Updated Sun and Mon Forecast on Sun 11:08 AM highlighted below

There are some changes and modifications in the forecast that are more clearly expressed with new post, rather than my usual edit of yesterday’s weekend forecast. See that post for more background info.

Current Sat 3:30 Water Vapor/Radar/RAP jet-level wind streamlines showing upper low—

Water Vapor image Sat 3:30PM depicting upper low. Also shows current radar and jet-level wind direction  (Click on image for a larger view.)

The weekend forecast is mostly on-track. (The hiatus in the rain in Philadelphia this afternoon was predicted by some but not all of the models.) More rain will be moving in later this evening and late night Saturday into Sunday.

The upper low will swing out of here somewhat faster on Monday than yesterday’s models were showing—

  • Monday's Upper Low Position from Friday Models
  • Monday's Upper Low Position from Saturday's Models

It will be cooler than previously forecast on Sunday. Dry on Monday.


Cloudy with periods of rain throughout the day. Difficult to identify a dry period, but mid to late morning until about noon, there may be a pause. (clearly not!) High 52.7º sd 4.7º (NBM Blue Bell)

Rain ends Sunday evening from south to north about 6 -8 PM 8-11 PM according to new HREF model.


Becoming mostly sunny, a few periods of clouds at times, especially during the mid afternoon. Cooler than previously predicted High 71.2º sd 1.6º High 72.1º sd 1.3º(NBM Blue Bell)

Updated Wind Forecast

NBM wind meteogram Blue Bell


Updated Friday 9:25 PM highlighted below

The forecast trend towards a cloudy, rainy and somewhat chilly Memorial Day Weekend continues. The only bright spot, literally and figuratively, will be Monday.

The current forecast weather setup, first took shape with Wednesday morning’s models. The ICON model was one of the first to pick up on the heavy rain.

An upper air trough develops over our area Friday through Saturday and lifts out by Monday. The upper trough will pinch into a closed low. Unfortunately, this upper air closed low will be far to our west, putting us in what’s called the “right entrance region” of a strong jet streak on Saturday.

Here’s the current (Friday morning) radar/water-vapor and RAP model jet stream flow showing the developing upper low—

By Saturday, this is the GFS forecast showing the upper low and jet streak entrance region—

GFS Model 250 mb level jet forecast 2 PM Saturday.  Note the position of the cold upper air low puts us directly in what’s called the “right entrance region of the jet streak”  (The red circles in the entrance region box are areas of strong vertical motion (vorticity) at the jet level.)  The southerly flow and orientation of the jet brings in moisture.  The vertical motion causes the rain.   (Click on image for a larger view.)

A strong moisture flow aloft will result in heavy rain. The latest GFS and NBM have joined the heavy rain forecast of the ICON model. Anywhere from 1.5 – 2 inches of rain now forecast from Friday afternoon through Saturday afternoon.

By Sunday morning, the same upper air low will induce the formation of a coastal low resulting in additional light rain, especially Sunday afternoon and evening and especially towards the coast.

GFS surface and 500 mb forecast showing upper low and coastal surface low.  More clouds and rain.(Click on image for a larger view.)

Memorial Day Weekend Forecast Details


Showers may start as early as noon- 2PM Friday, but the heavier rain moves in 3-6 PM Friday. 


Cloudy with rain early, then intermittent showers. There may be some breaks in the rain at times. It will be chilly with a northeasterly wind. High 54.6º sd 4.2º (NBM Blue Bell)


Cloudy with rain. The models are suggesting more rain over the entire area for much of the day on Sunday with the coastal low closer to the coast. An increasing chance of light Rain into the afternoon and evening, especially east into NJ. High 59.2º sd 4.6º (NBM Blue Bell)


The upper low moves out along with the coastal low Sunday, after midnight, allowing clearing.   Partly  sunny, some clouds .  Milder, but still cool for May. High 72.9º sd 1.8º  (NBM Blue Bell)

(The Canadian and ICON models  forecast slower clearing with considerable cloudiness lingering on Monday )

Wind Forecast
NBM model Wind/Gust forecast (Click on image for a larger view.)


A reader asks—

“If you have time to indulge, would you mind explaining the difference between the ICON and NBM models (the later I believe is a blend of models?), with an eye towards which rain forecast for tomorrow (modest or healthy) is the one you lean towards?”

Good question and I’ll do my best to [try] to keep it simple and straightforward.

First, I should say the latest NBM model has moved towards the ICON model with over 1.25 “ rain (ICON model- 1.6”) for Friday afternoon through Saturday morning. mid-afternoon, then cloudy. Sunday cloudy with light rain again towards evening. Clearing Monday early morning.

Complete forecast on Friday.


The NBM is the “National Blend of Models”, a blend of many different models that is run hourly by NOAA. It’s an increasingly accurate composite model that statistically evaluates numerous other model’s inputs, determines their biases and errors, and comes up with a better forecast. For temperature forecasts alone, it’s pretty complex—

Statistical step evaluation for temperature (MAE= Mean Absolute Error)  Source: NOAA


Numerous models are part of the blend.  While it is called the “National Blend of Models”, the NBM now incorporates the European Models, Canadian Models and the Australian models.  It includes all of the NOAA models- GFS, GEFS, NAM, NAM-NEST HRRR, HREF, SREF, HIRESW, RAP, GFS-MOS, NAM-MOS, EDKMOS, etc.

On an hourly basis, the NBM starts with the hourly HRRR model , adds each of these other listed models above on a regular schedule, and compares their forecasts to something called URMA, which is an error-corrected measurement of actual observed conditions six hours previously.  It basically sees which model is doing better from actual measurements six hours earlier and weights that towards a better composite forecast.  

The NBM model is available as a standard model in “grib” format and also available as a text file.

If you want to see the latest NBM forecast in text version you can click here for Philadelphia and Blue Bell.

ICON Model

The ICON model is a German global (and regional) weather numerical prediction model from Deutscher Wetterdienst.  It is run 4x daily, every 6 hours. What’s interesting about the ICON is its unique attempt to reduce computational errors by slicing/dicing up the atmosphere into icosahedrons (pyramids) instead of spherical cubes, which is what the new GFS-FV3 uses.   FV3= is Finite Volume Cube).

All numeric weather prediction models divide up the atmosphere three- dimensionally. The geometry of each model, and the number of vertical levels and the horizontal size (resolution)  determines its computational load and the amount of computational error created.   (Many of the equations for weather prediction become infinite series calculations that require cut off.  Successful geometric calculation can reduce the cut off error at the top of the troposphere. )

I’ve already gotten too technical, but the ICON is what is called a “non-hydrostatic model”, meaning it’s good at small scale vertical motions associated with thunderstorms and precipitation.

Why do I use the ICON model?  It seems to do very well with temperatures, cloud cover and high precipitation events.  I’ve been surprised how good it is. When the NBM shows high statistical spread (uncertainty), I lean towards the ICON forecast if it’s within the NBM spread. 

It’s been a challenge for me to learn how to download the ICON and convert it.  Since it’s icosahedral in geometry, it takes one of my old iMacs, used as a dedicated server, over an hour to re-interpolate the model into the standard latitude and longitude format and reduce its file size to something manageable.   And that’s the time needed to download and convert only a few parameters!

I don’t know why the ICON is not included in the NBM and I’m not aware of any research papers comparing the ICON to other models.  But I’m sure they’re out there somewhere on Google.