Does a dry fall mean a dry winter season? Less than you might think.

Last week, I wrote about the dry, cool beginning to winter that we're seeing this year in Paso Robles. In my research for the piece, I made a surprising and reassuring discovery. It turned out that having a totally dry October, as we did this year, didn't have any predictive effect on our future rainfall for those winters. There were five such Octobers in the 23 years since we installed our weather station in 1996, and for the rest of the rain season (November-May) we averaged the same 22.2" of rain that we did in the 18 not-totally-dry Octobers. Yes, we missed out on the rain we didn't get that month (an average of 1.5") but it didn't appear that the conditions that produced these dry months lingered in any meaningful way later into the winter.

That got me wondering: was that true for future months? Did low rainfall in October and November mean we were likely to see a drier December-May? Did low rainfall October-December mean drier January-May? Or were the weather patterns truly independent, as my first-pass analysis last week suggested? It turned out that a dry month or months does have some predictive effect, but it's less than you might think. I'll present my findings below, but first a note on my methodology.

I decided first that using "totally dry" as my measuring point wouldn't be reasonable. We do of course have winter months without any rainfall, but after October they're rare. So, I decided that for a period to qualify as "dry", we'd need to have seen less than half of our annual average winter rainfall to date. So, for the period through October, "dry" meant less than 0.85" of total rainfall since July. For the period through November, "dry" meant less than 1.8" of rainfall since July. And for the period through December, "dry" meant less than 4.1" of rainfall since July. This does mean that the results are correlated, since this is a cumulative total, but it seemed better than counting a winter like 2009-10 as "dry" through November because we received less than 1/10" of rain in November, while ignoring that we received a nearly 10" storm in October.

So, what predictive effect does a dry early season have? About 15%, in my calculations. Here's a quick recap of the averages. At the bottom I've added in some graphs that highlight how the dry early seasons have played out.

Period	Avg Rain, Rest of Winter	Avg Rain (Dry Years)	Avg Rain (Wet Years)	# of Dry Years
Through October	22.49	20.84	24.28	12
Through November	20.49	16.92	22.05	7
Through December	16.04	14.31	16.52	5

As you might expect, the data is noisiest when you're looking at early-season results, both because there are more dry years (12) and you'd expect to have received a lower percentage of your total precipitation. In the below graph, I've marked years that met my definition of dry through October with orange columns. Wetter years are blue:

While the averages are still somewhat predictive, there are some very wet winters that followed dry early seasons, including last year. Looking at years that are dry through the end of November shows a more obvious correlation:

In the seven years where we had less than 50% of our average rainfall at the end of November, we only saw one year where we recovered to even hit our long-term average (2005-06). Some of that is the lower average future rainfall (23% less) but a lot of it is that we'd already gotten through enough of the rainy season that the difference between what we've banked in a dry year (1.2") and our average from our wet years (4.7") starts to become more significant. Looking at the data through the end of December doesn't change the picture that much:

It's probably unsurprising that when it's been dry through December, we're likely to be looking at a dry winter. But even more than in the previous chart, the biggest difference isn't in the future rainfall we'd expect (we received an average of 13% less rain January-June in those years that were drier through December) but in how much rain we've banked, or not. We'd normally expect to have received 8.17" of rain through the end of December, a little more than a third of our annual total. In the five notably dry years, we'd only accumulated an average of 2.7" to date, while in the other eighteen years we'd averaged nearly four times that already: 9.6".

What does this mean for us this winter? In practical terms, not much. Like always, we're at the mercy of the weather patterns, and what we've seen so far this fall has been consistently dry, with a persistent area of high pressure diverting storms well north of us. But for the first time, forecasts are starting to sound more hopeful, and it looks like there's a chance that this pattern will break down by the end of the month: 

Multi-model agreement that extremely dry pattern will continue across CA for next 7-10+ days (outside of poss. sprinkles in north). Ensembles suggesting about ~ 1 in 3 chance that Westerlies could break through ridge last week in Nov, bringing more widespread rain.#CAwx #CAfire pic.twitter.com/8dVNmVmYvN

— Daniel Swain (@Weather_West) November 11, 2019

For all that, I feel like the results of my digging into the weather details have done some good for my state of mind. Each week without rain at this time of year feels long. And as nice as it is to be able to go out and enjoy the beautiful sunny afternoons, I enjoy them less because of this nagging feeling that it's wrong, and we really need the rain. Knowing that the predictive effects of past early season dry spells have been modest, and that we have 90% of our rainy season in front of us (and fully two-thirds after January 1st) is a good reminder to be patient. Storms will be coming. Fingers crossed that they'll come soon.