Best App for Weather Forecasts Anywhere in the US

Usually at Starpath we concentrate on marine weather, being the weather on ocean, coastal, and inland waters, but this is a note about weather on land, anywhere in the US. 

The National Weather Service (NWS) is the primary source, and there are indeed free mobile apps to access NWS products. But there are ways for us to get more specific and more timely information than available in standard NWS products, and some of this information we cannot get from the NWS, although the NWS themselves use this same data when making their forecasts. 

I refer to the numerical forecasts of the National Blend of Models program, and especially their continental US version, called NBM CONUS. This model gives  a forecast every 1.35 nmi (2.5 km) across the US and large parts of Canada and Mexico. 

Figure 1. Coverage of the NBM CONUS forecasts.

The forecasts are for every hour, out to 18 hr, then every three hours for 2 days. It extends farther out in time, but that is not pertinent, because the main value of these forecasts are their immediacy. The model forecasts are updated every hour, so the primary value here is for shorter term forecasts of say a day or so, then we just update this same forecast as often as we like.

The reason I title this "best app" and not just "best model" is the app I refer to (LuckGrib for iOS) is the only app I know that  can access and display what I believe is the best data set for land based weather.  It is also well suited to be set up for your area of interest, after which you can just turn it on and refresh for latest forecast, with all parameters laid out in a nice meteogram format.

Figure 2. Two iPhone views of LuckGrib, showing NBM 18-hr forecast for central Rhode Island. Left is showing wind only for the time at the blue line. Other parameters can be selected with the second icon down on the top left. Numerical values shown are for the location of the cross hair. Right is a plot of wind, rain, and temperature over the time period of the forecast.  This type of plot is called a "meteogram." Turn on the meteogram by clicking the wind data in the map view. There is a thin blue line around it meaning it is a link.

The meteogram display is a very convenient way to look at the forecast. This can be for just 18 hours, or ask for 3 or more days. The screen can also be captured in your phone or tablet to refer back to once you learn what really occurred with the wind, rain, temperature, and so on.

The LuckGrib app (LG) includes many sources of data, meaning many different numerical models. It is my belief that this source (NBM CONUS) is the best for inland weather, but I will leave that argument for another article; everyone can establish their own conclusions with experience. Screen capture the forecast then compare with what subsequently happens. 

Besides the fact that the NBM encompasses all data and numerical models that are available (it is a blend, not a single model), and that it is updated every hour,  a key attraction is it also presents the statistical dependability of the forecasts in the form of a standard deviation (SD) parameter.  This was not shown in Figure 2 (to simplify that first look at a meteogram), but we can select it in the list of parameters when setting up the forecast. For daily use of this app and forecast it would seem logical to include this important information.

Figure 3. Landscape view of the temperature forecast (solid line) along with the SD (dashed line)  at two different times.

From this we can say that the forecasted temperature at 2 hr into the forecast is 37º ± 1.1º and at 17 hr into the forecast the temp is 48º ± 1.7º.  In other words, we can think of the SD as an error bar or uncertainty in that simple manner, but the information in the SD is actually much more specific. The diagram below shows how the SD is used to characterize a statistical distribution of results.

Figure 4. Standard deviation illustrated. From our textbook Modern Marine Weather.

In the case of model forecasts, we are not dealing with "measurements" as referred to in the caption above, but rather different computations of the forecast. The NBM includes many models, each with its own forecast, along with several groups of forecasts called ensembles. A wind, rain, or temperature forecast at any specific time and place will be the average of many, maybe a hundred, different forecasts. What they report as the forecast is actually the mean value of all the forecasts in the blend. 

The other values of the forecasts will typically be spread about this mean value in a normal distribution that looks like one of those in Figure 3. When the values from all the different models are all near the mean value, then the SD will be small, and that will be a good forecast, but when the the variation of the results is spread out,  meaning different models came up with different values, then the SD will be large, and the forecast will be more uncertain. 

We see in Figure 3 that the temperature forecast was very good, with a small SD, even in those cases where it was not as good as earlier. This is typical of normal atmospheric conditions. Air temperature can often be well forecasted, within a degree or two.

Wind, on the other hand, is much more difficult to forecast in many areas around the country and its coastlines. The wind forecast of Figure 2 was updated about 5 hour later to get the data below.

Figure 5.  NBM wind forecast also showing the standard deviation (SD).

The peak wind remains at about the same time, but is now with this fresher forecast is slightly higher now at 15 kts compared to 14 before. The direction has not changed.

What we see new in this display (which we could also have seen in the earlier one had we turned it on) is the uncertainty in the forecast.  The wind at the time of the forecast (forecast hour 0H) and again about a day out is very good at about ± 1.5 kts or so, but at the peak the wind is more uncertain at ± 2.5 kts.  

This is actually a very good and dependable forecast. A standard deviation of the wind forecast of 1 to 2 kts is essentially as good as it gets.  In other words, the state of the art in meteorology cannot forecast  wind speed any better than that. Also we cannot hope for a direction much better than ± 15 to 20º. The cases we want to look out for are winds forecasted to be 15 kts ± 12 kts! 

But you will not know that without something like the NBM. We might see many different model forecasts tell us the wind is expected to be 15 kts, and the NWS report on the radio might also be 15 kts, but when you look at the SD from the NBM you see this is ±12 kts. In other words, they really do not know the wind forecast at all.  This is why the NBM is so valuable. This also brings to mind what we stress often in our weather course: there will always be a forecast... and unless you look in the right place, they will not be marked good or bad.

To follow up a bit on the standard deviation concept, we can look at more specific implications with another graphic from our textbook.

Figure 6. Statistics  based on the standard deviation. 

With a forecast of  15 ± 2, we know that 68% of all the computations put the wind within that range, and that there is a 16% chance that the wind is higher than 17 or lower than 13. Often for land-based considerations, it is more valuable to know that there is only a 2% chance that the wind would be stronger than 19 kts (2 SD).

That was a quick overview of the data for the most basic land forecast. Needless to say, this app will do very much more, including, for example, displaying forecasted HRRR model's simulated weather radar display, which is a good way to see how rain is likely to move across your region.  We can add something on that later.

LuckGrib is in the iOS App Store. There is a 14 day free trial to see it if meets your needs. After that the cost is $25. (At the end of the 14 days it just stops working and you can delete or buy as you choose, or do nothing.  There is no automatic charge or pending charge at the end of the trial period.)

Here is a short video demo of the steps discussed above.

There will also be a follow up video to show how this works on subsequent days with just 3 clicks to get a new updated meteogram, and how to set up multiple locations for weather watching.