Blog

Humidity Adjustments

It appears our humidity sensor’s maximum value has lowered over time. This is, unfortunately, common with our particular weather station’s sensor. The process is accelerated by 24 hour fan aspiration. We’ve seen this behavior in 3 other units we’ve deployed in the past.

When we first installed this sensor last July, the top possible humidity reading was 98%. The ceiling has since lowered to 96%.

After a round of tests using an independently calibrated device yesterday & today, and in an effort to stay within +/- 3% of true ambient humidity, we’ve programmed adjustments to the following humidity readings:

  1. 94% +2% = 96%
  2. 95% +3% = 98%
  3. 96% + 4% = 100%

Because we use a 5 minute rolling average for humidity at times you will see values between 96%, 98%, and 100%.

The good news is, based on experience with this sensor model, the biases should now be fairly solidified and predictable.

The bad news is that when humidity starts to descend from persistently moist conditions (>95% and >2 hours) the sensor can lag because of moisture uptake within the protective housing. Our programmed offsets exaggerate this process. We believe it’s more important to properly indicate the arrival of fog & dew than the drier air that immediately follows.

As always, though, we will test to verify sensor performance every so often and adjust accordingly.

****

NOTE: We send corrected humidities to CWOP & Weather Underground.

UPDATE 5/28/19: After an overnight period of saturated air, discovered the sensor’s ceiling is actually 96%; corrected offsets.

Cloudy With a Chance of Old School

I have swapped out the somewhat simplistic cloud icons that previously served as a background for navigation links with hand-drawn clouds. My wife and I both feel these new icons provide more visual appeal & warmth. Then again we like anything that reminds us of a mid-century movie poster.

I’m aware that using pictures for links is considered passe by most designers now.  So-called skeumorphic interface design was popularized in early Apple products decades ago. Round about 2012 everything started going “flat”.

As for this web app, we have crowned function king. The design is largely fit to purpose with a few embellishments here & there.

Did I mention I’m not a “designer”? Oh well, add that to the list of my faults.

***

The cloud images were acquired from Iconfinder & appear courtesy of a Creative Commons license. The artist is “Azuresol”. I removed some snowflakes under them since I wanted to use them year-round. I also created a variant that is lighter. A darker icon indicates the current page while a lighter icon links to other pages.

New Rain Gauge

Today I installed a new tipping spoon automatic rain gauge that will report real-time rainfall to the website. This is described by the manufacturer, Davis Instruments, as an “incremental improvement” over the previous tipping bucket gauge.

For the most part, our automatic gauge has run very close to our 4″ reference gauge which is 6 ft away & 2 ft lower. However, sometimes — particularly during large rain events — we encounter under reports of approximately 25% to 50%. Discrepancies can appear at any time regardless of wind speed or direction. Many others have reported similar, inexplicable anomalies with the old gauge.

Astute, long-time readers may notice the new “Aerocone” funnel in the picture. Supposedly, this cuts down on rain-induced measurement errors by reducing turbulence where rain is collected.

We’ll see how the new gauge performs. Early side-by-side comparisons from other station owners with reference equipment are encouraging. I’m definitely in the “trust but verify” camp nonetheless.

[UPDATE 5/20/19: The new gauge is performing marvelously so far. This weekend the tipping spoon reported 2.70″ of rain. Our reference gauge caught 2.67″. That’s about a 1% departure. Given the distance & design differences between the two gauges, small variances are normal.]

Minor Site Updates

You will no longer see the spinner image when the home page is refreshed. Previously iOS (iPhones/iPads) would sometimes display a blank page unless a slight delay was created while refreshing a page from device memory. The image let visitors know something was happening — no need to refresh manually which could make it worse. It appears that bug is gone now (knock on wood). So I have removed the image.

On the maps page, you’ll notice a couple changes. First, I had to replace the “Regional Radar” with a different image because the previous one hadn’t been updated in awhile. I had to completely remove the Surface Maps image for the same reason; however, there isn’t a freely available alternative available at the moment. If anyone would really like to see a replacement for a frontal map, let me know. Otherwise, I will leave everything as is. (Update 5/10/19: Both maps have been returned to service after being frozen for 3 days. We’ll keep an eye on them.)

The loop images on the maps page will now play continuously. I had to eliminate the play/pause button because on certain devices it was preventing the image from loading. It also seemed to slow down the loading process and sometimes interfere with the “user experience”.

I’ve addressed an issue where previously on iPhones the display would zoom when you selected an “Enhanced Radar” option. It should no longer do that (in portrait orientation at least).

I also fixed a bug affecting rain measurements on the home page that would only happen under rare circumstances. The malfunction also created incorrect totals for “Yesterday’s Rainfall” in the Almanac section, but only when metric units were selected in the page settings.

Finally, I’ve performed some back end improvements. The site now fully supports the HTTP/2 protocol resulting in better performance.

Snow Measurements On Hold

Due to a combination of intense recent snowfall, high winds, and limited space, we have nowhere to place a snow board where it is not subject to significant drifting. The city has become a labyrinth of snow piles that powerful lake winds are merging and leveling.

Therefore, effective 11 AM Friday 2/15, we are suspending snow measurements indefinitely.

Once the winds subside and/or the peaks reduce/even out we will resume snow measurement. Given the forecast of sunny to partly sunny days this weekend and limited snowfall in the next several days, hopefully that will happen by next week.

Note: a liquid equivalent reading taken from our gauge will continue to appear in the precipitation tables on our History page.

[UPDATE Saturday 2/16/19]: I have relocated a snow board to be further from any potentially drifting high spots to the north and northwest (two of the most common direction for snow in the winter here). After knocking on wood, I am now reinstating measurements. I have added 2″ to yesterday’s snow total as a minimum estimate of what fell after 8AM.

 

Putting Weather Alerts On A Diet

Synopsis: There are too many alerts containing overblown/non-applicable impacts and unnecessary instructions. Continue reading to see which alerts will now be displayed on the home page and why.

****

I’ve grown weary of all the alerts from the NWS that either 1) don’t apply within the city limits of Marquette or 2) don’t meet legitimate criteria for issuance.

Take, for example, Saturday morning’s wind chill advisory where there were virtually calm winds for hours while air temps were above 0.

Or consider the recent winter storm warning (for 2/7/19) where most parts of the county got 5-7″ and published guidelines require 8″ in 12 hours or 10″ in 24 hours. This occurred despite a drying trend in the model guidance even before the warning was issued. In isolation, this wouldn’t be a problem. Unfortunately, we haven’t had a single storm that meets warning criteria this winter (at least not in Marquette). Yet how many warnings have we had? I’ve lost count.

I’m sure any decision can be justified as following the spirit rather than the letter of the law. To that I say “if you can bend the rules one way, you can bend them the other. ”

I think, in general, more deference should be shown toward the unflappable, resilient spirit of Marquette and the U.P. We do not need alerts every time more than 2″ of snow is going to fall. Marquette normally receives ~10ft of snow per year. That’s a steady stream of statements & advisories notifying us that it’s, indeed, winter.

Furthermore, and I’m sorry/not sorry for the grumpy old man rant here, we don’t need to be constantly told how to dress, drive, and survive historically unremarkable winter events. Anyone who is that ignorant or helpless here is unlikely to fully read and heed the information anyway.

I may be told there is nothing that can be done because of mandates from on high. God (D.C.) has ordained these things and, therefore, all things are in their proper order. Well, in fact, there is something I can do. As of earlier today, I have suspended all but the most life-threatening notifications from the home page. These include: tornado watches/warnings, flood watches/warnings and ice storm warnings. Those are, indeed, unexpected events that can catch people unaware and endanger them without proper notification. The rest can be handled by the forecast.

The area page will continue to show all alerts for two reasons. First, most alerts are for the area rather than the city which is the focus of the home page. Secondly, I realize not everyone will agree with my decision, and I believe in freedom of choice.

****

[UPDATE Mon 2/11/19 12:15PM]: After further consideration, I am going to allow Winter Storm Warnings to post on the home page.  This is because Marquetters frequently travel outside the city. At least as important, the alert serves as a heads up to check the forecast and prepare for something other than the standard inch or two of snow.

Our Wind Cups Froze

Like every other object in its path, Monday’s ice storm did a number on our anemometer (wind gauge). The pole to which it’s mounted cannot be retracted because… you guessed it. So nothing can be done but wait.

It’s going to take either a big gust of wind (> 25mph), some sunshine and/or warmer temperatures to free the spinning cups. Until then, wind speeds will read 0. With temps barely cracking 20 in the next 5 days, it might be awhile before it’s fixed.

[UPDATE 2/5/19 6PM]: High winds forecast for Thursday may provide sufficient energy to loosen the ice. Fingers crossed.

The directional indicator is currently turning as it should.  So if you see “N”, that means winds are coming from the north.

Just FYI, the National Weather Service says this is only the 2nd time in the 2000s that an Ice Storm Warning has been issued. It’s just not something we are accustomed to up here. That’s doubly true for February. We got about an inch of freezing rain out of that storm!

Adjusted Humidity Readings

We are now adding 5% when humidity readings reach 95% since that is the maximum our sensor will read. For many years Davis Instruments’ sensors have struggled to achieve 100% humidity. Unfortunately, the problem seems to worsen with time. When we placed our sensor in service last November 28th, it would reach 98%.

Also, we are adding 1% to 92% (to make 93%) , 2% to 93% (to make 95%), and 4% to 94% to reduce the jump between 91% and 100%. Tests demonstrate these adjustments closely approximate reality.

You may occasionally see other readings besides 93%, 95%, 98%, and 100% due to 5 minute averaging of humidity values.

Raw values are available on the stats page.

Here’s some more info on this problem if you’re interested:

We have a backup weather station for emergencies. It also contains a Davis VP2 station with an SHT31 sensor that is less than 2 years old. It maxes out at 96% humidity, but it also has an average +8% bias below 80% during the warm season (May – Sept). So it has a wet bias in the middle range in addition to a dry bias in the top range. Keep in mind that the maximum error for this sensor is supposed to be 3.5% according to the manufacturer.

We are not the only ones to experience these issues. Many other Davis VP2 weather station owners have reported them (in a thread approaching 1200 posts). Davis says they are “looking into it”, although that was almost 2 months ago now. Some particularly brave souls have attempted to informally wire up older (SHT11/SHT15) and/or unsupported sensors (SHT75) with varying degrees of luck. We’re not keen on performing ad hoc experiments with our primary temperature/humidity sensor. Some combination of Murphy’s Law and common sense would argue that such slapdash “solutions” will inevitably break at the worst time. Indeed our various attempts with other sensors on the test bench have not gone well. Therefore, we await a viable, proven fix preferably from Davis Instruments.

[UPDATE 9/21/18: Most of those enterprising individuals referenced above are now indicating an array of problems with the SHT75 sensor ranging from sudden failure to gross errors. It’s uncertain whether the fault lies with the sensor or the implementation. Either way, at present there appears to be no good workaround to the issues mentioned above with the SHT31.]

[UPDATE 10/23/18]: We recently swapped out our SHT31 sensor suite as part of a preventative maintenance routine. The new humidity sensor arrived with a +6% average (wet) bias below 55F, an average +4% bias above 55F and a maximum reading of 98% (after several hours in saturated air). We have corrected these deficiencies with custom programming.

Noticed Some Changes Lately?

Last week I tweaked the color scheme to match the forecast icons. We’re keeping with the sky/water/sand theme only now just a bit brighter and more coherent.

Today, I removed the weather history icons that lived beneath the current conditions. Nobody was clicking on them. The images accounted for about 1/3rd of the data necessary to load the home page. That wasn’t a good tradeoff between useful information and bandwidth consumption. I’m always mindful that people have limited data on their mobile plans. Speed is another concern too.

I also moved the dropdown selector that was in the conditions section to the page settings (gear wheel icon in the upper right of home page). That’s where it belongs rather than cluttering up the current conditions. If you need to change the frequency of updates, it’s just one click to get there. Again, the data shows that few people adjust that setting, so why not put it where power users naturally go anyway?

One more tiny detail. I added the time zone to the conditions and forecast update times. When visitors travel, they shouldn’t have to wonder if the times listed are local time or Marquette time. The forecast will always be updated in Eastern time. The conditions will be updated according to the current time zone selected on your computer or mobile device. In Marquette, they should both show “EST” or “EDT” depending on Daylight Saving Time.

Also, I should mention that we recently changed hosts about 3 weeks ago. Regular, long-time visitors may remember some extended outages lasting hours, even days in one instance last year due to some lackluster planning by our old host, A2hosting. I picked out another Michigan hosting company, Liquid Web.  Not only do they own their data center in Lansing (rare), but their operation is prepared to withstand power & internet failures, weather events, and DDOS attacks (coordinated hacking) among other unfortunate events. That’s not to say the site will never go offline, but, hopefully, not as frequently or for as long. It’s much more expensive to host with them, but their speed and reliability so far have been phenomenal.

Well, thanks for visiting. If you enjoy the site, please tell your friends & family!

What Happened To The Forecast?

The website’s featured forecast provider, Weather Underground (WU), announced a change in policy last month. They were no longer going to allow free, limited access to their raw forecast data (API). Instead, they would begin charging hundreds per month even to those who contribute data like we do. Since this site generates no revenue, the new price tag is a nonstarter.

I’ve been exploring alternatives in hopes of continuing to offer dual forecasts. None of them have met my requirements of accuracy and affordability. Therefore, I’ve decided the National Weather Service (NWS) will be the sole forecast provider for marquetteweather.com.

In addition, the hourly forecast will not be replaced due to problems inherent in the forecasting process. Presently, the software-based weather models do not adequately represent the transient effects of the Great Lakes frequently leading to significant errors exceeding our forecast target of +/- 3 degrees. Plus, the forecasts don’t update often enough to keep pace with our weather. Routine changes in wind direction, cloud cover or precipitation can transform conditions in a matter of minutes — as most Marquetters intuitively understand.  Continue reading “What Happened To The Forecast?”