Tải bản đầy đủ

Bài giảng khí hậu học chương 10

G304 – Physical Meteorology and Climatology

Chapter 10
Weather forecasting
and analysis

By Vu Thanh Hang, Department of Meteorology, HUS

10.1 Forecasting methods
• There is no single “correct” way to forecast the weather.

• Forecasts based on long-term information in the absence of
any data about current weather are called climatological
• The reliability of a climatological forecast depends on year-toyear variability in weather conditions for the forecast day.
• A persistence forecast relies completely on current conditions
with no reference to climatology.
• This simple procedure might work for a little while but will
eventually fail to catch changes in weather.

10.1 Forecasting methods (cont.)
• The analog approach tries to recognize similarities between

current conditions and similar well-studied patterns from
before, assuming that what happened sometime in the past
provides a clue about the future.
• Numerical weather forecasting is based on computer
programs that attempt to mimic the actual behavior of the
• The numerical models typically used in weather forecasting
are very large and can only be run on the most powerful
computers, so-called supercomputers.

10.2 Types of forecasts
• The product (or result) of a forecast method can take a

variety of forms Æ type of forecast.
• Quantitative forecasts specify the “amount” of the forecast
variable. Ex.: “an cm of rain is expected”.
• Qualitative forecasts provide only a categorical value for the
predicted variable. Ex.: “rain/norain”.
• In a probability forecast, the chance of some event is stated.
Ex.: the probability-of-precipitation forecast (PoP forecast).

10.3 Assessing forecasts
• Forecast quality refers to the agreement between forecasts

and observations.
• Forecast value refers to the utility of a forecast and
necessarily depends on the application of a forecast to a
particular problem or decision.
• The quality of a forecast concerns forecast accuracy.
• That is, on average, how close is the forecast value to the
true value?

10.3 Assessing forecasts (cont.)

• Forecast bias concerns systematic over- or under-prediction.

• A biased forecast method is one whose average forecast is
above or below the true average.
• The mean absolute error (MAE), ignores the sign (positive or
negative) of the errors Æ over- and under-predictions are
treated the same.
• Forecast skill is defined as the improvement a method
provides over what can be obtained using climatology,
persistence, or some other “no-skill” standard.

10.4 Data acquisition and dissemination
• The starting point for almost all weather forecasting is

information about the current state of the atmosphere.
• The first process in operational weather forecasting is
acquiring the necessary data.
• The WMO collects data from land observation stations, ship
stations, moored and drifting buoys with automatic weather
sensors, weather satellites.
• Upper-air data from weather balloon sites, instruments
aboard wide-bodied commercial aircraft.
• The data from all countries are sent to three World
Meteorological Centers at Washington, Moscow, Melbourne,
which in turn disseminate the data to all member countries.

10.4 Data acquisition and dissemination (cont.)
• The member nations of the WMO maintain their own
meteorological agencies that obtain and process the data and
issue regional and national forecasts.
• Hydrogen-filled balloons carry weather instrument packages
called radiosondes.
• Radiosondes tracked by radar are called rawinsondes.

10.5 Forecast procedures and products
• The general procedure for all numerical models includes the

following three phases:
• The analysis phase, in which observations are used to supply
values corresponding to the starting state of the atmosphere
for all the variables carried in the model.
• Part of what analysis accomplishes is converting those
irregular observations into “uniform” initial values.
• Observed values also need to be consistent with the
particular model being used.
• All the models are approximations to the real atmosphere Æ
initial field not contain features that can’t be represented by the

10.5 Forecast procedures and products (cont.)
• The prediction phase begins with values delivered by the

analysis phase using governing equations to obtain new
values a few minutes into the future.
• The process is then repeated, using the output from the first
step as input for the next set of calculations.
• This procedure is performed over and over as many times as
necessary to reach the end of the forecast period (24hr, 48hr,
• Huge computational resources are needed for this.
• The governing equations can’t be solved directly but must be
broken down into simple operations that computers can

10.5 Forecast procedures and products (cont.)
• In the post-processing phase, conditions forecast by the

model at regular intervals are represented in grid form for
mapping and other display purposes.
• Forecasters study maps for each period and interpret the
conditions that would probably be associated with such
• In addition to gridded fields of model variables, forecasts for a
number of secondary variables are produced.
• The output products are called model output statistics (MOS)
and are designed to capture the effect of topography and other
factors that influence local weather conditions.

10.5 Forecast procedures and products (cont.)

Ensemble forecasting involves a number of
different model runs performed for
the same forecast period starting with
slightly different initial values.

10.5 Forecast procedures and products (cont.)
• If two model runs are made with slightly different initial
values, the results might be very different after a week or so.
• This behavior is now known to be typical of many natural
and human systems, and is referred to as chaos.
• Forecast at still longer lead times are called long-range
forecasts, ranging from a week to the limits of technical
• Another product is the “seasonal outlook”, a kind of forecast
for an entire season.
• In contrast to long-range forecasts that predict conditions for
particular days, seasonal outlooks predict average conditions
for an entire season.

10.6 Weather maps
• Although computers play a critical role in weather analysis,
ultimately meteorologists apply their knowledge to produce the
forecast that issued to the general public.
• Probably no tool is as valuable to a forecaster as a weather
• Weather analysis requires using a series
representing different layers of the atmosphere.



10.6 Weather maps (cont.)
• Surface maps of prevailing conditions present a general
depiction of sea level pressure distribution.

10.6 Weather maps (cont.)

Station models offer detailed knowledge of the conditions at a particular
location with over a dozen weather elements represented on each station

10.6 Weather maps (cont.)
A 850mb map resides at an average height of 1.5km above sea
level. Heights of the 850mb level are plotted with solid lines
(height contours in decameters).

10.6 Weather maps (cont.)
A 700mb map (~ 3km)

10.6 Weather maps (cont.)
A 500mb map (~ 5.6km)

Visible images view the atmosphere by registering
the intensity of reflected shortwave radiation and
are available only during the daytime.

Infrared images are based on measurements of
longwave radiation emitted (not reflected) from below.

Water vapor image obtained from GOES 8 satellite.

Radar images observe the internal cloud conditions
by measuring the amount of radiation backscattered
by precipitation (both liquid and solid).


Tài liệu bạn tìm kiếm đã sẵn sàng tải về

Tải bản đầy đủ ngay