Hey Everyone & Welcome to Miami!
*cues Will Smith* …Party in the city where the heat is on!… (Sorry, couldn’t resist 😊)
For those of you who don’t know me, I’m Chelsea, a meteorologist, sailor, and UMiami Alum/South Florida local. Find out what I’m up to with my company SeaTactics at the bottom of this blog and download our FREE Weather Resource Guide. Wondering what the best free sites are to use for this forecasting challenge? I promise this guide will help!
FORECAST LOCATION- KMIA
Miami International Airport is located approx. 5 nautical miles inland from the sea as the crow flies. (Or it can take an hour by car 😉) The weather station at the airport is on the southwest side – See images below.
One thing we can certainly expect in Miami is the sea breeze. The inland extent of the sea breeze boundary is often located quite close to MIA- which, as we’ll see later, can often queue up thunderstorms. However, keep in mind the sea breeze at the airport is certainly not as strong as it is on Biscayne Bay – 50% is probably a good estimate.
The end of Spring is a transition time for Florida from the drier winter season into the rainy season, marked by our lovely afternoon deluges you can set your watch by. (one of my favorite parts about Florida) Depending on the year, this transition can happen from late April/early May to mid-June.
This year, we are having a somewhat later transition. We actually had a couple of cold fronts push through South Florida in early May (!) and we are just now starting to move into our typical summertime pattern. The exception being: the unusually wet weather last week (May 23-27) from the tropical disturbance that eventually became Tropical Storm Bertha!
Just for fun, here’s the rainfall totals from South Florida over a 4 day period ending May 27 (courtesy of NWS Miami)…
…and this from the MIA station showing a total of 14.93 inches in the last 168 hours!! (as of May 28)
** You guys, be glad we were forecasting for PVD last week, not MIA! **
Okay, now that I’ve weather-geeked out for a minute… back to business where I’ll talk a little more about the typical summertime weather.
In the tables below you can see the normal temperatures don’t vary much. Temps may be one of the easier things to predict, while precip amounts will be more challenging. (more on that below..)
Daily Precip (in)
Highest Max Temp
Lowest Min Temp
Highest daily Precip (in)
Forecasting Tip: looking at the model guidance (such as the MOS tables) and the NWS Miami office will help get you in the right ballpark. The main reason for temps to drop significantly (sometimes 10 degrees) this time of year is due to a storm bringing cooling air from its downdraft (aka rain).
We know it’s becoming summer in South Florida when the weather is dominated by the Atlantic subtropical high, which is a semi-permanent feature (similar to the Azores high) that usually resides around 30 degrees north. It’s ridge axis is what provides the typical easterly winds along the coast.
*Side note: Interestingly, we haven’t had an uber stable subtropical high set up off the Southeast coast yet this year, hence the lack of a “normal” summertime pattern.*
When the subtropical high is farther away from the coast and more towards the middle of the Atlantic, or if the ridge axis is north of Miami, our gradient winds can be weaker which opens the possibility for onshore sea breezes.
How do we know if there will be a sea breeze? If the gradient wind in the morning is either very light/nil, or moderate from the W, AND the land is at least 5 degrees F warmer than the water, then we can get a sea breeze to develop in the afternoon. (see image below)
Getting the timing down will be the tricky part for a 12pm wind forecast. Often that is close to the time of the sea breeze fill on the coast, plus you want to allow some time for it to build inland towards the airport. It will all depend on the prevailing wind and how hot the land gets.
The sea breeze direction in Miami starts as E-SE and builds while clocking (clockwise) to become SSE-S late in the afternoon. Even when we have a gradient breeze and not a sea breeze, it is most often from the E to SE, so it’s a pretty safe wind direction to bet on. However, if a thunderstorm approaches from the Everglades and you get a gust of 30mph from the W, it can easily be a bust.
What about wind speeds? Typically this time of year, wind speeds are not very strong and usually run in the 8-15 kt range (especially inland at the airport- winds are stronger in Biscayne Bay). This is partly because of the dominance of the subtropical high extending over the area. It’s also partly because the water is quite warm and so is the land, so the sea breeze can be on the weak side.
The exceptions to the light winds are: wind gusts from thunderstorms that can reach up to 40-50kt (see my blog on thunderstorms below), and tropical systems.
The wind rose shows mainly SE breeze under 17kts because of the light sea breezes, and lighter winds from the subtropical high typically over the area.
Forecasting Tips: Keep in mind thunderstorm gusts for the daily high wind speed forecast – if a strong storm rolls by, you can easily see 35-45 knots!
Model preferences for wind would be the NAM, HRRR, and SailFlow’s 2km high resolution model.
by nature they are usually pop-up, isolated cells that are difficult to predict the exact location (although meteorologists can narrow down whether they will be focused inland, on the coast, or offshore, for example.)
A very common place for the thunderstorms to pop up is along the sea breeze boundary. This is where the sea breeze pushing inland meets a converging prevailing breeze, causing air to rise and storms to form. An interesting application of this is when the west coast FL sea breeze meets the east coat FL sea breeze, shown below. This leads to large convective storms in the middle of the state, that often hedge more E or W depending on the relative strength of the two sea breezes.
Hurricane season officially starts June, 1 but we are already teeing up the “C” storm after Arthur and Bertha both formed in the last weeks in May.
NHC and other syndicates are forecasting a more active than normal hurricane season, with potentially more US landfalls. Buzz is also generating in the meteorological world of an approaching La Nina, which usually means active an active Tropics.
In June and July (before the peak of season Aug-Sept), we see more storms that are “homegrown”, or develop close to the US like in the Gulf of Mexico or just off of Florida, like Bertha last week. As the season goes on, the generation area shifts more towards the Caribbean and Atlantic Ocean west of Africa.
Forecasting Tip: watch NHC and TropicalTidbits.com. Keep an eye on the tail ends of cold fronts or stalled/stationary fronts, where low pressures like to form.
Forecasting temps will be on the easier side with it being so warm and summer-like already, but precip will be a challenge along with max wind speeds.
The long-range models for next week are hinting at interesting weather with a possible cold front moving down the southeast coast ahead of a new high pressure cell heading towards Bermuda. If that holds, we’ll see a mix of light wind days along with fresher wind days.
Tropically speaking, we are also watching enhanced moisture and energy that could potentially be brewing in the western Caribbean Sea or Yucatan area. Should be interesting and plenty of fun!!
Best of Luck, Chelsea
Chelsea Carlson is a meteorologist, sailor, and the founder of SeaTactics, a company with the goal of providing useful and clear marine weather knowledge for safety and strategy. She has worked with mariners ranging from retired cruisers leisurely sailing around the world to high-performance racing teams. Chelsea is currently the US Olympic Sailing Team meteorologist for Tokyo 2020ne. In addition to her years forecasting marine weather, Chelsea is also a lifelong sailor who has sailed across the Pacific, navigated in various distance and day races, and is passionate about protecting our environment and oceans (which is why 1% of annual gross sales go to nonprofits specifically working to ocean conservation).
Ways to join in on the weather fun: – Watch our free webinar: “Weather Strategy 101” on www.sea-tactics.com
– Follow us on Instagram @sea_tactics and check out #WeatherWednesday where we post weather & sailing tips!
– Join one of our upcoming webinars or work with Chelsea as a private weather coach
UPDATED KPVD WIND ROSES
By Winn Soldani
It’s always good to have an adult in the room, and our good friend, Chris Bedford, noted something in my write up of KPVD yesterday and I thought I would share his insight with everyone.
The wind roses that we used below were from all the way back to 1942. This predates automated data collection and in fact was less precise than modern collection. In the pre-ASOS (the automated sensors used today) manual observers would collect the data using the 16 compass points.
Using a dataset that goes back that far biases the data towards the 16 compass points (you can see vestiges of this in the wind rose in yesterday’s blog post).
A better approach is to use wind roses only from the “modern” era, which I will post below after another point…
I got a little too precise time-wise–both in PVD and ORD–by posting the wind rose for only the few days we’re forecasting for. A broader perspective–perhaps a month, maybe even two in transitional times–makes more sense.
Chris was kind enough to send the roses using modern period data, and for the months of May and June. You can see May–with some southerly NE/NNE winds and some southerlies (likely as sea breeze “season” starts). Then, below is the June rose, where you can see that we’re definitely in sea breeze season with southerlies, off of the ocean/bay, quite common!
Welcome to week two of the CYC Wx Challenge! Matt and I are appreciative of everyone’s participation last week and hope we see even more of you this week (we’re talking to you, east coasters…we saw you sitting out KORD…don’t be afraid…).
This week, we’re forecasting for Providence, RI—the closest climatology station we have to the sailing mecca that is Newport, RI. We’re forecasting for the T.F. Green international airport (ICAO Code: KPVD).
Let me start by saying this—I’ve lived in/near two of the cities in this challenge—Chicago and Miami. On the other other hand, I’ve been to Rhode Island once (to Point Judith to see and buy my first keelboat, an X-yachts ¾ tonner). I’ve seen Narraganset Bay, but never really watched the weather in the area. So please feel free to jump in and add to/correct what I say!
Fighting local bias is key. For our Chicago competitors, this meant shifting from a lakefront focus to an inshore one, as O’Hare is off the lake.
Similarly, it’s easy to think about KPVD as “Newport”—and it’s not. In fact, as locals know, it’s not even Providence, really. The airport is in Warwick Rhode Island, south of Providence.
If you want to be a real weenie, like me, you can even see the actual AOS station where the data are collected in a google maps view of the airport. You can zoom in more than I did here, but to provide perspective it’s located to the SE of the main terminal (which you can see in top left of this photo).
Looks beautiful, right? But like ORD, looking at the extremes is eye opening!
Record Lowest high temp
Record highest low temp
As you can see, like Chicago, the word of the week is “volatility.” It can rarely get into the 90s. It can fail to get out of the 40s. Again, like ORD, identifying whether the pattern is benign or volatile is key to a good forecast.
The wind rose for the days we’re forecasting is remarkable:
Your attention is immediately drawn to the amount of time the wind is out of the S or SSE. This is clearly a sea breeze or at least has some sea breeze element to it. The temperature of the ocean and the bay, therefore, will be a major factor in temperature forecasts for PVD.
Watching and thinking about the fetch (e.g. a local thermal breeze vs a deep synoptic breeze) over water seems like an important issue to consider for PVD. In general, the wind direction seems like a major controller of temperature. Air masses originating over land will result in much different weather than bay or ocean breezes.
The Boston NWS says the following about precip in Providence:
Measurable precipitation occurs on about one day out of every three, and is fairly evenly distributed throughout the year. There is usually no definite dry season, but occasionally droughts do occur.
Thunderstorms are responsible for much of the rainfall from May through August. They usually produce heavy, and sometimes even excessive amounts of rainfall. However, since their duration is relatively short, damage is ordinarily light.
The thunderstorms of summer are frequently accompanied by extremely gusty winds, which may result in some damage to property.
Once again, we have a challenging forecast, with local effects of relatively cold ocean and bay water playing a role in the forecast. We wish you the best of luck in KPVD!
Congratulations to all of you who participated in the first week of the CYC Weather challenge. Matt and I really appreciate your enthusiasm and effort!
Results for the week are pending, but I took a moment to look at a few trends I noticed across a few variables. Hopefully you’ve been keeping track of your own forecasts and errors, but, even so, the averages I’m about to present here are perhaps instructive.
There were a total of 87 forecasts across the week that you made that were valid for the contest (i.e. not too late, etc). We examined the average forecast error across a few variables and then I totaled up a few interesting stats.
Average forecast error: -1.8 Degrees
# of forecasts that were correct: 7
Too high: 21
Too low: 59
Here we see a trend that will continue across the rest of this little analysis. On average, all forecasts were too low (temperature and wind speed).
Interestingly, though, one day alters our perspective on the analysis above. On Tuesday, May 19, out of 15 forecasts, 11 of them were too HIGH (one was right on, three were too low). Take that day out, and we now have an average error of -2.6 degrees, and only 10 out of the remaining 72 forecasts were too high, We had a definite negative bias on most days, except when we didn’t!
Average forecast error: -1.8 degrees
# of forecasts that were correct: 13
Too high: 19
Too low: 55
There was slightly less of a negative bias to the low temperatures on average. While the average error was the same as the high temperatures, there were more correct forecasts (nice work!) and fewer that were too low.
If a day stands out here, it’s day 4 (Thursday). Out of 16 forecasts, none were correct, and ALL were too low. The average error on this day was a whopping 4.9 degrees (!).
MAX WIND SPEED
I personally find that max wind speed can be the hardest variable to forecast. Most model guidance and MOS give us hourly looks at the wind speed however, the highest wind speed can—and really almost always does—fall between model guidance—either when the atmosphere “mixes out” as temps get warm enough or perhaps at the arrival of a front (synoptic or sea-breeze).
That struggle can be seen in the miss in our forecasts this past week!
Average error: -4.0 kts
# correct: 4
# too high: 11
# too low: 72
Wow! Almost every single forecast wind speed was too low!
What’s also interesting here is the magnitude of the miss. The temperature average miss of less than 2 degrees is really not bad. However, when we look at winds we can sort of calculate an average miss percentage (temperature is trickier since the scale doesn’t stop at 0).
The average max wind speed for the week was 17.8 knots—straight average across the 5 days. A -4 knot error is therefore a -22% error in wind speed forecast. Whether or not the percent error would be the same in other conditions is unclear, but this feels like a significant miss.
I hope this little analysis raises some interesting questions for you. For instance, if you followed the typical temperature misses I outlined, maybe take a look at what was different about the day(s) where you missed high vs low. What did you “get” and what elements of the pattern/local effects did you not? Were your errors larger or smaller than the average? Why?
And there are three variables I did not discuss here—rainfall, noon wind speed, and noon wind direction. Rainfall is harder to describe without getting into histograms and scatter plots (and this a volunteer operation, so…). Nevertheless, what can you glean from your forecast vs. what verified? Wind direction and speed at noon I’ll also leave to you but note that, like daily max wind speed, almost every single forecast of noon wind speed was too low! Did you have a negative “miss” tendency for the noon wind? How close were you on direction? Did you have a pattern to your misses on direction? Why?
Forecasting, and then validating and critiquing your own forecasts is the best way to become a better forecaster. We hope you’ve learned something this week and look forward to the two weeks to come. Please reach out in comments here, on Facebook, or via email with anything!
Chris Bedford is the world’s leading meteorologist for the sport of sailing. Chris currently serves as the meteorologist for American Magic, the New York Yacht Club entry in the 36th America’s Cup. With over thirty years of experience in marine and coastal meteorology, Chris has consulted with thousands of teams in boats as small as 8 feet and as large as 100+. Besides American Magic, Chris has been a part of ten prior America’s Cup Teams (winning 4) and has supported eight Volvo/Whitbread/Ocean Race Teams (also winning 4). Chris is the meteorologist for the Chicago Yacht Club Race to Mackinac and other CYC racing events.Chris just launched the Marine Weather Academy with Peter Isler, providing in-depth training in marine meteorology.
In my last post, I talked about the triad of theory, observation, and modeling as necessary elements of a good forecast. I also suggested that forecasting by looking at bunch of models and trying to select which to believe is a fool’s errand without understanding the meteorology. Today, I’ll conceptualize a forecast process which respects the triad and places model guidance it in its appropriate place in the forecast process.
A good forecast process follows a scientific approach which can be conceptualized by thinking about weather in terms of scale: Globally (or hemispherically) at first, then gradually narrowing your focus down in spatial coverage to continental, regional, then local scales. This approach is known as “The Forecast Funnel”, illustrated in the following figure – starting big picture, then scaling down to continental (synoptic) scale, then finally to what is called mesoscale (which further microscopes from regional to local).
To better understand the weather phenomena that fall within the scales, the following shows a representation for time and space with various weather/climate phenomena. Blue would fall into Planetary/Hemispheric classifications, green would be synoptic, and red would be mesoscale.
From those conceptual illustrations of the forecast process, let’s put some meat on it using a simple flow chart (from the Australian Bureau of Meteorology).
Note where in the forecast flow chart modeling (NWP a.k.a. Numerical Weather Prediction) falls! Pretty far down and even AFTER you have a basic forecast conceptualized.
In terms of time management, the amount of effort you expend at each step of the forecast process is more or less determined by the type of forecast you are producing AND the specific forecast problem of the day. For example, if your problem of the day is the absorption of a tropical cyclone along the Carolinas into a mid-latitude frontal system moving off the US East coast, you will spend more time reviewing the synoptic scale. Alternatively, if your problem is whether or not thunderstorms will become severe/damaging in a particular area, you’ll focus your time on the smaller, mesoscale.
Next time, I’ll run through a live example of this process.
As I write this, we’re done forecasting for KORD. Results from today’s forecasts will be posted sometime Saturday (hopefully) or Sunday (worst-case). You get a break until Sunday night, when forecasts are due for day one of KPVD (Providence, RI).
We hope to have posted over the weekend a few blog posts on forecasting in general as well as a climatology for KPVD, but of course you can do your own climatology research!
If you didn’t start forecasting already, now is a great time! New city, fresh start, and close to a venue we all know well, Providence, RI. Providence will provide a different set of challenges than did Chicago.
As always, please feel free to post any questions here or as comments on the blog.
Chris Bedford is the world’s leading meteorologist for the sport of sailing. Chris currently serves as the meteorologist for American Magic, the New York Yacht Club entry in the 36th America’s Cup. With over thirty years of experience in marine and coastal meteorology, Chris has consulted with thousands of teams in boats as small as 8 feet and as large as 100+. Besides American Magic, Chris has been a part of ten prior America’s Cup Teams (winning 4) and has supported eight Volvo/Whitbread/Ocean Race Teams (also winning 4). Chris is the meteorologist for the Chicago Yacht Club Race to Mackinac and other CYC racing events.
The flood of always changing data, observations, models, and circumstances make the task of weather prediction extraordinarily challenging. Personally, I feel that every forecast I make is obsolete the instant I send it out as there is always new information coming that will alter the forecast. Every meteorologist has developed their own approach and process to making a forecast. But there are common aspects that every trained forecaster follows before they apply their own spin on the problem. The common process is scientifically based. The individualized portion is the “art” of weather prediction, and that is unique to a particular forecaster.
Meteorology is a hard science. The atmosphere follows established laws of physics and chemistry. Chuck Doswell, a renowned severe weather meteorologist, refers to a good forecaster as one that can balance the “triad of components of a healthy science: 1) Theory, 2) Observation, and 3) Modeling.”
If your forecast process is comprised primarily of looking at a bunch of models and deciding which to believe, then you are a) not forecasting, and b) wasting your time. Of the myriad of models available (and there are literally over a hundred you could look at to make a single CYC Weather Challenge forecast), how can you know which is the “correct” one or, as some people refer to it, “the model of the day.” The goal is to ADD VALUE over the model, and that can only be accomplished by analyzing data (observations) and applying your understanding of meteorology (theory). Models are a GUIDE in that process (In fact, meteorologists refer to models as “Guidance”).
Weather forecasting is not black and white. “Add value” is – in and of itself – an interesting phrase. Adding value doesn’t necessarily mean getting the lowest error score. You can have the lowest error score, but make one wrong forecast at the wrong time and the impact on the user could be huge. You could predict the minimum temperature above 32F on 10 straight days and be correct 9 out of 10 times (90%!). But the farmer will only care about the one day you were wrong and it was 29F (resulting in a damaged crop since he/she did not initiate frost/freeze protocols based on the forecast).
The real emphasis is on providing actionable information for a user. Let’s consider another example. Say there is a crane operator who has an established limit of 25 knots, above which operations must cease to maintain a suitable safety envelope. Predicting whether or not the wind will exceed 25 knots is key and quite frankly an easier “GO/NO GO” forecast than predicting the maximum wind speed for the day. But for this particular user, you add value by identifying WHEN during the day that limitation will be exceeded AND communicating it effectively. Will it be over 25 knots all day or can some of the day be salvaged and, if so, when will that be so that work crews and related equipment can be scheduled to minimize downtime?
So, as you are sitting down to make your forecast today, think about your process. Have you reviewed the observations and analyzed the existing state of the atmosphere? Do the models adequately and consistently reflect the initial state of the atmosphere? Can you identify the processes at play (without models!) and understand their causes and potential outcomes based on meteorological theory? What is/are the forecast problem(s) for the day? Am I respecting and adequately reflecting uncertainty in my forecast and adding value over, say, a model consensus forecast?
When I next write, I’ll talk more about the forecast process and perhaps provide you some structure around which to make your CYC Weather Challenge forecast.
Forecasting precipitation, as you’ve seen, is really challenging when attempting to forecast for a specific point. Forecasts are generally good for an area – e.g. it’s likely to rain in Chicago on a given day with a 70% probability, but that doesn’t really help with coming up with a forecast for a specific amount of precipitation at a specific point in that area (the rain sensor at KORD). Models such as the GFS or NAM as shown on the Penn State E-Wall help you understand broad areas which are likely to see precip, and even give you a sense of the amount of QPF you should expect, but they aren’t much help in providing the detailed forecast you need to provide in the Challenge.
Today I want to highlight a resource available which can help with your QPF forecast. However, you should always remember to first understand the big picture setup, and then look to the big picture models, before turning to these specialized tools. (I’ll hit another available resource for this in the next couple of days, SREF).
The first tool are grid interpolations of the models, available from the Texas A&M site. They do exactly what it sounds like – they interpolate between grid points in the models to provide you a solution for the exact amount of QPF forecast for a given airport. To find them on the A&M site, click on Model data, and then pick GFS or NAM or RUC grid interpolation (all three models provide grid interpolations, and you should check all 3!). Put in the station identifier (KORD) and click on ‘get data’. Your results will look like this:
Interpreting this chart is a bit rough. The DAY / HOUR line tells you the calendar day and period in Zulu time – so in the example above, the first column with data is for 19/18, or May 19 for 18Z. Looking down the left, you’ll see the various parameters that the model interpolation provides. Look for ‘total precip (IN)’ and read across. You’ll have to sum the amounts shown for the periods in question (in this case, 6 hour periods). To get the total for a given calendar day using this example you’d add the 20/06, 20/12, 20/18 and 21/00 columns. (In this case, 0.00”).
Remember this is NOT “the answer”. It is one more source of info, but model grid interpolations do at least give you precise numbers to look at (but maybe not accurate ones). You need to start with the big picture and then model charts to make sure the interpolations make sense. Also, don’t rely on one model – look at all available ones and see if they’re consistent. Only then should you start thinking about a specific number forecast for QPF for your forecast location. Then look to our old friend MOS and see what guidance it provides. Your job as forecaster is to take all these (differing) sources of guidance and using them to formulate a forecast that makes sense.