Visualisation of data on the map - best practices

[DenoBeno]

Related to: https://github.com/clarity-h2020/data-package/issues/8#issuecomment-542592439_

We are often running into a problem with data visualisation. Two main reasons:

1. Scale.
2. Colour coding.

Recommendations:

1. Use the same scale and color coding for all the variants of the same indicator, at least within one study. E.g._
   • The "number of heat days" hazard indicator should have the same scale for the past climate and for the various future climate predictions, for 1 year, 5 year and 20 year frequency.
2. Assure that the scale has high contrast so that the users can recognise the changes on the map. E.g.:
   • scale that only uses yellow/brown/red is less usable than one that also uses green and blue
   • scale that uses blu(ish) color at two ranges of values will confuse the users.
3. Choose the scale wisely, so that:
   • the "interesting" parts of the continent do show some cahanges of color. For example, all of France should definitely not be the same color.
   • the difference between tome periods, emission scenarios, emission scenario etc. should be visually obvious.

[DenoBeno]

Some examples of badly chosen scales:

1. Hot days: almost half of the scale is bluish. Furthermore, the two blue strips can't be differentiated.

Past climate Worst case future

2. Hot days, standard deviation: scale is way out of bounds. Anything beyond 10 is probably to be considered "junk data", so the contrast should be somewhere in the 0-10 or maybe 0-2 scale.

3. No green & blue colors used => less contrast available

[DenoBeno]

For EMIKAT, I have noticed that scales for the heat indices are not good. E.g. the Ageos dimitrios study:

historic data, yearly events

• T_MRT = 74 to 103 °C
• T_UTCI = 51 to 58 °C
• T_amb = 35 to 41 °C Worst case, 20 years events
• T_MRT = 86 to 114 °C
• T_UTCI = 63 to 70 °C
• T_amb = 43 to 49 °C

Forr all three indices, the scale is the same, 20-100°C, although T_MRT never goes below 74°C and T_ambient never above 50°C. Consequently, almost nothing is seen on a map.

E.g. T_UTCI:

The green-yellow-red scala used here is not bad by the way. We could use this and add lila & blue below green and brown above red for even more contrast.

[LauraMTG]

Hi @DenoBeno

Considerations

The initial selection of the scale is certainly random, since its availability in geoserver was massive. I will now make some points that are already being taken into account:

1- Use the same scale and color coding for all variants of the same indicator, at least within a study. Make sure the scale has a high contrast so that users can recognize changes in the map. E.g: Yes, these two points have already been started. You can check this in the following layers: HI_summer-days... Two base legends have been made for media and desviation layers that are applied to the whole equally for ease of comparison. 2- the scale that only uses yellow / brown / red is less usable than one that also uses green and blue. However, when the variable that is represented alludes to hot days, the green or blue color is really confusing and not advisable. Therefore this recommendation is not always applicable.

3-for visual examples 1 and 2, these are not definitive scales. these layers must be updated with the new mask we are considering. Then we proceed to its revision and adjustment of ranges.

4-I will consider greater contrast for stakeholders in next steps.

[DenoBeno]

ok, thanks

On Mon, 21 Oct 2019, 16:55 LauraMTG, notifications@github.com wrote:

Hi @DenoBeno https://github.com/DenoBeno

Considerations

The initial selection of the scale is certainly random, since its availability in geoserver was massive. I will now make some points that are already being taken into account:

1- Use the same scale and color coding for all variants of the same indicator, at least within a study. Make sure the scale has a high contrast so that users can recognize changes in the map. E.g: Yes, these two points have already been started. You can check this in the following layers: HI_summer-days... Two base legends have been made for media and desviation layers that are applied to the whole equally for ease of comparison. 2- the scale that only uses yellow / brown / red is less usable than one that also uses green and blue. However, when the variable that is represented alludes to hot days, the green or blue color is really confusing and not advisable. Therefore this recommendation is not always applicable.

3-for visual examples 1 and 2, these are not definitive scales. these layers must be updated with the new mask we are considering. Then we proceed to its revision and adjustment of ranges.

4-I will consider greater contrast for stakeholders in next steps.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/clarity-h2020/data-package/issues/53?email_source=notifications&email_token=AAWTC7QVHAX4F7UHECLAO43QPW7E7A5CNFSM4JC32DL2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEB2TQUI#issuecomment-544553041, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAWTC7Q4557V6BPOTM2TX7LQPW7E7ANCNFSM4JC32DLQ .