The purpose of this web page is to describe several methods for forecasting snowfall amount. Upon completion of this material you will be able to:
Forecasting snowfall amount is one of the most difficult challenges that a meteorologist faces. Although four factors that affect snowfall amount and three explicit snowfall methods are discussed below, these methods are not exact and leave room for improvement. Nevertheless, snowfall amount is a very important part of your forecast and can have a significant impact on the lives of many people.
| What does snow amount depend upon? |
| Available Moisture |
| Snow Duration |
| Storm Intensity |
| Warmth of the Ground |
When we talk about available moisture, we are referring to the moisture that is feeding the storm and producing the snow, i.e., the air that is flowing into the area of best upward vertical motion associated with the storm.
Available moisture does not refer to the conditions at the earth's surface where the snow is expected. The point of snow occurrence is usually in the cold air while the moisture that is producing the snow is coming from the warm air via some form of the warm conveyor belt.
Available moisture is best measured by examining the dew point at the earth's surface and at the 850 mb level in the source area for the moisture. The following table lists typical values that are considered adequate moisture for significant snowfall. Significant snowfall is 4 inches of snowfall or more in a 12 hour period.
| Adequate Warm Sector Moisture |
| 850 mb dew point: -2 deg C or greater |
| surface dew point: 36 deg F or greater |
When examining surface and 850 mb dew points, you need a wind component that will advect the warm sector moisture into the storm and beyond the 850 mb 0 deg C isotherm (where the heaviest snowfall is most likely). Ideally, this moisture should feed into the area of a storm where the lift is strongest.
Although there is no statistical correlation between available moisture and snowfall amount, as dew point values drop below those listed in the table above, snow amounts will diminish.
Remember ...
This message is particularly important if you are forecasting for an area away from a good moisture source such as the ocean.
The second factor that affects snowfall amount is obvious but still must be stated.
Based on model dynamics and storm movement, you can usually estimate fairly accurately the duration of a snowfall event. You then combine the snowfall duration with anticipated snow intensity to estimate snowfall amount.
A study at LaGuardia Field (KLGA) in New York City statistically correlated snow intensity (light, moderate, or heavy) with the snowfall accumulation rate. The results of this study are shown in the following table.
| Average Snowfall as Function of Intensity | |
| -SN (light snow) | 0.2 inch/hour |
| SN (moderate snow) | 1.0 inch/hour |
| +SN (heavy snow) | 1.6 inch/hour |
Let's assume that system movement indicates that snowfall duration will be 12 hours long. During that period, storm dynamics indicate two hours of moderate snow is likely as the better upward vertical motion moves across your forecast area. This produces the following snowfall estimates:
| 10 hrs light snow | at 0.2 inch/hour produces | 2.0 inches |
| 2 hrs moderate snow | at 1.0 inch/hour produces | 2.0 inches |
| Total Snowfall Estimate | 4.0 inches | |
Let's apply this method to actual snowfall event that occurred at Topeka, Kansas.
| Type | Began | Ended | Duration | Calcuation |
| -SN | 0000 | 0340 | 3.67 | 3.67 x 0.2 = 0.73 inch |
| -SN | 0640 | 0905 | 2.25 | 2.25 x 0.2 = 0.45 inch |
| SN | 0905 | 0920 | 0.25 | 0.25 x 1.0 = 0.25 inch |
| -SN | 0920 | 1019 | 1.00 | 1.00 x 0.2 = 0.20 inch |
| SN | 1019 | 1420 | 4.02 | 4.02 x 1.0 = 4.02 inch |
| +SN | 1420 | 1530 | 1.17 | 1.17 x 1.6 = 1.87 inch |
| SN | 1530 | 1830 | 3.00 | 3.00 x 1.0 = 3.00 inch |
| -SN | 1830 | 2030 | 2.00 | 2.00 x 0.2 = 0.40 inch |
| -SN | 2120 | 2230 | 1.17 | 1.17 x 0.2 = 0.23 inch |
Using this method we get an estimated storm total snowfall of 11.15 inches. The actual snowfall during this period was 11.3 inches.
The example shown above and the original development of the correlation were done in the pre-ASOS era when reported snow intensity was related to visibility. ASOS uses a different method for determining snow intensity. Nevertheless, this general approach to estimating snowfall amount should still be valid.
When using this method to estimate snowfall amount be aware that thunderstorms producing snow will cause locally heavier snowfall than indicated.
These last two factors are rather qualitative statements. With regard to the first, in general, the stronger the storm system, the heavier the snow, other factors being equal, and assuming adequate moisture is available. This factor really does not help in most situations.
The warmth of the ground is a factor that comes into play early in the winter before the ground has had time to cool off or freeze. If the ground is warm enough snow may melt upon contact and reduce the accumulation. For example, if you expect 3 inches of snow but ground warmth melts one inch of it, a 2 inch snowfall forecast would be appropriate. Similarly, snow may accumulate on the grass but not on the roads. Some weather offices have access to ground or road temperature data that may help in evaluating this impact on snowfall accumulation.
The first of the three methods for forecasting snowfall amount is the Magic Chart. This technique was developed during the AFOS era (NWS during the 1980s) and requires two charts that were available on AFOS:
The area of greatest NVD overlaying the -3 to -5 deg C band is where the heaviest snowfall is likely to occur. The 12-hour snowfall amount is the NVD (in mb) divided by 10. For example, if the NVD = 60 mb, 6 inches of snow would be forecast for the 12-hour period. If the NVD = 120 mb, 12 inches of snow would be forecast.
Experience with this techniques shows that it works best for mature or developing synoptic low pressure systems, i.e., organized systems. It does not work well for overrunning situations (large areas of isentropic lift). In terms of forecast confidence, it is rated moderate.
The second method for forecasting snowfall amount is the 200 mb Snow Index. This method is rather involved in that there are 10 steps that must be followed before arriving at a forecast. Instructions refer to the current 200 mb chart unless otherwise specified.
Step 1: Locate the area of warm advection downstream from the short wave trough producing the snow.
Step 2: Find the coldest area (usually in the ridge).
Step 3: Find the warmest area (usually in the trough).
Step 4: Find the temperature difference from cold to warm.
Step 5: The 200 mb snow index or 24-hour snowfall amount is:
Step 6: The maximum snowfall during the next 24 hours is near the coldest 200 mb temperature that is downstream from the warmest 200 mb temperature.
Step 7: Heavy snowfall bands are parallel to the 200 mb contours and downstream from the maximum warm advection.
Step 8: The width of the heavy snow is about the same as the distance between adjacent contours (at 200 mb). This assumes a 120 mb contour interval.
Step 9: The east or northeast end of the heavy snow is located near the eastern end of the coldest 200 mb temperature (but never more than 14 degrees of latitude from the warmest 200 mb air).
Step 10: The west or southwest end of the heavy snow is located just downstream from the maximum warm advection.
The third method for forecasting snowfall amount is Garcia Method that was developed at the National Weather Service office in Milwaukee, WI. This methode involves evaluation of isentropic lift and available moisture in the form of mixing ratios.
Step 1: Based on the general weather situation, identify the "area of concern" for heavy snowfall.
Step 2: Run a cross section through the "area of concern". This cross section should be parallel to the low to middle level flow through the area.
Step 3: From the cross section, determine the isentropic surface that best crosses the "area of concern" in the 700 to 750 mb layer.
Step 4: Run an isentropic surface for the value determined in Step 3. On this surface analyze the following fields:
Step 5: Using the analysis from Step 4, identify areas of uplift, that is, areas where the wind is flowing from higher pressure values to lower pressure values.
Step 6: Determine an "effective mixing ratio" for the next 12 hours by averaging the mixing ratio from the "area of concern" upstream for a distance corresponding to a 12 hour advection fetch. For example, use an average wind and recall that for every 10 knots, a fetch of 2 degree of latitude will result during a 12 hour period.
Step 7: The "potential snowfall" for the "area of concern" is equal to twice the "effective mixing ratio" values determined in Step 6. For example, if the "effective mixing ratio" is 2.5 g/kg, the "potential snowfall" is 5 inches. This assumes that there is sufficient lift to produce the snow.
Cautions:
Northwest flow refers to situations where the upper tropospheric flow (500-300 mb layer) is flowing from the northwest toward the southeast. In general you do not expect significant snowfall with these systems because they are usually between the upstream long-wave ridge and the downstream long-wave trough. However, snow can occur.
One study that looked at northwest flow situation developed the following rules of thumb:
What sometimes happens with northwest flow is that the mid-troposphere (500 mb) flow is relatively weak while a vigorous jet streak is present in the 300-200 mb layer. This jet streak is sufficient to develop upward vertical motion that produces snow when the jet streak interacts with a low-mid level moist source. Typical snowfall amounts are around 3 to 4 inches.
Numerical models do not predict snowfall directly. Models do, however, forecast liquid precipitation amount, that is, the accumulated condensation produced by lifting air beyond its saturation point. This forecast is commonly known as the Quantitative Precipitation Forecast (QPF). If other factors indicate that the precipitation type will be snow, the QPF can be used to estimate snowfall amount.
The ratio of snow to liquid water equivalent is, on average, around 10. This means that 10 inches of snow typically yield 1 inch of liquid water. This ratio can be used to convert QPF to a snowfall estimate. For example, if the model QPF says that 0.25 inches of water will fall during a 6-hour period, this could be interpretted to mean 2.5 inches of snow will fall.
There is a catch, however. The snow to liquid ratio is not always 10. In dry snow situations, when available moisture is sparse, the ratio can be considerably higher with values around 20 not uncommon. On the other hand, if there is abundant moisture and the snow is wet, a ratio of 5 may be more representative. The bottom line is that you have to use your judgement as to available moisture and select a ratio that might be appropriate for the situation.
The other thing about model QPF is that verification of QPF values is typically poor. So be cautious when using QFP.
Even though we have expended a considerable amount of verbage on forecasting snowfall amount, the bottom line is that snowfall forecasting is more art than science. Nevertheless, the concepts presented here should help you make an educated estimate that blends science with qualitative judgement.
The location of the heaviest snowfall is addressed in a separate lesson.
Have you ever listened to a television weather presentation that made a snowfall accumulation forecast 3 days ahead of the expected event? This type of long range forecast raises a philosophical question: How far ahead should you place snow accumulation amounts in your forecast?
From the discussion above, it should be evident that snowfall forecasting, in general, is an inexact science. Some might even argue that calling it science is stretching the issue. So, if you have that much difficulty forecasting snowfall amounts in general, why would you do it 3 days ahead?
As a snow storm approaches an area, computer models usually give an indication of snow potential 3 to 5 days into the future. As the event comes closer in time, track and intensity tend to change and different models often give different solutions. A track change of a little as 60 miles for a low pressure center can signficantly modify the amount of snow that falls or the precipitation type. The availability of moisture is often not clear until the storm is relatively close. On the other hand, in most cases, by the time the snow is 24 hours from falling, things are usually fairly well defined.
Due to the uncertainities just mentioned, it is well within the state of the science, and strongly recommended, to limit snow accumulation amounts to the first and second periods of your forecast. Beyond that point in time, you mention snow potential but don't commit to amounts.
There will be people who disagree with the statements just made, particularly news directors who want to scoop the competition, but the current state of the science for snowfall forecasting is considerably uncertain beyond 24 hours into the future.
Instructions: Place the cursor over the answer of your choice. If you are correct, it will be highlighted in green; if you are incorrect, it will be highlighted in red.
Which of the following factors affect snowfall amount?
If you are examining an 850 mb chart, what values would you look for to determine if sufficient moisture is available for heavy snowfall?
It snows for 6 hours. During that period the snow is light except for a one hour period when moderate snow falls. Estimate how much snow fell.
One factor used by the Garcia method is:
Chaston, Peter, 1989: The Magic Chart for Forecasting Snow Amounts. National Weather Digest, 14, 10-22.
Cook, 1980: A Snow Index Using 200 mb Warm Advection. National Weather Digest, 5, 29-40.
Garcia, Crispin, Jr., 1994: Forecasting Snowfall Using Mixing Ratios on an Isentropic Surface, An Empirical Study. NOAA Technical Memorandum NWS CR-105, 31 pp