Dear Paloma,

I guess it depends on what precisely you mean by "automated". As far as I can see, a great deal of weather stations are already running at least semi-automatically by employing automatic, continuous data logger etc. Many can also be managed and programmed remotely. However, as with all machines, there will always be a at least a minimal degree of on-the-ground human maintenance necessary. I hasten to add, though, that at the end of the day, I still see a lot of value in the "human eye" and - as relates to forecasting, issuing warnings etc. - in the "gut feeling" of the meteorologically trained observer/forecaster and his or her long-time experience in "reading" the weather. He/or she will be well advised to combine both models and personal outlook. But I guess your question pertained more to data collection, rather than forecasting...

Thanks,

Julius

To add to Julius, automated weather stations need periodic maintenance (insects inside boxes, change dessicant for maintaining low humidity for electronics, troubleshooting power/sensor/communication issues, etc. Then there are open source instrumentation solutions too that may require a bit more maintenance and keeping up with changes in tech. Much of the world though does not have a good network of automated weather stations because of cost/logistics, and indeed an ideal model would be a mix of automated stations and attended stations ( like schools in remote areas).

To add to previous answers. You can get a good picture on number, quality and use of in situ surface and upper air observations in several global NWP centres on WMO data monitoring website: https://wdqms.wmo.int/

Hi Paloma,

Great question! There are certain weather parameters that are handled just fine by a machine and some that are not.

The basics that are very well handled by a machine include: temp/dewpoint, wind direction and speed, peak wind, and pressure (all three).

Machines are a DISMAL failure at ascertaining sky condition though. Evaluating how much of the celestial dome is covered by clouds, and at what level those clouds are, is best done by a trained weather observer.

Currently, machines are also pretty bad at freezing precipitation. Though they can do an ok job at detecting freezing precip (not perfect by any stretch though), they can't measure ice accumulation due to freezing precip (or fog).

They tend to do ok, but not great, at liquid precip measurement also. The tipping cup style is especially prone to dust build up that then turns to mud with the first rain. The weight of the dirt in the cup means that it tips (empties itself) too soon which results in a wrong precip amount reading.

It will be interesting to see the trends in obs, especially here in the United States, after about 2005-ish when many weather stations went automated (AOS). The sky conditions will be radically different from previous observations that were done by a human. I suspect this will throw off climatology dramatically!

If I had a choice, I would mandate human observing.

A little background on me...20 years USAF weatherman with thousands of observations and forecasts around the planet in very diverse conditions. I've set up numerous automated weather sensors and serviced them as well.

Hope this helps Paloma. Let us know where you're going with this...

I second John's point of certain weather parameters being best observed and recorded by the naked eye of a trained observer. I think this is true for percentage of cloud coverage in different atmospheric layers as well as identification of cloud types. As I'm sitting here outside on a windy day, I can clearly see Cumulus clouds moving at high speeds from an almost perfect westerly direction and with quite a low base (perhaps 2500 to 3000 m or so above ground). I am not sure if an automated machine could or would produce such a reading. In relation to the measurement of overall precipitation volumina over, e.g., a 24 h period I tend to be perhaps a little more optimistic than John. But I agree that it gets more difficult, the more one moves to precipitation types such as snow, icy rain etc. I'd further add visibility (maximum range/distance of visibility) as another parameter which automated machines may find difficult to correctly identify. Out on sea, wave heights are also quite trick, I think.

Best,

Julius

How many advantages can imagine from the Meteosat Third Generation (MTG) as concerns flexible combined imager and in particular the LI Liaghtning Imager? Do you think it should be most valuable cloud to cloud lightning, and lisghting at the top of the cloud so valuable as cloud to grounf ligtning detection? Can you imagine a lot of spacial trips lookig for the LI outputs at the top of e.g a cumulonimbus? Please explain us diferences and innovation of LI in 2023 from the actual ground lightning detectors. (MSG doesn´t have LI in the SEVIRI) EUMETSAT researcher answers are also reguired. Many thanks in advance

Excellent addition on the topic of visibility Julius!! I forgot to mention it.

Machines do an ok job with vis.

We've relied on machines for runway visual range (RVR) for decades and they do a VERY good job. This is very much a specific and highly localized usage as the machine measure light transmissivity (aka transmisometer). Often a fog bank will obscure one portion of a runway (or airfield complex) and not the entire runway. If the RVR machine just happens to be in the fog bank, then you're in luck! If not, well, this is why the old school rules and procedures were in place.

Modern visibility measuring devices use the same principle and are subject to the same limitations. They are HORRIBLE at determining actual visibility. I've seen obviously reduced visibilities well under 7 miles (9000 meters or less), but above 1/2 mile (800 meters) and the machine is reporting 7+ miles (9999+)!

This too WILL alter climatology and make it seem that fog is occuring less and visibilities are great! When of course it ain't.

Again, some weather observation parameters are simply NOT appropriate for a machine to measure.

Hello Paloma,

As to your question about lightning detection in general...

Check out the GOES-R Quick Guide on "Geostationary Lightning Mapper: Definitions and Detection Methods"

https://www.star.nesdis.noaa.gov/GOES/documents/GLM_Quick_Guides_May_2019.pdf

This is an excellent summary of the visible flash detection capability of the -R series GOES birds. It is FAR superior to ground based detection systems in my opinion. I have no empirical data to back this assertion up, but I have dealt with a variety of ground based systems for 20 years and they are problematic.

Ground based systems usually only detect cloud-to-ground strikes. Not always, and some of the cheaper hand-held units are actually pretty good at short range (under 20-30 miles or so). Any networked systems are subject to going down (usually when you need them the most!) Additionally, since most of these systems are solid-state, maintenance is usually non-existant...on the sensors themselves. The network that sends that data out to the world IS a maintenance item though, and it usually gets neglected. You simply don't know its down or glitching until you're expecting to see lightning data pop up and it doesn't. Terrestrial units / systems I think are just ok, but should be used as backups or supplemental to the GOES lightning mapper (primary).

Since the GOES-R series sat birds detect the actual visual flash (due to intensity / over saturation of the sensor), they detect all versions of lightning: Cloud-to-cloud, cloud-to-air, and cloud-to-ground. The graphical output products are amazingly useful for forecasting and climatology purposes.

Hope this helps!

Many thanks for expanding on the point of visibility, John! It's interesting to hear about this parameter from an aviational point of view. I've been mostly familiar with visibility issues from seafaring (where it is obviously very important) as well as from our German Autobahnen. Public weather reports used to include a note on visibility here (if it was relevant during the reporting period), but I've heard it being mentioned less frequently ober recent years. It is luckily still included in the ocean weather forecast, though. Best, Julius

The role of meteorlogist is still very important even with fully automated instruments. People operate, analyse, interprete, make maintenance, give their sense. In Iraq, the meteorological organization works in the least level of skill. It does not launch radiosonde on a regular basis, does not have its own numerical weather prediction models, and does not have any remote sensing measurement ( radar, satellite, lidar, etc).