WEATHER FORECASTING
ATMOSPHERIC MODELS
scientific model- approximate representation of simulation of a real system
-eliminates all but the essential variables or characteristics that can change
specific types
conceptual model-describes the general relationships among components of a system. geostrophic wind is such a model that relates the interaction of certain forces operating in the atmosphere to straight, horizontal air movement above an altitude of 1 km
graphical model-assembles and displays data in an organized format that can be readily interpreted - an example would be a weather map
physical model is a miniaturized version of some system e.g. dishpan model
numerical model--consists of one or more mathematical equations that portray the behavior of a particular physical system such as the atmosphere. Variables in the numerical model such as temperature or humidity may be manipulated, individually or in groups in order to assess the impact on the system.
ASSEMBLING WEATHER DATA
10,000 land based stations and hundreds of ships
Readings taken 4 times a day
-every hour at airports
-radiosonde at 0000 and 1200 GMT
World Meteorological Organization
-responsible for international exchange
-checks observation procedure
-data collected and sent to WMO centers in Moscow, Wash. and Melbourne
-in Washington goes to Suitland Md
-charts prepared and sent to WSFO ( Weather Serv. Forec. Off)
- one per state
-regional forecasts prepared
- 200 WSO (weather service offices) issue local forecasts
DATA ANALYSIS (synoptic)
Synoptic charts
Isobaric Analysis
Frontal Analysis
Isallobaric Analysis
-pressure changes in past three hours
Upper air analysis
-stream lines
-isohypses-lines of constant height (reverse isobars)
HISTORY OF WEATHER FORECASTING
A. Empirical Era (l860-l920)
1. -first regular issue of storm warnings by a national weather service
was begun in Netherlands in l860 and England in l86l
2.-United States was l870 when Army Signal Service made them
3.-basic tool was synoptic weather map
4. -first maps were made in late l700's but it took too long to gather
the information for them to be useful
5.-for the remainder of the century experience gained in the use of
weather maps remained almost the sole basis of prediction
6. -past movements of pressure systems, statistics on storm tracks and a host of empirical rules were employed to project low pressure centers and other features forward in time
6a -during this time the two fundamentals of forecasting were
-weather travels
-the character of the weather is largely determined by atmospheric pressure distribution
7. -basic physical concepts and theoretical concepts played little if any role to WWI
-Ferrel began his work on general circulation in the l850's and in the
l860's and l870's papers were presented on atmospheric thermodynamics
-problems with lack of upper air measurements
B. Transitional Era (l920-l950)
-in l9l8 Vilhelm Bjerknes founded the Geophysical Institute in Bergen
-due to the war and lack of weather info a dense network of stations were established in Norway itself
-Jacob Bjerknes, his son, was interested in direct analysis of the surface wind field and in studying moving cyclone structure
-identified the warm and cold fronts, precip. and cloud patterns associated with same
-identified the source of kinetic energy of storms
-four years later Bjerknes began work on the life cycle of cyclone and the polar front theory of atmospheric circulation
-was l938 before ideas were adopted by National Weather Service
-development of upper air observations with radiosonde
-as a result Carl Rossby at Chicago introduced the concept of
jet streams in the l930's and documented waves in the westerlies
-weather radar in the late l940's
-development of dynamic principles such as the conservation of
absolute vorticity
C. Scientific Era (l950 to present)
-marked the attainment of successful weather prediction by physical
numerical methods into routine operational use
-marked the end of meteorology as an art
-must determine state of atmosphere at the present time and then
arrive at future conditions through solving hydrodynamic equations.
-necessity for high speed electronic computer
TYPES OF WEATHER FORECASTING
Synoptic Weather Forecasting
-based on interpretation of synoptic weather map
-synoptic means coincident in time
-primary method used in making weather predictions
until the late 1950's
-synoptic weather charts are the basis of the forecast
-empirical rules
-importance of upper air data
Steady State or Trend
-based on continued movement of systems in same direction
Statistical (Analog)
-also known as prediction by weather types
-analyzing past data and maps to establish predictive relationship
(cycles) then utilize current data to obtain predictor.
-use of analogues (past similar weather)
-most often used to determine one aspect of weather at a time
-e.g. maximum temp.
-done by compiling statistical data relating temp to wind speed
and direction, cloud cover, humidity and season
Dynamic (Numerical Weather Prediction)
-the process of solving the equations that govern the behavior
of the atmosphere, starting with approximately known
initial and boundary conditions
-l904 Vilhelm Bjerknes
-stated the central problem of numerical prediction
-made first explicit, coherent recognition that the future
state of the atmosphere is, in principle, completely determined
by its detailed initial state and known boundary conditions,
together with Newton's equations of motion, the Gas Laws
principle of mass continuity and thermodynamic energy equation.
General Comments
-numerical weather prediction is best for general flow
-modified using statistical forecasting
-further modified using synoptic
NUMERICAL WEATHER ANALYSIS AND FORECASTING
First step in developing a forecast is to summarize values
of different atmospheric variables such as wind speed and
pressure measured at a large number of scattered locations
as the average values that apply at the intersection of an
imaginary 3 dimensional grid wrapped around globe.
Problems with small scale atmospheric phenomena.
-relatively energetic
-fall within grid cells
-i.e. most of vertical motion n tropics occurs in isolated
storms covering .1% of area.
Keeping track of 7 atmospheric variables on a grid 200 km on
a side and 10 layers deep involves 1 million variables. 500
arithmetic operations are required to compute interactions to which
1 variable is subject so that one half billion operations occur
for each 10 minute step.
-10 fold increase in spatial resolution increases variables
1000 fold and number of operations would increase 10,000 fold
-when distance between grid points is halved, 8 times the number of computations is required and time required goes up by a factor of 16
Starts with a current analysis of state of atmosphere
Composed of individual analyses of the several meteorological
variables
-involves charts displaying the spatial distribution of quantities
-simultaneous observations of meteorological variables
-surface temp, d.p. pressure, wind, clouds, visibility and weather
-data are encoded and sent to Wash
-data are retrieved, analyzed and electronically plotted on
certain analyses charts
Each analysis consists of a family of lines, each connecting points
of the same value of the particular datum
-includes upper air charts at 1000 mb 850 700 500 300 200
Much of the meteorological science involves boundary value or boundary
condition problems
-requires intimate knowledge of initial meteorological conditions
at boundary locales
-invariably data are missing, incomplete or sparse
The mathematical expressions on which computer programs are founded are
intractable because of complexity
-therefore we simplify
-omit some of the mathematical variables under the assumption that
because they are small in magnitude their importance is negligible.
-often we have to deal with very small differences between or changes
in very large quantities
-instrumental or observational errors may exceed the desired quantities
-errors are propagated inward to the center of the problem area and
are magnified over time
-after 3 to 4 days computer programs depart seriously from reality
One can think of Numerical Weather Prediction like a movie camera
-takes pictures at small but finite intervals
-the shorter the more precise
-grid point representations
-tough to show small scale phenomena
Mathematical Equations used in Numerical Models
-Newtons laws
-first law of thermodynamics
-continuity equations
-water vapor
Computer Generated Prognoses
-in the US the National Meteorological Center of the National
Weather Service prepares basic large scale forecasts of 48 states
-prognoses are transmitted to local forecast offices
-some are manual and some are computer generated.
In the computer generated prognoses
-data from randomly distributed observations stations are
computer interpolated to regularly spaced points on a grid
covering the forecast area
-final product is called a prognostic chart or prog
- 2 models
a. seven layer primitive equation (PE) model
b. LFM or Limited Area Fine Mesh
-tighter interpolation over smaller area
-grid points are 130 km apart
-can't predict thunderstorms or hurricanes directly
though with a special 60 km mesh hurricanes are possible
-originally 6 levels with 125,000 grid points at each level
-4.5 million equations with 4.5 million unknowns and is
good only for 4 minutes
PE Model
-refers to equations of motion for a fluid where the primary
dependent variables, those that depend on the instantaneous
state of the atmosphere are the velocity components
-values for variables are assigned to grid points for each time
step in the forecast
-equations are filtered to remove effect of physical disturbances
believed to have negligible effect on weather.
Accuracy of National Weather Service
-best with wind direction, speed and temp
-when off with temp it usually has to do with frontal movement
and nighttime cloud cover
-worst with precipitation amount
TYPES OF WEATHER FORECASTING (Time)
l. persistence forecasts
-6 hours of less
-projection of past into future
2. Meteorological Forecasts
-most accurate
- 6 hrs. to 4 days
3. Climatological Forecasts
-upper air circulation
-sunspots
WEATHER SATELLITES
I. Polar Orbiters (Tiros and Essa)
-ITOS (Improved Tiros operational satellites) and NOAA 1,2
-Landsat is a polar orbiter
II Geostationary Satellites
-35,000 km above equator
-GOES Geostationary Operational Environmental Satellites
RADAR
Backscatter from liquid and solid water particles depends strongly
on their size and wavelength of radio waves. With radio wavelength
of 1 cm rain and snow of greater than 1 mm diameter can be detected.
Small cloud droplets scatter too little to get a return of signal
Significant developments of the l970's
-radar meteorology is no longer the exclusive domain of the radar
meteorologist and is being used by almost all meteorologists
as observing tools
-the three dimensional kinematic structure of precipitating systems
can be determined by Doppler techniques
-Doppler radar measures the frequency shift between a transmitted
microwave or light (Doppler Lidar) beam and its echos from
water droplets or dust particles in the atmosphere
-the frequency is shifted upward if the target is moving
toward the radar and downward if it is moving away
and not at all if it is moving sideways
-the wind velocity component along the line of sight between
the instrument and the target is proportional to the magnitude
of the frequency shift
-resolution is currently limited to 300 m by l75 m pixels
-Doppler Lidar has better resolution because it has a
narrower beam
-acceptance of Doppler radar as a viable tool for operational
applications
-initially used for severe storms (tornadoes)
-further uses
-winds in mid latitude cyclones
-thunderstorm gust fronts
-hurricane winds
WEATHER FORECAST ACCURACY
AMS l979
l. Up to 48 hours
-considerable skill in forecasts of cloudiness, air quality, temp
and precip when weather is dominated by large scale weather
systems
-general area of severe storms can be predicted up to 24 hours
in advance
2. 2 to 5 days
-daily temp forecasts of moderate skill and usefulness are possible
-precip forecasts of moderate skill up to 3 days
- at 4 to 5 days is only as good as climatology
3. 5 days to a month
-average temp conditions can be predicted with some skill
6-l0 days
-slight skill for 6-l0 day precip forecasts
-greater than 10 days no skill
4. longer than one month
-minimal skill exists in seasonal outlook
FUTURE IMPROVEMENT
Probably less in resolution than in calculations discriminating more
accurately between phenomenon that have a significant impact.
PROBABILITY FORECASTS
-a forecast stating there is a 60% probability of rain means that there is a 60% chance that any point in the forecast area will receive precipitation
ACCURACY AND SKILL
-problem of determining what constitutes a right or wrong forecast. e.g. is a forecast bad if it is 1, 2 or 3 degrees off
-skill versus what could be predicted from climate data
-meteorological forecasts show skill when they are more accurate than a forecast utilizing only persistence or climatology
AFOS
Automation of Field Operations and Services
Now Working onAWIPS Advanced Weather Interactive Processing System
-will integrate NEXRAD data
Automated Surface Observing System (ASOS)
-continuous information on wind, temp, pressure, cloud base height and runway visibility
Ensemble Forecasting
-baed on running several forecast models with slightly different weather
information. If they match fairly well the meteorologist can have a high
degree of faith in the forecast