January Banter String

[RevWarReenactor]

I think a trace to an inch is the best bet at this point.

Its driving me crazy the people who try to claim "this is normal" in terms of snowfall. Its not normal 2 to 3 years with nothing but a single storm to our name. Sure the pattern might argue that we really shouldn't be seeing much, but people try to make out like 6 inch snowstorms are a fluke in DC. They aren't.

[Lowershoresadness]

2 minutes ago, usedtobe said:

The bad thing is it the Euro EPS almost did it at 72 hours and supported the Euro idea of a 500hlow closing off to our south.  I used to have faith in the euro ensembles but now not so much.

 

 

4 minutes ago, usedtobe said:

The bad thing is it the Euro EPS almost did it at 72 hours and supported the Euro idea of a 500hlow closing off to our south.  I used to have faith in the euro ensembles but now not so much.

 

Wes it is 2018. The cubs and redsox have won world series. Drones can deliver packages to your door step. We have a a spacecraft that has left our solar system. Why on earth don't we have a super duper computer model that can be accurate as hell within a week???

[MN Transplant]

16 hours ago, supernovasky said:

I’ll be honest. I don’t feel that terrible. We had 22 players on IR, for the first time have a TON of cap space, and get like 5 rookie starters for pennies on the dollar the next 4 years. We have 2nd year Kamara and Ramczyc and Lattimore to look forward to. If we can get another good year or 2 out of Brees we will be back next year.

 

Good game to the 2-3 Vikes fans I spoke to here.

Kamara and Thomas are outstanding.  Bodes well for the future.

[usedtobe]

57 minutes ago, dtk said:

Yeah, it's a huge problem with the GEFS and we get complaints from the regions all the time.  There is lots of work ongoing to address this in the next GEFS upgrade (v12).  Interesting that it's being observed in the EPS for certain events.

Yes, at least in the two runs that closed off a 500h low to our south, one around day 5 and the other a 72 hour forecast.   The EPS mean also closed off a center. The individual members were more dispersive but the really amped up 1/3 outweighed the other 2/3 or at least that is what is looked like.  It's nice to see the GFS being closer to the truth or seemingly closer to the truth (we won't know for sure until Wed) but having both model ensembles exhibiting herd instincts along the east coast makes it tough too really hone in on a forecast with much lead time unless both herds stick together for several runs. 

[Bob Chill]

1 hour ago, showmethesnow said:

Yeah, it has been a very good run. But the market is substantially over valued at this point and when the correction comes I think we are going to see a steep adjustment. In many ways this reminds me of the housing bubble just a few years ago.

A combination of technology and too many dollars chasing too few assets will keep throwing infinite logs on the fire until an unexpected monsoon comes in and puts it all out. Then the cycle will start right back up again. In some ways it's long term disaster or crisis proof at this point. Yea, a monsoon is going to sucker punch a lot of poor saps in the face but in the grand scheme it's an infinite money machine. I used to worry. I don't worry at all anymore. 

Plus I have a really good life insurance policy so if my old ass is broke when I die then my kids still win unless I live too long. If I see that possiblity happening then I'll just start drinking again. 

[usedtobe]

6 minutes ago, Lowershoresadness said:

 

Wes it is 2018. The cubs and redsox have won world series. Drones can deliver packages to your door step. We have a a spacecraft that has left our solar system. Why on earth don't we have a super duper computer model that can be accurate as hell within a week???

DTL could answer it better than me.  There are loads of issues, you need to parameterize physical processes that occur at scales below which are resolved by the grid, no matter how many obs you have, there is no way to know perfectly what the initial conditions are.  The equations that describe the atmosphere in the models are non linear so small differences can amplify with time pretty dramatically.  Heck, as you move towards higher resolution how you do the math might even bias the forecast a bit (at least I think that is true).  I have a slide from a presentation that deals with those issues.  The remarkable thing is how good models actually are.  Last year I wrote and article for the Captial Weather Gang on how much model forecast and forecaster tools have improved since the 1970s.  It's pretty remarkable.

[Jandurin]

Yeah, it has been a very good run. But the market is substantially over valued at this point and when the correction comes I think we are going to see a steep adjustment. In many ways this reminds me of the housing bubble just a few years ago.

First bit coin

[Lowershoresadness]

16 minutes ago, usedtobe said:

DTL could answer it better than me.  There are loads of issues, you need to parameterize physical processes that occur at scales below which are resolved by the grid, no matter how many obs you have, there is no way to know perfectly what the initial conditions are.  The equations that describe the atmosphere in the models are non linear so small differences can amplify with time pretty dramatically.  Heck, as you move towards higher resolution how you do the math might even bias the forecast a bit (at least I think that is true).  I have a slide from a presentation that deals with those issues.  The remarkable thing is how good models actually are.  Last year I wrote and article for the Captial Weather Gang on how much model forecast and forecaster tools have improved since the 1970s.  It's pretty remarkable.

Thanks Wes. Look into your crystal ball for a min. Our society seems to technically advance faster than humans can catch up. How accurate do you see weather models getting in say 20 years from now. What advances have to be made to achieve this?

[psuhoffman]

2 hours ago, dtk said:

It's all good.  If someone wants to call me out for coming off as a jerk, that's fine.  It wasn't intentional and I'm not that bothered by it.  However, I stand by the basic premise.  It's pretty crazy to me that people are suddenly calling the Euro "useless".

I don't mind you're style but I can be that way too sometimes. People are too sensitive sometimes. And you accept discourse so it's fine imo. 

You are right to call my claim it's useless silly because without qualifying that it is. But I do think right now, for whatever reason, the euro has been struggling and throwing out way over amped solutions with too much frequency that for the very very specific purpose of determining the strength of a specific threat in the medium range it's lost its value to me. 

This is a very narrow thing but it's so all over the place lately with every threat that I would have been better off ignoring it. It would have added less noise and more confidence in what actually happened. Several times the euro has simply head faked us with crazy tangent runs. 

Im sure that will change and this is probably a temporary pattern specific issue. So I'm not trashing it permanently. And it continues to have usefulness in a broader sense in general pattern identification. But for my purposes it's hurt more then helped for the last month or so. 

I realize this is not an objective way to score a model. That the euro might have nailed 80% of the pattern and weather and scored better than the gfs but we also know what most of us care about is snow and for the 4 or 5 snow threat it's been a train wreck.  It's also not objective because from run to run overall maybe it scored ok but to me stability matters and when it has 2 or 3 crazy stupid runs that head fake us during the 5 day run up to a storm that bothers me more than if it's just a bit off consistently. I can manage that by compensating for how I imagine it will correct better then when the thing puts a 980 low over ACY when in reality it's gonna be a weak 1005 wave off the outer banks. 

So a lot of this is personal preference. And it's not objective. And it's uber specific. But that's what I meant by it's useless. I shouldn't be as flippant though. 

[dtk]

7 minutes ago, Lowershoresadness said:

Thanks Wes. Look into your crystal ball for a min. Our society seems to technically advance faster than humans can catch up. How accurate do you see weather models getting in say 20 years from now. What advances have to be made to achieve this?

We continue to gain about about one day of lead time per decade of research and development.  We are getting pretty close to having some breakthroughs in data assimilation to better handle nonlinearity and non-Gaussianity.  While AI (maching learning, neural nets) has gained some footing in model post-processing, I think this is going to be an area of active development within the models themselves.  More/better computing will continue to be a driving force, as we can get higher and higher resolution, more complex physics, coupled components, etc.

The problem is, there are fundamental predictability limits that we are going to run up against.

[showmethesnow]

24 minutes ago, Jandurin said:

1 hour ago, showmethesnow said:

Yeah, it has been a very good run. But the market is substantially over valued at this point and when the correction comes I think we are going to see a steep adjustment. In many ways this reminds me of the housing bubble just a few years ago.

First bit coin

Not sure how I feel about the bit coin craze. All I know is it doesn't pass the smell test with me. But I could very well be wrong.

[psuhoffman]

Regarding the upgrade date I do remember the new euro head faked us with that late march threat in 2016 being way over amped. I think at that time we were seeing both Euro runs leading in and the upgrade really went overboard.  But other then that we didn't get to test it much and one example and then a one month sample isn't statistically significant enough to make conclusions just bringing it up. 

[Lowershoresadness]

7 minutes ago, dtk said:

We continue to gain about about one day of lead time per decade of research and development.  We are getting pretty close to having some breakthroughs in data assimilation to better handle nonlinearity and non-Gaussianity.  While AI (maching learning, neural nets) has gained some footing in model post-processing, I think this is going to be an area of active development within the models themselves.  More/better computing will continue to be a driving force, as we can get higher and higher resolution, more complex physics, coupled components, etc.

The problem is, there are fundamental predictability limits that we are going to run up against.

Thanks dtk.  Advancing in other technologies seems to be moving at light speed. We have gotten so much better at forecasting. I just wonder what the ceiling is if there is a ceiling in forecasting

[dtk]

5 minutes ago, psuhoffman said:

I don't mind you're style but I can be that way too sometimes. People are too sensitive sometimes. And you accept discourse so it's fine imo. 

You are right to call my claim it's useless silly because without qualifying that it is. But I do think right now, for whatever reason, the euro has been struggling and throwing out way over amped solutions with too much frequency that for the very very specific purpose of determining the strength of a specific threat in the medium range it's lost its value to me. 

This is a very narrow thing but it's so all over the place lately with every threat that I would have been better off ignoring it. It would have added less noise and more confidence in what actually happened. Several times the euro has simply head faked us with crazy tangent runs. 

Im sure that will change and this is probably a temporary pattern specific issue. So I'm not trashing it permanently. And it continues to have usefulness in a broader sense in general pattern identification. But for my purposes it's hurt more then helped for the last month or so. 

I realize this is not an objective way to score a model. That the euro might have nailed 80% of the pattern and weather and scored better than the gfs but we also know what most of us care about is snow and for the 4 or 5 snow threat it's been a train wreck.  It's also not objective because from run to run overall maybe it scored ok but to me stability matters and when it has 2 or 3 crazy stupid runs that head fake us during the 5 day run up to a storm that bothers me more than if it's just a bit off consistently. I can manage that by compensating for how I imagine it will correct better then when the thing puts a 980 low over ACY when in reality it's gonna be a weak 1005 wave off the outer banks. 

So a lot of this is personal preference. And it's not objective. And it's uber specific. But that's what I meant by it's useless. I shouldn't be as flippant though. 

Fair enough and I appreciate the response.  One of the things I have observed over the years is just how much expectations have grown, which is a testament to our successes.  Wes is right in pointing out just how good the models really are and how much we continue to improve.  

I am really fascinated by the conversation about the Euro being bouncy and inconsistent, as I/we dealt with hearing that about the GFS for so long.  I am a little worried that our new model (the FV3-based core) will have more of this characteristic given that it is a gridpoint model and we'll be using the non-hydrostatic version. 

 

[psuhoffman]

1 hour ago, Mdecoy said:

I think a trace to an inch is the best bet at this point.

Its driving me crazy the people who try to claim "this is normal" in terms of snowfall. Its not normal 2 to 3 years with nothing but a single storm to our name. Sure the pattern might argue that we really shouldn't be seeing much, but people try to make out like 6 inch snowstorms are a fluke in DC. They aren't.

It's normal in a 2 year Nina. We score big in some ninos and try to get lucky the rest of the time. Once in a long while we luck into a big snow year like 1996 and 2014 that doesn't fit that typical pattern. But that rule works 90% of the time. 

[psuhoffman]

13 minutes ago, dtk said:

Fair enough and I appreciate the response.  One of the things I have observed over the years is just how much expectations have grown, which is a testament to our successes.  Wes is right in pointing out just how good the models really are and how much we continue to improve.  

I am really fascinated by the conversation about the Euro being bouncy and inconsistent, as I/we dealt with hearing that about the GFS for so long.  I am a little worried that our new model (the FV3-based core) will have more of this characteristic given that it is a gridpoint model and we'll be using the non-hydrostatic version. 

 

Totally agree on the expectations.  I made a post a week or two ago going off on tenman when he launched into his typical paranoid "the NWP is designed to be wrong" crap.  The fact we can do as well as we do is a marvel of modern science math and technology.  People don't seem to respect the level of difficulty in modeling the atmosphere given all the variables (some we can't really accurately judge so it's a prediction on a prediction) and with incomplete data.  Then the fact that the partial differential equations aren't 100% solvable so there is chaos inherent in the process.  And still we can get a pretty good accurate idea of general weather out far enough to significantly reduce life threatening events and that's priority number one not picking out snow threats in D.C. 8 days out.  I'm sorry my flippant comments came off as disrespectful in a way.  It event intended that way.

The gfs has been a rock lately. It's not always right but it's become a stubborn model and that has some valie if you know it's traits and biases and can self correct when you see a possible issue.   Jumpy is harder to compensate for imo. 

 

[WEATHER53]

Weather forecasting cannot be solved by mathematical input alone  so enhancing the grids to microscopic level may be making things worse.  Binocular view may In fact be the better route 

 

 

 

 

 

[dtk]

23 minutes ago, psuhoffman said:

Totally agree on the expectations.  I made a post a week or two ago going off on tenman when he launched into his typical paranoid "the NWP is designed to be wrong" crap.  The fact we can do as well as we do is a marvel of modern science math and technology.  People don't seem to respect the level of difficulty in modeling the atmosphere given all the variables (some we can't really accurately judge so it's a prediction on a prediction) and with incomplete data.  Then the fact that the partial differential equations aren't 100% solvable so there is chaos inherent in the process.  And still we can get a pretty good accurate idea of general weather out far enough to significantly reduce life threatening events and that's priority number one not picking out snow threats in D.C. 8 days out.  I'm sorry my flippant comments came off as disrespectful in a way.  It event intended that way.

The gfs has been a rock lately. It's not always right but it's become a stubborn model and that has some valie if you know it's traits and biases and can self correct when you see a possible issue.   Jumpy is harder to compensate for imo. 

 

No apology necessary and there was no disrespect.

[dtk]

12 minutes ago, Tenman Johnson said:

Weather forecasting cannot be solved by mathematical input alone  so enhancing the grids to microscopic level may be making things worse.  Binocular view may In fact be the better route 

 

 

 

 

 

OK, I'll recommend that NCEP just has someone run the old single-layer barotropic model on their phone and we can  get rid of our supercomputers.

[nw baltimore wx]

I missed a good afternoon here. Nice seeing dtk drop by to set us weenies straight.

[RDM]

1 hour ago, dtk said:

Fair enough and I appreciate the response.  One of the things I have observed over the years is just how much expectations have grown, which is a testament to our successes.  Wes is right in pointing out just how good the models really are and how much we continue to improve.  

I am really fascinated by the conversation about the Euro being bouncy and inconsistent, as I/we dealt with hearing that about the GFS for so long.  I am a little worried that our new model (the FV3-based core) will have more of this characteristic given that it is a gridpoint model and we'll be using the non-hydrostatic version. 

 

Fascinating exchange gentlemen - (dtk, PSU and Wes)

In my 50+ years of following the weather, I've seen a lot of changes in the systems and their forecast abilities.  No doubt there's an enormous amount of work that goes on behind the scenes that would likely boggle our minds.  That said, and I mean this constructively, it's still perplexing the amount of variance and swing between the various models, even when within a few days (or whatever the sweet spot is for each model). 

Realize there's a lot of subjective theory that goes into trying to make the models as objective as possible.  The most brilliant mathematicians often have a different way to solve a problem, just as engineers have different ways to design a car (I'm a mere mortal cursed engineer who at one time longed to become an immortal Pro Met, whom I idolized as a youngster).  What I'm at a loss to conceptualize is how wide the variance can be between the major models on a particular event and also how quickly something can materialize on one or two models only to be lost with the next run of the same model.

Unfortunately, my yeoman's understanding of weather modeling is wholly insufficient to understand the extremely complex terminology that is used to describe the theory.  If there is some sort of tutorial that outlines the various models, their strengths, weaknesses, the logic behind their computational design, and perhaps the limitations of the same, a point in the right direction would be appreciated. 

 

[snowfan]

Drought talk in January is exciting.

[nj2va]

My bar this year is set at my office closing for snow. 

[psuhoffman]

16 minutes ago, snowfan said:

Drought talk in January is exciting.

A drought can happen any time of year, didn't you hear?

[BTRWx's Thanks Giving]

Drought talk is better than hearing family members and relatives talking to me on the phone like they are worried I'm dying.  I'm still so confused about how anyone would consider that I'm depressed.  Sorry to beat the dead horse, but it's bothersome.

[87storms]

drought talk is fine, but i don't like when it's used in place of meteorology.  i see that on here sometimes.  even i do that.  let's be honest, it's mostly a way of saying i don't know how to forecast the weather.  we're not dealing with a gulf low, so of course it's going to be a relatively moisture starved system.  hopefully we're close enough to the lift where we can squeeze out whatever moisture is in the upper layers. 

[North Balti Zen]

There’s a drought thread now, take it there, all who wish to track dryness.

[BTRWx's Thanks Giving]

1 minute ago, 87storms said:

drought talk is fine, but i don't like when it's used in place of meteorology.  i see that on here sometimes.  even i do that.  let's be honest, it's mostly a way of saying i don't know how to forecast the weather.  we're not dealing with a gulf low, so of course it's going to be a relatively moisture starved system.  hopefully we're close enough to the lift where we can squeeze out whatever moisture is in the upper layers. 

It's part of the meteorology, pattern persistence. 

[87storms]

1 minute ago, BTRWx's Thanks Giving said:

It's part of the meteorology, pattern persistence. 

it is, but you gotta take the storms on a case by case basis.  is the reason we're not getting a major storm because of a drought?  no, it's because it's a clipper.

[psuhoffman]

Just now, North Balti Zen said:

There’s a drought thread now, take it there, all who wish to track dryness.

is that where we talk about best moisturizers