Terpeast

I’d compare DCA to BWI, SBY, and EZF though. The lows at DCA are often warmer than all three. BWI is closer to DCA at 59 for the same reason, while SBY and EZF are both in the 56-57 range. Like raindance says, the 60F line is probably down closer to richmond. It’ll take a while for it to creep up, even if locally it gets there first at the river and bay areas.

We’re most definitely not in the yellow yet. Our recent 15-year average temp is 56.6, and even the last 7 years averages out to 57.3. Still solidly green. Maybe well SE of DCA towards the bay is where you’ll see the green/yellow line creeping up, but it will be a long time before I-95 and west becomes yellow.

Backed off to 94 with mid-level clouds rolling in after hitting a high of 98.

96 imby, dews much higher today at 69

My yard could really use this

That’s what I am hoping. I do think we still have some loaded winters coming up

That line was real. Saw a few raindrops in my morning run. Not even a trace though, just made it even more humid.

Thanks. sometimes I do wonder if the “tipping point” had already happened after the last super nino, or if we’re just in a down cycle. I still think it’s the latter, but this coming winter should be a good test if the nino plays out as most think it will. But even if we’re in a down cycle now, these kinds of winters will become the norm anyway. Any double digit winter will be a win for this forum.

Agree, and it’s also not just how strong or deep the trough is, it’s also how strong the accompanying ridge is. I’ve been watching the MA summer here and we’ve had a lot of NW flow due to the persistent central US ridge (which oddly doesn’t show on the map above, but it was there). With all the fires in Canada and extreme heat in the west, a weak or even moderate trough in the east isn’t going to bring much colder temperatures.

Can confirm. I don’t think I’ve seen an entire summer where IAD constantly beat DCA for daily maxes 97 for the high imby

94 now. Outpacing the forecast so far

Yeah, I don’t think I’m making a lot of friends here haha but I rather be honest and call it for what it is. Positive thing to draw from this is that CC makes extreme events more likely. So if it’s going to snow, it may be extreme.

My guess would be that storm tracks would be shifted north. But I have no way, or resources, of proving this hypothesis.

Yeah, I’m already thinking of a vacation home up around there or even Vermont

High of 90 imby, same as IAD. With the dryness, I wouldn’t be surprised to see temps over performing this week a bit. 100 isn’t off the table.

Looks like my analysis supports this.

Okay, you’ve been warned. This is a long post, so get comfortable. Several months ago, I decided to undertake some research and statistical analyses to tease out what we can expect in future winters in the mid-Atlantic with a warmer background state. Thanks @psuhoffman for the inspiration. The conclusion is… sobering. Short answer? We will still get snow, just less than we used to… not surprisingly. The real question is “Just how MUCH less?” Here’s how I did my analysis. DATA PREP I downloaded all daily temperature, precipitation, and snowfall data at KIAD from NWS’s NOWData. I chose KIAD even though it only goes back to 1962 because DCA is a poorly sited measuring location, even with a longer record. I could have chosen KBWI for a longer record, but I’m not sure how relevant weather records before 1960 would be due to those earlier years being in a vastly different climate state. 1962 through 2023 is 60 years worth of data, which may not seem like much, but found it sufficient for this purpose. After downloading, I put it together as an entire series in one Excel tab. There were something like 22k+ rows of data. TRENDS IN TEMPERATURE AND PRECIPITATION Next, I analyzed a linear trend in temperature and precipitation to get their slopes. At KIAD, we have warmed by 3.9 degrees F between 1962 and 2023 - from 52.9 to 56.8 F. At the same time, we have had a 16% increase in precipitation from 3.23”/month to 3.74”/month. Warmer and wetter. So, using the slopes of these trends, I applied a reverse delta-T and reverse delta-P to adjust the entire time series warmer and wetter to match today’s climate. But not all days or years are adjusted the same way… the earlier in the time series, the bigger the adjustment. The most recent years prior to 2023 only had negligible adjustments just because they are closer to today in time. The “adjusted” time series would have no trend (flat trendline), and are thus used in methods 1 and 2 to derive “new” snowfall totals for every winter “event”. I’ll explain each method… METHOD 1: SNOWFALL ADJUSTED BY TEMPERATURE First, I studied the daily average and minimum temperatures for each snow event (using the original dataset). At first, I tried using daily average temperatures, but the data was too noisy. There were snow events even with daily averages between 35 and 40 degrees F. And it was hard for me to tell what daily average temperature do snow totals begin to decrease. So I switched to daily minimum temperatures. Now the picture becomes clearer: From visualizing this graph, I can make assumptions: 1) For the biggest events, snowfall increases 50% every 4 degrees until around 25 degrees (due to higher moisture content and better dendritic growth). Then snowfall levels off from 25 degrees and up, and then decreases past 30 F. 2) If MIN temps increase from 25 to 30, there is not much difference in snowfall. But we get a 30-50% decrease from 30 to 32 degrees. Minor snowfall events (5" or less) get cut in half when MIN temps go from 32 to 33 Major snowfall events (>5") become minor (<5") when MIN temps go from 32 to 33 Minor snowfall events become rain when MIN temps go from 33 to 34 (important!) So when I apply these assumptions to the “warmer” time series, I adjust past snowfall events to new snowfall totals as if they happened today. METHOD 1 RESULTS: From Jan 1962 through April 2023, a cumulative total of 1325 inches had fallen at KIAD. With adjusted snowfall using method 1, the cumulative total becomes 1142 inches. This is a 14% decrease in snowfall at KIAD. So if KIAD averaged 22”, then our new average in TODAY’s climate is now 18.9”. I can hear you saying now: "Oh, that's not so bad, we'll still average almost 19" of snow even in a warmed world!" Ah... not so fast... remember we are not accounting for FUTURE warming - more on this later. Back to the analysis. I also cataloged changes to each winter “event” as to whether their snowfall decreased, increased, or stayed the same (within 10% of the original snowfall). If my adjusted snowfall for an “event” became zero (i.e. all rain), I log it as an “Event Lost”. Here’s how all of our past snow events would have changed if they happened in TODAY’s climate: Increased Snowfall: 17.1% No Change: 35.4% Decreased Snowfall: 32.6% Event Lost: 15.0% Note: I only consider any snow day with 0.5” or more of snow an “event”. I omitted the rest. In other words, one-third of past snow events wouldn’t have happened much differently today. Another one-third would have resulted in less snow, but still a snow event nonetheless (perhaps with more mixing if the conditions are right). More notably, 17% of past events would have resulted in increased snowfall, while 15% of them would have been all rain today. Here are some examples: Of those 17%, I can highlight a few examples of very cold storms where a few degrees increase wouldn't flip it to rain, but would actually juice up the storm: Feb 1979 (PD1) would have dropped 22" if it happened today rather than 16" at IAD\ Feb 1983 would have dropped 30" today instead of just 23" (!!) Jan 1996 would also have dumped 30" instead of 24" Then as we get into more recent storms like 2010 and 2016, the effect would be minimal because the temperature difference would be minimal. Interestingly, in the first 2010 storm, day 1 yielded 4" less, but day 2 yielded 2" more, so the result was 30". The actual was 32.4", so the loss was minimal, but more on the front end of that storm when the BL temps were still warm-ish and lots of snowfall was lost to melting. The 2016 storm had almost no change (obviously). Now, on the flip side... how many major storms did we lose? 1) We lost a footer in Feb 1987... assuming my method is correct (more or less), that storm would be the perfect track rainstorm. Total shutout. Sobering fact: This is the most notable example of losing a great snowstorm to a perfect-track rainstorm as the climate warms. We already have examples of this like 1998 (which was a preview into the future during that super El Nino) and more recently in 2013, 2018, etc. with more of these “lost events” yet to come. And here’s yet another kick in the gut: 2) There were four 8" storms throughout the 1960s... and we lost them all! Four 1960’s era 8-inchers all zeroed out. (Though there were other 3-4 storms in the 60s that made up for those losses by adding more snowfall... one 10" storm would produce 15" today, for example). Pretty sobering, eh? Now, we haven’t looked at changes in precipitation and SWE due to CC, so let’s try another method… METHOD 2: SNOW WATER EQUIVALENT ADJUSTED BY TEMPERATURE AND PRECIPITATION This time I used average (adjusted) daily temperatures to apply treatment to SWE values based on adjusted precipitation/QPF to derive new daily snowfall values. I made this treatment only to a small window of temperatures, though. If the adjusted average daily temp was 25 degrees or colder, I leave it unchanged and only adjust snowfall based on precipitation. But when the temp is between 26 and 32, I apply the decreasing slope on this graph to SWE and derive snowfall from both that and precipitation. Here’s where it gets tricky, though. If an event had an average temperature of 24 degrees, and the adjusted temperature is now 28 degrees, I don’t apply the full adjustment to SWE. I only use the portion from 26 to 28 degrees, so in such case, only one-half of the full adjustment is applied. I also made other adjustments for when the average temp is pushed above freezing (e.g. a 31 F event now becomes 35 F): SWE falls by 30% when AVG temps go from 32 to 34 SWE falls to below 5 when AVG temps go from 35 to 36 SWE falls to zero when AVG temps go from 36 to 37 Note on potential mixed/icing events: A) SWE of between 3 and 7 with AVG temps of 32 or less may be mixed events (partially filtered out by only considering days with >1" snowfall). SWE <3 with AVG temps of 32 or less may be sleet or ice events (filtered out by only considering days with >1" snowfall). METHOD 2 RESULTS: Remarkably similar to method 1. Total cumulative snowfall is now 1144 inches, again a 14% decrease in snowfall. Here’s how the “events” shake out: Increased Snowfall: 9.5% No Change: 40.8% Decreased Snowfall: 30.7% Event Lost: 18.9% The main difference is that method 2 is less generous towards top-end snowstorms in terms of increased totals. Instead, most of these events don’t change much. Also worth noting that we lost almost 19% of past events with this method. ChatGPT’s Take: What AI Has to Say About This… I input all this data into GPT’s code interpreter to do some analyses and fill in any gaps I might have missed. I started with something super basic. I asked it to create histograms of average daily temps between two time periods: 1962-1990 and 1991-2023: You can notice a shift towards the right with daily average temperatures. So now we know that GPT can see the warming trend. Then I asked if our hypothesis of increased snowfall in discrete events within a warming world is valid: The response: Yes, that's a scientifically valid hypothesis. While a warmer climate can lead to overall reduced snowfall, it can also result in more intense, albeit less frequent, snowstorms. Here's why: 1. Increased Moisture Content: Warmer air can hold more moisture. For every 1°F increase in temperature, the atmosphere's capacity to hold moisture increases by approximately 4%. When conditions are cold enough for snow, this additional moisture can fall as more significant snow accumulations. 2. Oceanic Effects: Warmer temperatures can lead to less sea ice and warmer ocean waters. Open and warmer waters, particularly in the early winter, can be a source of added moisture to the atmosphere. Coastal areas, or regions downwind of large bodies of water, can experience heavy snowfalls when cold air passes over these warmer waters, picking up moisture and subsequently dropping it as heavy snow over land. 3. Storm Patterns: Climate change can influence atmospheric circulation patterns, potentially leading to changes in the frequency, intensity, and tracks of snowstorms. Some research suggests that certain regions could see more frequent and intense snowstorms, even if the total number of snowy days decreases. 4. Temperature Fluctuations: Even in a generally warmer world, there will still be periods of cold temperatures. If these cold periods coincide with moist conditions (as might be more common due to the reasons above), the result could be significant snowstorms. 5. Changing Landscapes: As the climate warms, the line between rain and snow can shift. This means that regions previously receiving rain might start receiving snow if the conditions are right, leading to unexpected snow events. However, while these factors can contribute to larger snow accumulations in specific storms, they don't necessarily contradict the potential for reduced overall snowfall. It's a matter of frequency versus intensity. There might be fewer snow events, but when they occur, they could be more intense. Cool, nice response. Very plausible. Now let’s try to validate it using the KIAD data series: To better understand the trend in the intensity of snowfall events, we'll proceed with the next step: Check for Extreme Snowfall Events: We'll identify and count the number of days with snowfall amounts exceeding a certain threshold over the years. Specifically, we'll consider days in the top 5% of all recorded snowfall amounts and see if there's an upward trend in their frequency. The plot illustrates the number of extreme snowfall days (days exceeding the 95th percentile of snowfall amounts) for each year from 1962 to 2022. From the plot, we observe some fluctuations in the frequency of extreme snowfall days over the years. There are periods with a higher number of extreme snowfall days, but there isn't a clear, consistent upward trend throughout the dataset. To further explore, let's look at the next approach: Yearly Maximum Snowfall: Analyzing the maximum daily snowfall for each year can provide insights into whether peak snowfall events are becoming more intense. The plot showcases the maximum daily snowfall for each year from 1962 to 2022. A few observations: 1) Fluctuating Intensity: The intensity of the maximum daily snowfall fluctuates from year to year without a definitive upward or downward trend. 2) No Clear Increase in Intensity: There doesn't seem to be a clear increase in the intensity of peak snowfall events over the dataset's span. While there are some years with notably high maximum snowfall amounts, they are interspersed throughout the dataset without a consistent pattern. While the dataset does indicate the occurrence of significant snowfall events, it doesn't conclusively support the idea that the intensity of individual snowstorms is increasing over time. However, it's essential to note that more comprehensive datasets, encompassing larger geographical regions or longer periods, might yield different insights. Additionally, factors like localized climate influences, geographical variations, and data collection methodologies can also play a role in these observations. So, not much of a trend in either graph. Maybe it’s due to my prompting skills, or maybe we’re looking for a trend that isn’t really there. Next, I asked it to project our future annual average snowfall assuming the world warms a further 1 degree F over the next 20 years (I’m being conservative here): It first produced a graph like this: Then it said: 10.91” annual average snowfall? Ouch… And that’s the average over the next 20 years… meaning that by 2043, single-digit averages with a ratter every other year will become the norm. And that’s assuming we warm only a conservative 1 degree F over the next 20 years, while a ton of evidence points to a faster or accelerated rate of warming. How’s that for a real kick in the balls? Now, if you want to poke holes in my study or tear it apart, allow me to do it for you: WEAKNESSES and/or BIASES Many studies have limitations or biases, and this is no exception. Here are the weaknesses: We are using a point snapshot of sensible weather in one location. We do not consider any spatial alterations in weather patterns over time. We do not consider changes in storm tracks, jet stream placement or strength, or the placement or strength of H5 anomalies We do not consider the state of ENSO, PDO, AO/NAO, etc. We do not consider mixed events (if SWE is less than 5 or 6 while the average daily temperature is colder than 29 degrees, it is likely a mixed event) This is only a statistical analysis that relies on hypothetical changes in snowfall based on a warmer climate This does not take into account non-linear changes imposed by a warming climate, which needs to be modeled and validated This study uses ChatGPT's code interpreter to draw certain conclusions, which may or may not contain bias or imperfections In my use of ChatGPT, my prompting may or may not have introduced bias The “gold standard” would be to run op models with ensembles starting in 1962 (or whatever past date you prefer) using an adjusted initialization state with parameters that factor in today’s warmer climate, without allowing the models to revert to the previous/colder base state. Then repeat for every 6-hour intervals all the way through today’s date. That would be the proper way to do such research, but I (and most people on here) don’t have the resources to do that. And finally, this assumes warming will continue into the future. There’s an outside chance that I could be totally wrong, but it’s hard to argue against the trend. CONCLUSION Although this probably won’t pass the muster of an academic peer review, I’m glad I did this exercise to satisfy my curiosity and reset my own expectations of our winters to come. On a positive note, due to increased QPF and moisture with warmer SSTs, we may experience at least one, maybe two more record-breaking snowstorms in the next 10-15 years… One of them could be an extreme and biblical 1 in a 10,000-year event. And then, sadly, as the world warms even further and maybe faster than expected… it’s lights out. Meanwhile, let’s enjoy what we get… and we will get quite a few memorable storms - hopefully this coming winter.

Looks even worse than here in Ashburn

Low of 53

And when does it peak?

Voted strong. Thinking 1.8 or 1.9

I respect a forecaster who admits their own biases much more than someone who armchair quarterbacks other people’s forecasts.

You’re not wrong. I’m still investigating the pdo data to see if the correlation is real, or if pna is still sustainable in a neg pdo regime. It was in the 1960s. It may have been in 09-10, too. I see weak negative values for that winter, maybe -0.5 or so. But if the marine heat wave off Japan doesn’t cool down a tad, I’d wager that’s going to have an impact.

The only thing that keeps me from getting over the moon excited over CanSIPs is the PDO. looking at the sst map, it’s looking even more negative on this run than last month’s run. out of all sst indices, the PDO has the strongest correlation to temps, precip, and snowfall in the MA. I don’t know if the models are picking this up, or if other tele’s are muting (or overpowering) the PDO impact, but this is something to watch and take into account.

This is… loaded. Hope this holds for the next… 6 months