to define the dimensions of the rectangles, but you do need to define the vertical range of the rectangles to be beyond the range of your data. You can't use Scaled in just one dimension. The code below shows how to have multiple rectangles, though in this case I have used recessions and milder economic downturns in Australia rather than summers as the dates for the rectangles. (I happened to have a similar graph lying around.) If you want different colors for the different rectangles, just intersperse the colors in between the Rectangle objects in the list.

DateListPlot[fakedata, Joined -> True, 
 Prolog -> {GrayLevel[0.8], 
   Rectangle[{{1981, 6}, 0}, {{1983, 7}, 90}], 
  Rectangle[{{1989, 11}, 0}, {{1992, 3}, 90}], 
   Rectangle[{{2008, 1}, 0}, {{2009, 7}, 90}]}]

Answer 2

Here is the version I use (it is quite similar to Verbeia's, but I don't want to compute the vertical range every time):

ClearAll[dateListPlotShaded];
Options[dateListPlotShaded] = 
  Join[DeleteCases[Options[DateListPlot], GridLines -> _], 
        {GridLines -> None, ShadeColor -> LightBlue, ShadeOpacity -> 1}];

With[{opt = First /@ Options[DateListPlot]},
  dateListPlotShaded[ddata_, ShadeDates_, OptionsPattern[]] :=
     Module[{yRange, yPadding},
        With[{plot=DateListPlot[ddata, Sequence@@((#->OptionValue[#])& /@ opt)]},
           yRange= Last[PlotRange/.AbsoluteOptions[plot, PlotRange]];
           yPadding = If[OptionValue[Frame] === True, - .2 Subtract@@yRange, 0];                                
           Show[plot,
             Prolog -> {{OptionValue[ShadeColor], 
               Opacity[OptionValue[ShadeOpacity]], 
                 (Rectangle[
                    {AbsoluteTime[#[[1]]], 
                       yRange[[1]] - yPadding}, 
                    {AbsoluteTime[#[[2]]], 
                       yRange[[2]] + yPadding}
                    ]&) /@ ShadeDates}, 
             {OptionValue[Prolog]}}]]]];

Plotting the data:

fakedata = 
  Transpose[{DatePlus[{1977, 1}, {#, "Month"}] & /@ 
    Range[0, 33*12 - 2], 
    FoldList[#1 + #2 &, 59.5, 
    RandomVariate[NormalDistribution[0.002, 0.5], 33*12 - 2]]}];

dateListPlotShaded[fakedata, {{{1981, 6}, {1983, 7}}, {{1989, 11}, {1992, 3}}, 
   {{2008, 1}, {2009, 7}}}, Joined -> True]

You can use dateListPlotShaded with all the options of DateListPlot (including Prolog and with Frame -> False).

Answer 3

Here's another suggestion - there are (at least) two caveats:

1. you need rather dense data (I fill points from axis, so not really good for only single points in a year)
2. I can't get it working for DateListPlot

Nevertheless, it looks quite ok to do that sort of coloring for e.g. financial data. And maybe it helps you with your fires.

So in that sense, we can load:

data = FinancialData["IBM", "Jan. 1, 2004"];

And then, we define colors per month (group of months):

setup = {{{11, 12, 1}, ColorData["SolarColors"][1/3]}, 
{{2, 3, 4}, ColorData["SolarColors"][2/3]}, 
{{5, 6, 7}, ColorData["SolarColors"][3/3]}, 
{{8, 9, 10}, ColorData["SolarColors"][0/3]}};

Then:

DateListPlot[data, Joined -> True, 
    ColorFunction -> ((Which[MemberQ[setup[[1, 1]], #], setup[[1, 2]], 
    MemberQ[setup[[2, 1]], #], setup[[2, 2]], 
    MemberQ[setup[[3, 1]], #], setup[[3, 2]], 
    MemberQ[setup[[4, 1]], #], setup[[4, 2]]] &)@
    ToExpression@DateString[#1, "Month"] &), Filling -> Axis, 
    ColorFunctionScaling -> False, ImageSize -> Large]

and get: