|
|
Post by nei on Jan 16, 2018 0:14:56 GMT -5
for Umeå, 1997-2017. Will do a few more cities later. Darker the color, the more common the temperature is. The temperature bins are 2°C wide [-20°C to -18°C is the first bin, next one is -18°C to -16°C, etc]. Each vertical pixel represents an hour. As Babu has said, there's not much of a diurnal cycle in winter. The plot shows that just above and below 0°C is most common. Anything much warmer than that nearly never happens. Much colder, happens but not as often. the 3:00 pixels are roughly the same color-wise as 15:00, showing there's no diurnal cycle. Watches as it goes to summer and the diurnal cycle appears. Pixels in summer are darker than winter as there's less of a range so each 2°C bin is more frequent  Here's the monthly means for Umeå  Extreme is average hottest & coldest a typical month gets, 24 hour mean from averaging over all hours, monthly mean the usual (Daily Max + Daily Min)/2 |
|
|
|
Post by shalop on Jan 16, 2018 1:20:45 GMT -5
I would have thought that -20C is more common up at 63 degrees North latitude in winter, but it seems like most of the probability is overwhelmingly concentrated in the -10 to 2 C range. I guess Umea is moderated by the sea, I'm sure if you did Tavelsjo instead, there would be a bigger variance and lower means. Another interesting thing is the incredibly sharp dropoff at 2C in Jan-Feb (i.e. there is a heavy left tail but a light right tail; why?).
Anyways, it's a really cool density plot.
|
|
|
|
Post by Babu on Jan 16, 2018 2:23:50 GMT -5
Oh, and this is the University station and not the airport. Amazing work, Nei!
|
|
|
|
Post by Babu on Jan 16, 2018 2:35:03 GMT -5
Now we just need for Lommaren to add these into another weatherbox on Wikipedia  Oh and btw, what was the annual avarage extremes? |
|
|
|
Post by nei on Jan 16, 2018 10:08:24 GMT -5
I would have thought that -20C is more common up at 63 degrees North latitude in winter, but it seems like most of the probability is overwhelmingly concentrated in the -10 to 2 C range. I guess Umea is moderated by the sea, I'm sure if you did Tavelsjo instead, there would be a bigger variance and lower means. Another interesting thing is the incredibly sharp dropoff at 2C in Jan-Feb (i.e. there is a heavy left tail but a light right tail; why?). Anyways, it's a really cool density plot. Thanks. Umeå has a really variance is really high in winter; it's just skewed in one direction. Makes a probability density plot a bit misleading as each individual bin much colder than average isn't very high but the total % of times it's at or below a cold temperature is higher. % of times (by hour) Umeå is below a given temperature, just January and February: 4°C: 97.3% 2..81°C: 95.0% [two standard deviations above if a normal distribution] -1.44°C: 68.3% [one standard deviation above if a normal distribution] -4.44°C: 50% [median winter tempertaure] -8.08°C: 31.7% [one standard deviation below if a normal distribution] -15°C: 9.8% -18.1°C: 5.0% [two standard deviations below if a normal distribution] -20°C: 2.9% -22°C: 1.5% -25°C: 0.49% -30°C: 0.05% [running into sample size issues] The skew I'm guess is oceanic influence weather from the Baltic comes from a sea that's near freezing, so it's rather stable. Airmasses from the east or local subarctic weather can bring much colder temperatures if the setup is good but there's no easy way of making much warmer than freezing temperatures. Helsinki has similar pattern, and probably the whole baltic region. |
|
|
|
Post by shalop on Jan 16, 2018 12:28:35 GMT -5
^Ya, for sure Umea has a large variance in winter, I just meant that it'd (presumably) be even larger for an inland location like Tavelsjo, as much lower temps would occur with greater frequency, but more moderate ones would still occur as well (judging from Baba's posts in the photo thread).
Also, thanks for those percentiles. I guess "overwhelmingly" was inaccurate for the range -10 to 2 C. More like -20 to 4 C. I also think the skew would be present in most of Scandinavia.
|
|
|
|
Post by Steelernation on Jan 16, 2018 15:19:10 GMT -5
Can you do this for Rochester?
|
|
|
|
Post by AJ1013 on Jan 17, 2018 21:24:11 GMT -5
This is really cool; Can you do this for Miami Beach or tell me how to?
|
|
|
|
Post by nei on Jan 17, 2018 23:32:35 GMT -5
Alaska at a similar latitude has no diurnal cycle, but one is just starting
|
|
|
|
Post by nei on Jan 17, 2018 23:36:13 GMT -5
Steelernation AJ1013I found a site with hourly US data for download but it's not exactly an hour, once in a while an hour is skipped, some hours have more than one observation. I'll see if I can find a good "fix" so I can get a decent plot out of it this weekend. Think I've found a way, but don't feel like playing with at the moment.
|
|
|
|
Post by Babu on Jan 18, 2018 6:37:31 GMT -5
|
|
|
|
Post by nei on Jan 21, 2018 12:04:31 GMT -5
|
|
|
|
Post by nei on Jan 22, 2018 23:38:18 GMT -5
figured out to make the same plots for hourly [ASOS] US stations. Here's the nearest one to me, Chicopee. I rounded the number to the nearest °F and then binned by every 2°F, there were some odd looking patterns otherwise; charts aren't good enough for anything more precise but the pattern is more visible. I also have dewpoint, precipitation and skycover; so I could do more plots [maybe just clear hours? or just a dewpoint by hour plot?]  Could do plots for other weather stations; takes little time to do more once I got one. I'll do Miami and Rochester airports next. Did Hartford's airport which should be similar but slightly warmer than Chicopee as a check; they look similar enough to me. our diurnal cycles weakens in winter more than I expected even if the diurnal range doesn't. Nowhere as much as Umeå; summer looks very monotonous with little range in possible temperatures compared to winter. I cut it off at -6°F even if we get colder, because colder happens so infrequently [only last for a few hours on a few days] it'd be too faint to see on the plot. If you look carefully, in the warmer months there's a faint tail of much colder than usual sunrise temperatures but not earlier in the night, clearly radiational cooling. |
|
|
|
Post by nei on Jan 23, 2018 8:13:29 GMT -5
and for dew points of Chicopee  |
|
|
|
Post by nei on Jan 23, 2018 9:08:12 GMT -5
and here's Miami AJ1013. Used the airport, for some reason when I downloaded Miami Beach I got a blank document. I need a better color scheme, as it's very hard to see infrequent colors. You can kinda make out there's a skew in winter temperatures.  |
|
|
|
Post by nei on Jan 24, 2018 10:58:09 GMT -5
and another for Melbourne, my same source had 15 or so years of hourly data for Melbourne. Has skycover, no sunshine. If Beercules or jgtheone can think of anything else interesting to do with Melbourne data, I'd try. Doesn't look like hot nights are that frequent; this is from the airport station. Do my results look reasonable?  |
|
|
|
Post by rozenn on Jan 24, 2018 15:30:07 GMT -5
Awesome stuff nei! You can almost tell Miami's July averages by looking at the graph, impressive consistency. I imagine Nice would be quite consistent too.
|
|
|
|
Post by nei on Jan 30, 2018 19:07:01 GMT -5
Thanks! Here's Mt. Washington:  And Whitefield, 10 miles to the northwest at about 1000 feet. Great frost hollow, has the most range in temperature of any station I plotted so far; enough to make the -20°F to -30°F range light up  difference plots, implying area lapse rate. Since it's a frost hollow, not quite ideal for measuring lapse rates. It's a good mountain valley - summit measure. The x-ticks mark every 9°F (5°C) difference between the summit and Whitefield; 9°F = 3.13°C/1 km. Dry adibatic lapse rate is a 28.2°F difference, you can see a summit - valley difference much higher than above than doesn't happen; occasionally a bit more, guessing on sunny, calm dry days, when the ground heats faster than the air can mix. Dewpoint lapse rate shows much less variability, and with little seasonal variability. Seems more set by physics of the atmosphere? For temperature difference. Positive means Whitefield is warmer than Mt. Washington; negative Mt. Washington is warmer.  Notice dewpoints are always lower at the summit vs the valley.  tempted to do more difference plots. Either stations in the same region. Or the few mountains vs valley stations that have hourly data. |
|
|
|
Post by 🖕🏿Mörön🖕🏿 on Jan 30, 2018 23:28:41 GMT -5
Can you do Vancouver and Anchorage please?  |
|
|
|
Post by nei on Jan 31, 2018 15:09:06 GMT -5
. % of times (by hour) Umeå is below a given temperature, just January and February: 4°C: 97.3% 2..81°C: 95.0% [two standard deviations above if a normal distribution] -1.44°C: 68.3% [one standard deviation above if a normal distribution] -4.44°C: 50% [median winter tempertaure] -8.08°C: 31.7% [one standard deviation below if a normal distribution] -15°C: 9.8% -18.1°C: 5.0% [two standard deviations below if a normal distribution] -20°C: 2.9% -22°C: 1.5% -25°C: 0.49% -30°C: 0.05% [running into sample size issues] for Chicopee, MA. Cold was less common than I thought from the lows; perhaps because lows tend to occur for a few hours late at night / early morning 11.7°C: 99% 6.7°C: 95% 4.4°C: 90% 1.1°C: 75% 0.6°C: 68.3% [one standard deviation above if a normal distribution] -2.2°C: 50% -5.6°C: 31.7% [one standard deviation below if a normal distribution] -7.2°C: 25% -11.7°C: 10% -14.4°C: 5% -20°C: 1% |
|
