Making Weather Personal

“Weather is a literary specialty, and no untrained hand can turn out a good article on it.”
    Mark Twain

Outdoor photography and weather go together like thunder and lightning; you can't have one without the other, and one of them can kill you. To avoid Mark Twain's truism above, this article isn't about weather (or photography, either). It's more about how we obsess over weather, whether or not we can do anything about it. We watch TV weather reports, we have weather apps on our phones, we visit various weather-related Web sites, we listen to weather-related podcasts, comedians joke about it, we talk about it when we have nothing better to talk about. And we complain about it. Endlessly.

We also complain about weather forecasts, mainly about how bad they are. Yet we often make a point of tuning in the local TV news program, more to hear the forecast than the “news.” Where I live in western Montana we have “small-market” TV, where the on-air staff isn't well paid, is mostly young kids who move on to better markets when they demonstrate a bit of talent, and where low-budget commercials are repeated, often back-to-back, because the stations' marketing folk aren't very good at selling air time. The weather forecasters, whom I call “TV Weather Kids,” (TVWK) are fun to watch. When new to the job they are uniformly terrible at ad-libbing, which is mostly what TV weather personalities do. The kids usually grow into it as they gain confidence and experience. In some cases they become quite good; when we see that happening we know they'll be moving on to larger markets before long.

As for other forms of consuming forecasts, we have apps like AccuWeather (which my brother-in-law in Michigan accurately calls “InAccuWeather”). Here, and, apparently, where bro-in-law lives, AccuWeather's accuracy leaves a lot to be desired, but to make up for that, each updated version of the app seems to bring an increasingly weird user interface. Maybe the theory is, if you can't present good data, present it in a confusing or oddly-formatted and graphically bizarre way.

For some time I had the Dark Sky app on my Android phone, and I liked it. Then Apple bought it and the Android version is no more.

Personal Weather Stations (PWS)

For many years I've had a simple La Crosse weather station. It's a basic, monochrome LCD panel displaying time, indoor and outdoor (via wireless sensor) temperature and humidity, and barometric pressure. It shows real-time numbers and stores minimum and maximum values until manually reset. It eats batteries for a living. I also have a simple tube-type rain gauge with read and empty powered by me, no batteries required. I check the rain gauge in the morning when we've had rain, which is not a regular or frequent event (typically about 14 inches (36 cm) annually, most of that coming in May and June). For many years I've recorded the day's min and max temperatures and rain or snow totals, in little calendar books (very old-school). We sometimes look up entries in those books along with my archived PotD to compare to the present.

The La Crosse weather display has been reliable and useful, but I've long wanted to upgrade. I'd like a PWS that measures what the La Crosse does plus rain amounts and wind velocity, provides a data-rich and nice-looking display, and automates some of the data collecting I've been doing manually. It'd be nice, but not mandatory, to be able to see this data remotely, when we're away from home. Over the last couple of years I've sporadically read reviews and user comments about a dozen different PWS. As with every other product category, each system had its strengths and weaknesses, but all seemed to omit one or more features I wanted; some even require a paid subscription to allow access to one's own data. Then, in late 2019, a friend sent a link to a crowd-funding site advertising an upcoming PWS from Weatherflow (WF), which has been around for a long time providing professional hardware and services to business and government customers. They also have some consumer PWS products. The new device, the Tempest, had not yet shipped, but the page had lots of good technical information. It's an integrated (that is, a single device), solar powered weather-hardened hardware package to collect the expected metrics, with unique systems for measuring wind velocity and rain totals and detecting lightning. It is completely wireless, solar-charging internal batteries and communicating with one's wi-fi router via an indoor “hub” device. The Tempest does not include any sort of display; users must provide a tablet or smart phone (iOS or Android) to set up a new unit. Once installed, the Tempest's data can be viewed with those same devices or on just about any device that can browse the Web. Tempest data can also integrate with Weather Underground's site, IFTTT (If This Then That) scripts, and more, including Alexa, Google Assistant, and Siri, and software from independent developers. There's an active, knowledgeable, and helpful user community. The Tempest sounded like a winner, but I decided to wait until real units had reached the hands of real users.

In August, 2020, I placed my order directly with WF, and a week later had my unit. While waiting that week I built the outdoor mounting I'd use, based on photos and other information I gathered about installing the Tempest. That worked perfectly, requiring only a few minutes (and a climb up on the roof) to install, after completing the easy set up of the Tempest, powering it up and getting it connected to my network via wi-fi.

Real-time Data and a Forecast

The Tempest (via its indoor hub) sends its data to WF's servers, which they call their “data processing center.” The app (or Web site) displays data from those servers rather than directly from the Tempest. WF's Web site has a nice graphic (top of page) that explains this. Pat and I have several iPads of various vintages, and one each mothballed Android and Apple smart phones in addition to the phones we carry, so we had plenty of devices to use as displays. I bartered some print work for a used Samsung Galaxy Tab 8 tablet, which replaced the old La Crosse panel on the kitchen counter. The only missing information now is indoor temperature, which, I've been told, may someday be resolved by a future product from WF. The apps are very nice, displaying current data, a forecast, and weather history both graphed and numerically. The forecast is derived from data aggregated from a number of sources and then manipulated through WF's machine learning algorithms, which they claim produces more accurate results than modeling alone. Because we get the data via the Internet (from WF's servers) we can see “our” data anywhere we happen to be.

So can you, on our station's data pages. Click/tap any of the horizontal “cards” (for example, the top temperature card) to see a graph of that data's history. Click (or tap) the little cloud icon at upper-right to see the forecast page. You can also see historical data and a map of nearby stations. Note that one can disable making the data public if privacy is a concern. I mentioned (optional) integration with Weather Underground, which I've enabled. There you can see our station's data (and many others) presented a bit differently. If I display a full month's data in table form I can copy and paste that into a spreadsheet, a once-per-month task which eliminates the need to record our daily readings on the calendar.

The Tempest's forecasts are dynamic, and fun (well, fun for this soft-core weather geek) to watch. These change hour by hour; in the morning the day's forecast may show an expected high temperature of 70° (F, in my case). By 3:00 pm the measured temperature might be 73°, and the forecast might then show an expected high of 75° at 4:00. Predictions for rain and wind are equally variable throughout any given day. I suppose it's a way to insure forecast accuracy: predict what's happening right now. In any case, it routinely beats our TVWKs' forecasts.

It's Not Perfect

The Tempest has been fun and useful, installation and setup were simple, it's as easy to use as glancing at a display, and it saves me the minor task of manually recording min/max temperatures and precipitation. But as always, there's room for improvement.

The haptic rain sensor has, so far, been less than accurate, consistently underreporting rain amounts when compared to my tube rain gauge, and also to a nearby MesoWest automated station. This is consistent with consensus on the WF user forums. According to WF, their RainCheck “Continous Learning” software improves the sensor's accuracy over time. To be fair, it's been extremely dry here since late June, so my station's not had much data to work with. I hope it improves, as it's been disappointing so far. And by its very nature it'll be useless when covered with snow and ice. Otherwise I've found little to complain about, and have been happy with my choice of PWS.

How About That Weather?

Our professional meteorologists and forecasters have billions of dollars worth of satellites, millions of dollars worth of supercomputers, and decades of historical data, but they still can't reliably tell you if it'll rain tomorrow. Maybe they need better satellites. Maybe they need better computers. Maybe they need better people. Maybe they just need to look out a window.

September, 2020

All products and brand names mentioned are trademarks or registered trademarks of their respective owners.