** A NOTE ON SPC SREF PLUMES **
DISCLAIMER: This tool is provided as a proof-of-concept demonstration project
and should not be considered operational. Data are updated in a timely
manner, but the contents are not deemed operational and processing is
not supported 24/7. The statistical methods involved in this project
remain under development, both conceptually and statistically, and as such
contents on this page are provided as a proof-of-concept experiment for
the advancement and utilization of SREF diagnostics, and not as an
operational forecast tool. In summary, information is intended for
evaluation purposes only.
The plume diagnostic tool is probably best suited for point forecasting, but
it may prove useful for parameter evaluation on the national scale. All
available parameters are first bias corrected and then statistically
dressed based on errors over the past 30 days in an attempt to produce
calibrated (i.e., reliable) SREF forecasts. Thus, evaluation of these
probabilities over a long period of time *should* yield statistically
reliable results. At this time, only the calibrated 2m temperature
forecasts have been verified over the CONUS yielding very reliable
statistical forecasts of temperature.
** WEB INSTRUCTIONS **
The web interface allows you to view the calibrated SREF output in
three ways: plume diagrams, probability diagrams (or "prob-o-grams"),
and combined prob-o-grams (i.e., "comb-o-grams," which are
the product of prob-o-grams. For example, if concerned about red
flag conditions one could consider the probability rh <= 20%
multiplied by the probability wind >= 20 mph). The comb-o-grams are
useful for determining the likelihood multiple ingredients may occur
at the same time.
Point selection is accomplished through one of two ways. Use the map
in the lower left to select any CONUS point via the mouse, or choose
a station from the drop down station list.
Generally, you want to fill in the information from the top frame down.
Start by choosing the SREF run (the latest available is the default), the
type of diagram (plume, prob-o-gram, comb-o-gram), the field to graph, and
a map background. The map background consists of the default blank map,
or the corrected median (50%) forecast of the field you selected. Then,
select your options in the middle frame and at some point, choose a
location (via the mouse or drop down list) in the lower left frame.
The available options depend on the type of plot you selected.
Plumes: Calibrated probabilistic forecasts with time. Choose the
probability(ies) via the checkbox buttons. You may choose
multiple values. A good starting starting methodology
is to select the 50% (median, or most likely) value, and the
20% and 80% values. This will plot the corrected median
forecast (i.e., 50% chance truth will lie above or below this
point, and the 20% and 80% curves which together form the 60%
confidence interval, meaning more than likely (60% chance) truth
will fall inside the two outer curves). If you add the 10% and
90% thresholds, there's an 80% chance truth will land inside
these extreme values, and a 10% chance truth will be higher or
lower than these extreme values.
Prob-o-grams: Enter the values of interest (up to 3) by entering values
inside the text boxes. You can plot one to three curves
simultaneously. For example, if you're interested
in the probability the temperature will exceed 35, 40, or
45 degrees F, then enter the following:
min_1 = 35 and max_1 = 150 (or some very high value)
min_2 = 40 and max_2 = 150
min_3 = 45 and max_3 = 150
The three curves that are graphed are the exceedance
probabilities for 35F, 40F, and 45F. Likewise, if
you're interested in the probability the wind is
between 10 and 25 mph, select 10 m wind in the top
frame and then enter the following.
min_1 = 10 and max_1 = 25
min_2 = and max_2 = (blank out these options)
min_3 = and max_3 = (blank out these options)
Combo-o-grams: Chose the field and enter the probabilistic range for
each variable similar to the above description. Then,
the product of the fields is also graphed. For example,
if you're interested in the probability of accumulating
snow then you might choose the following.
Field_1 = 2m Temperature
min_1 = -100 and max_1 = 34
Field_2 = Skin Temperature
min_1 = -100 and max_1 = 33
Field_3 = Precipitation
min_1 = .01 and max_1 = 100
Prob-o-grams for each individual field are graphed, and
the product of the three prob-o-grams (i.e., the
comb-o-gram) is also be graphed. Thus, from the product
one can deduce the relative likelihood that all three
of the ingredients will occur simultaneously. The product
of the probabilities forms the joint probability if the
assumption of independence is made. In other words, the
product of the probabilities may not be an actual probability
if some degree of dependence exists between the predictors.
Nonetheless, experience at the SPC has shown that the
combined probabilities are quite useful and can still serve
as a tool for assessing forecast confidence (or lack thereof).