Wind Direction Aster Plot

[kbdw]

Hello,

I am trying to create an aster plot (or similar) showing wind direction at a weather station. The measurements were every 15 minutes, from which I have created an average daily measurement. I have then created a column converting the measurement (presented in degrees) into a direction i.e. N, E, NW, SSW etc. Therefore I want to present a plot of showing the number of averaged days within each category of wind speed direction. Below shows my output so far.

There are two main issues with this output :

• I cannot sort the "order" of direction categories on the plot to match the compass layout.
• Any direction categories without any datapoints e.g., N, NE, do not plot at all.

Hope someone can assist.

[lbendlin]

both issues can be mitigated. Easiest is to have  a static dimension table with all desired directions in the right order (add an index column if needed and "sort a column by another column").  Then choose the direction from that table and say "show items with no data".

Please provide sample data that fully covers your issue.
Please show the expected outcome based on the sample data you provided.

[kbdw]

Thank you for your response. I'm a beginner to Power BI so please let me know if any further information is needed.

Below is a sample dataset showing the daily averaged wind direction in degrees, converted to a direction.

Dt.1CountDirection

20/06/2023	218.255102	SW
19/06/2023	225.3489583	SW
18/06/2023	204.5104167	SSW
17/06/2023	194.9947917	SSW
16/06/2023	198.8020833	SSW
15/06/2023	179.8125	S
14/06/2023	191.6041667	SSW
13/06/2023	197.2447917	SSW
12/06/2023	185.725625	S
11/06/2023	184.21875	S
10/06/2023	183.609375	S
9/06/2023	118.2463542	ESE
8/06/2023	89.62833333	E
7/06/2023	128.9645833	SE
6/06/2023	168.300625	SSE
5/06/2023	207.6360417	SSW
4/06/2023	246.9727083	WSW
3/06/2023	286.3077083	WNW
2/06/2023	297.9644792	WNW
1/06/2023	211.5104167	SSW
31/05/2023	199.5104167	SSW
30/05/2023	228.2552083	SW
29/05/2023	278.5677083	W
28/05/2023	208.4166667	SSW
27/05/2023	193.6979167	SSW
26/05/2023	214.1041667	SW
25/05/2023	188.5729167	S
24/05/2023	190.0677083	S
23/05/2023	189.8020833	S
22/05/2023	198.7135417	SSW
21/05/2023	201.9947917	SSW
20/05/2023	229.9427083	SW
19/05/2023	217.9583333	SW
18/05/2023	209.640625	SSW
17/05/2023	177.4895833	S
16/05/2023	190.2291667	S
15/05/2023	189.8385417	S
14/05/2023	187.9895833	S
13/05/2023	188.953125	S
12/05/2023	208.9166667	SSW
11/05/2023	194.9791667	SSW
10/05/2023	204.390625	SSW
9/05/2023	284.578125	WNW
8/05/2023	260.7239583	W
7/05/2023	241.5729167	WSW
6/05/2023	318.96875	NW
5/05/2023	238.125	WSW
4/05/2023	222.0260417	SW
3/05/2023	272.6875	W
2/05/2023	300.546875	WNW
1/05/2023	218.6041667	SW
30/04/2023	186.3541667	S
29/04/2023	176.21875	S
28/04/2023	196.7760417	SSW
27/04/2023	184.8125	S
26/04/2023	175.640625	S
25/04/2023	173.4791667	S
24/04/2023	191.921875	SSW
23/04/2023	176.0625	S
22/04/2023	204.0520833	SSW
21/04/2023	213.0052083	SSW
20/04/2023	214.5104167	SW
19/04/2023	226.3177083	SW
18/04/2023	220.5260417	SW
17/04/2023	177.5104167	S
16/04/2023	190.2604167	S
15/04/2023	206	SSW
14/04/2023	177.8177083	S
13/04/2023	179.9635417	S
12/04/2023	201.4010417	SSW
11/04/2023	262.8854167	W
10/04/2023	211.82125	SSW
9/04/2023	220.2760417	SW
8/04/2023	194.7916667	SSW
7/04/2023	170.5364583	S
6/04/2023	189.1458333	S
5/04/2023	154.34375	SSE
4/04/2023	195.390625	SSW
3/04/2023	198.109375	SSW
2/04/2023	181.595	S
1/04/2023	257.9895833	WSW
31/03/2023	198.75	SSW
30/03/2023	220.5260417	SW
29/03/2023	207.7291667	SSW
28/03/2023	195.4427083	SSW
27/03/2023	229.9583333	SW
26/03/2023	164.5885417	SSE
25/03/2023	189.21875	S
24/03/2023	226.28125	SW
23/03/2023	215.75	SW
22/03/2023	144.5681818	SE

The expected outcome would resemble the below, except the directions would be arranged so as to resemble a compass rose.

---

@lbendlin wrote:

Easiest is to have  a static dimension table with all desired directions in the right order (add an index column if needed and "sort a column by another column").  Then choose the direction from that table and say "show items with no data".

---

Understand this in principle but would be grateful for directions as to how to achieve this.

[lbendlin]

The standard Aster plot is not entirely satisfactory.  Probable would make sense to use Deneb for this instead.

See attached.

[kbdw]

Happy to use alternative plot.

The categories on that plot still need shifting so that "N" is at the north point of the diagram, and so forth. In addition - can the position of the category boundaries be altered within the circle? At the moment for example, "N" would "start" at 0 degrees whereas in fact (with my direction categories defined as they are) it would need to span 348.75 - 11.25 degrees - as below.

[lbendlin]

@dm-p can you please help me over the hump? I found theta but don't know how to draw partial arcs.

[dm-p]

Hi @lbendlin - you also have theta2 (end angle) available for encoding. Here's the doc link (with interactive example).

[lbendlin]

Hmm. Can't seem to get it to turn the right way, even with the theta adjustments. Also tried to add placeholders for the missing directions.

{
  "data": {"name": "dataset"},
  "layer": [
    {
      "mark": {
        "type": "arc",
        "innerRadius": 20,
        "stroke": "#444"
      }
    },
    {
      "mark": {
        "type": "text",
        "radiusOffset": 50
      },
      "encoding": {
        "text": {
          "field": "Direction",
          "type": "nominal"
        }
      }
    }
  ],
  "encoding": {
    "theta": {
      "field": "Theta",
      "type": "quantitative",
      "stack": false
    },
    "theta2": {
      "field": "Theta2",
      "type": "quantitative",
      "stack": false
    },
    "radius": {
      "field": "Sum of Count",
      "scale": {
        "type": "sqrt",
        "zero": true
      }
    },
    "color": {
      "field": "Direction",
      "type": "nominal",
      "legend": true
    }
  }
}

[dm-p]

Hi @lbendlin,

If you want to explicitly bind the theta (and theta2) values to the radian representation from the data, you will need to use the value definition in the encoding (which maps the explicit value). If you bind a field, Vega-Lite will work out the angle based on the range of values from minimum to maximum (although you could probably manually set the range and domain on the scale to match the radian value, if you specifically wanted to still use the field definition).

Not sure if this is exactly what you're after but may get you closer:

{
  "data": {"name": "dataset"},
  "layer": [
    {
      "mark": {
        "type": "arc",
        "innerRadius": 20,
        "stroke": "#444"
      }
    },
    {
      "mark": {
        "type": "text",
        "radiusOffset": 50
      },
      "encoding": {
        "text": {
          "field": "Direction",
          "type": "nominal"
        }
      }
    }
  ],
  "encoding": {
    "theta": {
      "value": {"expr": "datum['Theta']"},
      "type": "quantitative",
      "stack": false
    },
    "theta2": {
      "value": {"expr": "datum['Theta2']"},
      "type": "quantitative"
    },
    "radius": {
      "field": "Sum of Count",
      "scale": {
        "type": "sqrt",
        "zero": true
      }
    },
    "color": {
      "field": "Direction",
      "type": "nominal"
    }
  }
}

Cheers,

Daniel

[dm-p]

Updated version with a ThetaMid column (which is just the angle value in radians for the text mark):

{
  "data": {"name": "dataset"},
  "layer": [
    {
      "mark": {
        "type": "arc",
        "innerRadius": 20,
        "stroke": "#444"
      }
    },
    {
      "mark": {
        "type": "text",
        "radiusOffset": 50
      },
      "encoding": {
        "text": {
          "field": "Direction",
          "type": "nominal"
        },
        "theta": {
          "value": {
            "expr": "datum['ThetaMid']"
          }
        }
      }
    }
  ],
  "encoding": {
    "theta": {
      "value": {
        "expr": "datum['Theta']"
      },
      "type": "quantitative",
      "stack": false
    },
    "theta2": {
      "value": {
        "expr": "datum['Theta2']"
      },
      "type": "quantitative"
    },
    "radius": {
      "field": "Sum of Count",
      "scale": {
        "type": "sqrt",
        "zero": true
      }
    },
    "color": {
      "field": "Direction",
      "type": "nominal"
    }
  }
}

Cheers,

Daniel

[ToriSugden]

Hi Daniel,
I have been trying to extend this. I would like to colour code by wind speed (which then gives me labels for each one) and have the directions equidistant from the centre. I have created two charts one on top of the other. But wonder if there is a way to do it in one visual. I would also like the colours to be in wind speed order

[lbendlin]

Thank you for the insights. Need to bottle this up somehow. You're not planning to write a book or something on the finer details of data binding by any chance?