Which is the best weather forecast?
There is a big catch in this question that we don't always realize.
The weather forecast - be it wind forecast, rain forecast, wave forecast among others - is the result of processing numerical forecast models.
It is these forecast models that analyze various environmental data, process them in complex equations and, as a result, produce the sea and weather forecasts to which we have access.
And here's the catch.
Most forecasting websites and apps present the same information in different formats, which you more or less appreciate. Therefore, the best forecast website or app for you is: the one that gives you the information you need in the easiest and most complete way.
But not only that.
Most of the time, we want to know which is the most accurate sea and weather forecast and which is the best forecast for a specific use. In this case, we need to look a little deeper into such numerical models.
Which is the best weather forecast model?
Those who are regulars on forecasting websites and apps know that there are different sea and weather forecasts. These forecasts are generated by numerical forecasting models developed by large regional or national government agencies, such as the US NOAA.
You may have already seen the acronyms of the main global sea and weather forecast models on one of the forecast sites: ECMWF, ICON, GFS, WWIII.
It is the results of these models that most “gurus”, “winds” and “weather” websites and apps use and make accessible to the public, often with a prettier and simpler layout.
Again, choosing between free - and some paid - forecast sites depends on how you like to look at the information, because they all use more or less the same data. To go further, it is important to evaluate which forecasting model has the best performance for your region and for your needs.
There are a few ways to assess which model is more accurate:
1. Ideally, you should have access to an analysis of the accuracy of the model for your location and your needs. Like our wave forecast accuracy analysis on Dan Tysk's wind farm for Vattenfall.
This information provides the confidence needed to make decisions that impact the safety of people, the environment and assets. But no generic forecasting service is going to provide a custom accuracy analysis.
2. Another option is to check the overall forecast accuracy. This information is more common to obtain on some sites - but not all. Although it is useful, it is difficult to understand what it represents in practice.
For example, a forecast might be very good for light winds up to 15 knots, but not picking up any winds above 25 knots. So, if your problem is winds above 25 knots, the forecast will let you down even with a high overall accuracy.
3. In the absence of access or clarity of forecast performance, there are some important features of numerical models that help you to increase the confidence of your decision.
These characteristics allow us to know, for example, what the model is capable of predicting and whether the results will be close to reality. In addition to giving you more clarity on which forecast to trust when there is divergent information between them.
How to know which is the best forecast?
To assess the best sea and weather forecast for your location and activity of interest, it is important to look at some basic characteristics of the numerical models that generate these forecasts:
-Spatial resolution
-Temporal resolution
-Update frequency
It is these characteristics that will tell you which prediction is more reliable in different circumstances.
What is the spatial resolution of the forecast model?
The first step to understand a forecast is to know the spatial resolution of the model that will be used. Spatial resolution is similar to the number of pixels in a photo. The more pixels, the smaller the distance between them and the higher the image resolution.
We talk a lot about spatial resolution in this post about hyperlocal predictions.
Briefly, the spatial resolution can vary from a few meters for hyperlocal models to tens of kilometers in global models. And it indicates the ability of the model to predict spatial variations of sea and weather conditions.
The best spatial resolution for you will depend on your goal.
If the intention is to know if there is a storm approaching the coast of Brazil, a low spatial resolution model works well. But if you need to know what's going to happen on a specific stretch of the coast, like a port, the higher the resolution, the better.
What is the temporal resolution of the forecast model?
It is not only the spatial resolution that matters, the temporal resolution of the forecast increases or decreases the ability of a model to predict a given weather event.
The most common time resolutions are hourly - 6am, 7am, 8am, 9am and so on - or every 3 hours - 6am, 9am, 12pm, 6pm and so on. And the impact of that is simple.
If you only have information on sea and weather conditions every 3 hours, any phenomenon that occurs within that interval will not be identified in the forecast. So a strong wind lasting minutes or even an hour may not be captured in a forecast with temporal resolution every 3 hours.
Likewise, when reading an hourly forecast for 12:00, we are seeing the model result for this time, which may not be the same as the forecast for 12:45. However, free predictions do not have results every 15 or 30 minutes. The next result will be at 13:00 and will not catch the event.
This detail is very important to understand what is possible and what is not possible to expect from the forecast.
A weather forecast with hourly results cannot be expected to predict a transient atmospheric phenomenon lasting 10 minutes.
How often are weather forecasts updated?
The third important factor is the frequency of updates during the day.
Let's think of it this way, every forecast is seeking to diagnose the future through information from a past time. The fewer updates during the day the numerical model has, the more out of date the forecast data will be in relation to what is happening, as nature changes so fast.
The most common frequency of updating models ranges from 1 to 4 times a day - that is, every 6 hours. This means that environmental conditions around the globe have been processed by the numerical model, and a new forecast will be made available to users in that period.
The frequency of updates directly affects how many times a day it is worth or not to check the forecasts and how up-to-date this information will be. Generally speaking, the more daily updates, the better the forecast quality.
There are other factors that influence the quality of the forecast, but with these three - which are the minimum information that every site should offer - you can get a good idea of what is the best sea and weather forecast for your location.
Let's see?
What is the best forecasting model for my region?
Although the three characteristics that we explained above are important, the spatial resolution is undoubtedly the one that has the greatest weight. Because it's no use updating a model with low resolution 10 times a day - that won't make it see better the cutout of the coastline, for example.
Therefore, although the GFS is the most common free atmospheric model on websites and apps, as well as its oceanic pair, the Wave Watch III, both have a spatial resolution of 22 km, equivalent to 48.4 football fields.
As a consequence of this lower resolution, the GFS can fail a lot in mountainous regions and not perform well in the rain and cloud pattern. Likewise, despite performing well in the open sea, the Wave Watch III may fail in shallow areas such as harbors, bays and inlets.
ICON is a modern model, with a spatial resolution of 13 km and presents good results around the globe. The point of attention here is that ICON performs much better in Europe than elsewhere, as it is optimized for the European continent.
Among the main forecasting models available, ECMWF has the highest spatial resolution, 9 km. This means that every 9 km (approximately 8 football fields) the model provides the results of atmospheric (wind, rain, air temperature, humidity, etc.) and oceanic (wave height, wave period, etc.) forecasts.
This is currently the highest spatial resolution among available global models, and ECMWF is considered the most effective model. It offers good results for both weather forecasting and sea forecasting. In addition, your results have hourly resolution and update every 6 hours.
Therefore, when choosing between global models, there is a high probability that ECMWF will give better results - especially if you are not based in Europe.
But of course it has limitations. And that's why you need to be careful when making decisions based on sea and weather forecasts.
How to make better decisions based on weather forecasts?
After this brief description of the models and their main characteristics, what we can expect from a forecast is that… it will fail.
Don't get me wrong, we can't expect one forecast to get them all right. It is still technologically impossible. Nature is extremely complex, and in order to be able to provide a forecast for 10 to 15 days ahead in a timely manner, it is necessary to carry out several approximations.
Still, proper sea and weather forecasts are the best tool available to you to plan for the future, manage risk, reduce costs and seize opportunities. See the benefits of climate intelligence in the operation of Santos Brasil.
To use forecasts in the best way, you need to be aware of some points:
1. Global models are very good at predicting very extreme conditions. Even if the magnitude is wrong, the sign that something impactful will happen is evident.
2. They are also very good at predicting the most frequent site conditions, ie the common weather patterns of that region.
3. However global forecasts will not be able to identify different local wind and wave conditions. For example, between the open sea navigation channel and the interior of the port or even between one berth and another in the same terminal.
A resolution of 9 km means that the data you are viewing can be up to 9 km away from your location. This makes a lot of difference in coastal regions!
Let's go to a practical example.
In port areas, it is very common to have breakwaters, jetties or infrastructures that reduce the intensity of the waves. None of these global models are capable of identifying the difference between the wave pattern in a sheltered area and an unsheltered area. The same occurs in regions where there are bays, coves and other particularities on the coastline.
This is because the spatial resolution - even 9 km - does not allow "seeing" nuances of the coastline, depth or infrastructure of the place. In these cases, specific high-resolution models are needed for your place of interest: the hyperlocal models.
5 tips for making better decisions with sea and weather forecasts
Aligned expectations of what to expect from each type of forecast, let's look at some tips to get the best out of the information you have at hand:
1. Look for websites and applications that, in addition to being pleasant, provide you with information such as the time of the last model update, spatial resolution and show time information.
2. Check forecasts right after the most recent update and continue to monitor changes with each new update.
3. Adjust your planning to make a final decision as close to the event as possible, as the data will be more accurate.
4. When conditions differ between sites and applications. Be clear that different conditions can occur and make more than one plan.
5. When available, look at data from local sensors to see if the forecast has underestimated or overestimated reality.
Finally, if you really need safer information, a weather intelligence platform is ideal. On i4cast® you will have access to the best forecast for your location, with temporal resolution of minutes, spatial resolution of meters and results translated into impact and risk to give the best support to your decision - in addition to personalized assistance, customization and data transparency.