Isoline filtering is harsh

[did-g]

Hi, Currently isolines rendering is using every nth points which create artefacts:

while with all points

Options:

• Always use the full dataset, but it produces jagged lines.
• Spline lines.
• DouglasPeucker reduction.

Thoughts?

[rgleason]

Doesn't look that bad really.

[dominiquehausser]

I notice quite rarely artefact... Personally I will prefer the use of the full dataset (raw output of the model), creating the artefact. If we start to treat them there is a risk (probably low) to create new artefact or to modify "too much" at least partially the "original output".

[norulz]

I tend towards using more data points but with some smoothing. Looking at the "big" picture in the Atlantic does not justify the issue. In smaller geographic areas such as the Med West or East basins it is easier to recognise the synoptic pattern with smoothing. This can be seen by comparing GFS (which include smoothing) to ICON that does not.

Splines or reduction? -- Whatever works best and fast.

[dominiquehausser]

1 hPa between isolines 4 hPa between isolines (the most frequent presentation of met offices. Smoothing looks nicer, but ... What is GFS using for smoothing ? We should try to use the same to avoid a different bias between the 2 models.