Pressure stair step

[wd40]

Does everyone gets the small Pressure stair steps from the fine offset pressure sensor about .01inhg?

Randy

[steve]

The Fine Offset measures to 0.1 mb, which Cumulus then converts to inHg when required, to the nearest 0.01 inHg. Is that what you're referring to?

[wd40]

Here is the pressure chart showing a stair step like sensor output. I wondered if all the fine offset pressure sensors put data like this.

[Ned]

That looks typical of what I get when zoomed in. There is a fair bit of 'hunting' with the FO pressure sensor.

[steve]

Here is the pressure chart showing a stair step like sensor output. I wondered if all the fine offset pressure sensors put data like this.

Yes, it's because of the 0.1 mb resolution which Cumulus translates into a 0.01 inHg resolution. It's the same with all the data, effectively; they all have a particular resolution, so you'll never get a completely smooth graph. I'm hoping to provide an option to draw smoothed graphs from the data.

[nking]

I'm hoping to provide an option to draw smoothed graphs from the data.

That would certainly be great for the average wind direction graph - but I imagine this is less likely.

[steve]

That would certainly be great for the average wind direction graph - but I imagine this is less likely.

The wind direction graph is a problem, because it has to be discrete points. If I make it a line graph, then stupid things will happen when the wind is fluctuating around the top and bottom of the graph. If anyone has any ideas about that...

[Gina]

The wind direction graph is a problem, because it has to be discrete points. If I make it a line graph, then stupid things will happen when the wind is fluctuating around the top and bottom of the graph. If anyone has any ideas about that...

In my Linux software I create 8 graphs of the wind direction, each one centred on a main compass point. Then I use the results from a 6hr wind rose analysis to choose the one with most values near the middle line (in the last 6 hours) and upload that. It's not perfect and when the wind direction changes a lot, the older values tend to wrap around the top/bottom; and it's complicated, but it's the best I've come up with so far. If anyone has any other ideas I'd be very interested.

This is a plot of my last 24 hours wind direction using this approach. Here the points are joined up by a simple line. For longer time periods I just use the step type of plot as the wind direction usually varies too much to do anything else.

24hrs-wind-dir.png

[wd40]

Always interesting answers to my questions.

While I was getting a copy of the graph I realized I could right click on the graph and change line colors. That is nifty.

Randy

[Meteoalmoster]

That would certainly be great for the average wind direction graph - but I imagine this is less likely.

The wind direction graph is a problem, because it has to be discrete points. If I make it a line graph, then stupid things will happen when the wind is fluctuating around the top and bottom of the graph. If anyone has any ideas about that...

I think that you can somehow work it out by making the line go "past" the edge of the graph and then "reappear" at the other end. Imagine that the chart center is 180º, and that the top and bottom of the chart represent for instance S direction, that is both 0º and 360º.

Then before drawing the line that joins two consecutive points, you need to determine which of the two "paths" is shorter: the straight line that simply goes from point i to point i+1, or the line that goes "around" the chart, that's to say from point i to the closest edge and then from the opposite edge to point i+1.

For instance, if point i has a wind direction of 320º and point i+1 of 170º, then you calculate 320-170=150 and 360+170-320=210. In this case, the straight line is shorter and therefore this is what you draw.

However, if the two consecutive points are respectively 350 and 50 then 350-50=300 and 360+50-350=60, what means that you draw a line from point i to the top edge and then another "continuing" line from the bottom edge to point i+1.

I'd say that most of the time this should render a reasonably readable chart, although obviously there might be days when, if the wind would oscillate, in the example, aroound the S direction, the chart would appear kind of "empty" in the middle. This could be minimized by making the center of the chart coincide with the dominant direction in your area, though. What do you think?

[steve]

I think that you can somehow work it out by making the line go "past" the edge of the graph and then "reappear" at the other end. Imagine that the chart center is 180º, and that the top and bottom of the chart represent for instance S direction, that is both 0º and 360º.

Thanks for the suggestions, but Cumulus doesn't draw the lines. It uses the Delphi 'TeeChart' charting component.

[nking]

I guess you can't have a line graph for the average value only as that is less likely to jump around (unless the avg. oscillates from e.g. > 0 to < 359 deg!).... certainly no easy solution.

[mcrossley]

Plot it on a circular chart with time on the radial axis?

[Gina]

Say, starting in the middle and moving outwards with time? Yes, that's a thought - think I'll try that - should be interesting

[mcrossley]

Gina

I just mocked one of these graphs up, the latest plot is on the outside - no background image at the mo, but you can see the pattern shift towards the centre with each update.

http://weather.wilmslowastro.com/winddi ... r_plot.htm

[I'll take this down after a couple of days]