A kind of random physical perturbation motion method based on the conservation of energy
Technical field
The present invention relates to weather forecast field, more particularly to a kind of random physical perturbation motion method based on the conservation of energy.
Background technology
A series of past researchs show that storm scale systems DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM is feasible and effective, but each storm ruler at present
The perturbation motion method for spending ensemble prediction system is also to be improved.Such as the analysis of U.S.'s storm and the storm scale systems collection in forecasting centre (CAPS)
Close forecast system, by improving number of members and resolution ratio can improve forecast score, but its gather disturbance scheme be it is fixed,
The initial value disturbance of i.e. each set member is matched using fixed response excursion with specific physical schemes;Partial set into
The lateral boundary conditions of member are provided by the short-period forecast of NAM patterns, and identical lateral boundary conditions will limit to a certain extent between member
The development of dispersion;On the other hand, which does not consider to disturb the mismatch problem of scale between lateral boundaries disturbance and initial value disturbance.
Above-mentioned typical problem has embodied in each storm scale systems ensemble prediction system, initial value disturbance, physics (parameter) disturbance, side
The success or not that the combination problem that boundary is disturbed between initial value disturbance is for storm scale systems DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM plays a key effect.
Since initial error with the differentiation of time has significantly difference, mid-term in baroclinic instability and convective instability
The initial disturbance of set perturbation motion method construction can not quickly increase in contracurrent system, namely initial disturbance structure and storm system
The development of system is incompatible, and then causes set member's dispersion horizontal less than rational;On the other hand, the initial value of storm scale systems
The scale of disturbance and lateral boundaries disturbance is incompatible will also to be limited initial disturbance and gathers the growth of dispersion.More physical processes and
Multi-mode set helps to improve dispersion problem, but Rainfall forecast and Microphysical mistake in the Microphysical scheme of pattern
The threshold condition of journey judges it is different, and precipitation intensity and scope forecast of settling in an area are significantly affected, occurred for real-time
Different type storm system, using it is changeless set configure associativity choose different mode Microphysical scheme be blindly
's.
The content of the invention
Present invention aims to overcome that the deficiency of the above prior art, there is provided a kind of random physical based on the conservation of energy is disturbed
Dynamic method, is specifically realized by following technical scheme:
The random physical perturbation motion method based on the conservation of energy, includes the following steps:
1) according to the selected ARPS patterns for carrying out numerical forecast are actually needed, each ginseng in model predictions equation is then determined
The position of the final tendency item of numberization scheme output;
2) disturbance of the tendency item is determined by integrating further according to disturbance form after the position of the parameter is determined
Scheme, the expression formula for disturbing form are:XP=(1+rX)Xc,
Wherein XcRepresent the time tendency item of all parameters, XpRepresent the time tendency item after disturbance, rXRepresent increase
Disturbance, rXThe form of Autoregressive Functions isWherein α is the decorrelation time
Scale, t represent that the present mode moment be t, and ε (t) is the random number for meeting uniform distributions, when each time use it is different with
Machine generation, the maximum and minimum value of uniform distributions are set as 0.5 and -0.5, and σ is the variance function of disturbance, with height
Degree change, distributional pattern and the homomorphosis of tendency of above-mentioned time;
3) two-dimension fourier decomposition is carried out to ε (t) in the horizontal direction, then filtering out high-frequency noise will wherein 1 to 4 times of lattice
Away from the coefficient of ripple of wavelength be assigned a value of 0, the coefficient after adjustment is finally synthesized into new ε (t) by inverse Fourier transform;
4) repeat step 3) until obtaining the corresponding coefficient of disturbance r of each variableX, and bring into the table of the disturbance form
Up to formula, the variable includes u, v, w, one temperature variable T of a three wind field components and steam variable qv.
The further design of the random physical perturbation motion method based on the conservation of energy is that the prognostic equation of u is:
Wherein can be disturbed time tendency item beRepresent time u variables of grid turbulence parameterized procedure output
Time tendency.Remaining item cannot be disturbed, wherein (ρ * u)tIt is inclined to for the full time of u variables,For pressure gradient-force item ,-ADV (u) is advective term,For geostrophic deviating force item.
The further design of the random physical perturbation motion method based on the conservation of energy is that the prognostic equation of the T is:
The tendency item for including time grid turbulence parameterized procedure that can be disturbed in formulaWith Microphysical scheme and
The tendency item of the nonadiabatic process of irradiation protocolWhereinTemperature tendency including cumulus parameterization scheme output
Item and the temperature tendency item of irradiation protocol output;Its remainder cannot be disturbed, wherein (ρ * θ)tIt is inclined to for the full time of position temperature
,For convective term ,-ADV (θ ') is the advective term of position temperature.
The further design of the random physical perturbation motion method based on the conservation of energy is, the prognostic equation of the qv
For:
The tendency item for including time grid turbulence parameterized procedure that can be disturbed in formulaLed with Microphysical scheme
The tendency item of the steam change of causeRemaining item cannot be disturbed, (ρ * q)tIt is the full time tendency item of steam ,-
ADV (q) is the advective term of steam, (ρ * Vqq/zζ)ζIt is the flux term of steam.
Advantages of the present invention is as follows:
Random physical perturbation motion method provided by the invention based on the conservation of energy inclines completely self-contained physical process
To disturbance, it is therefore intended that conservativeness of the tendency disturbance of Assured Mode variable on energy so that this method can improve including side
The set dispersion of interlayer and convective region internal schema variable, this will significantly improve the forecast water of storm scale systems DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM
It is flat.Compared with traditional disturbance being inclined to full time, disturbance scheme of the invention can be balanced in itself by pattern to be added
The disturbance entered so that significant change will not occur for the gross energy of pattern, and (weather system especially contracurrent system just wraps when developing
Contain the factor for making it wither away, therefore during a part of physical process in only perturbation mode, other physical processes of pattern
The adjustment that can be responded, offsets the energy introduced due to artificial disturbance to a certain extent.And traditional scheme is to full time
Tendency is disturbed, and allow for that the values of all physical processes all unified zooms in or out, therefore can not ensure keeping for energy
It is permanent).Meanwhile disturbance scheme of the invention can use different disturbance shapes to different physical processes according to the characteristics of the process
State so that disturbance is more representative compared with conventional method.
Embodiment
Below in conjunction with the accompanying drawings and specific embodiment, the present invention is furture elucidated, provided by the invention based on the conservation of energy
Random physical perturbation motion method mainly includes:
1) according to the selected pattern for carrying out numerical forecast is actually needed, parameters in model predictions equation are then determined
The position of final tendency item;
2) disturbance of the tendency item is determined by integrating further according to disturbance form after the position of the parameter is determined
Scheme, the expression formula for disturbing form are:Xp=(1+rX)Xc,
Wherein XcRepresent the time tendency item of all parameters, XpRepresent the time tendency item after disturbance, rXRepresent increase
Disturbance, rXThe form of Autoregressive Functions is
Wherein ε (t) is the random number for meeting uniform distributions, secondary when each to be generated using different random seeds,
The maximum and minimum value of uniform distributions are set as 0.5 and -0.5, and σ is the variance function of disturbance, with height change, distribution
The homomorphosis of form and tendency of above-mentioned time;
3) two-dimension fourier decomposition is carried out to ε (t) in the horizontal direction, then filtering out high-frequency noise will wherein 1 to 4 times of lattice
Away from the coefficient of ripple of wavelength be assigned a value of 0, the coefficient after adjustment is finally synthesized into new ε (t) by inverse Fourier transform;
4) repeat step 3) until obtaining the corresponding coefficient of disturbance r of each variableX, and bring into the table of the disturbance form
Up to formula.
The specific implementation step of the above method is given below:
(I.1) according to the selected pattern for carrying out numerical forecast is actually needed, the present embodiment is by taking ARPS patterns as an example.
(I.2) position of the final tendency item that parameters scheme exports in model predictions equation is determined.Any one
The prognostic equation of pattern variable has multiple times tendency items, the tendency of final time of variable be above-mentioned time tendency item and.
The time tendency item of random process only disturbance parameter scheme output.The prognostic equation that random perturbation is related to includes three wind fields point
Measure u, v, w, one temperature variables T, a steam variable qv.Wherein u is similar with the form of v, and by taking u as an example, its expression formula is such as
Under:
WhereinFor the time tendency of the u variables of secondary grid turbulence parameterized procedure output.DetermineIn program
In position.Similarly it needs to be determined that v and wWithPosition.
The prognostic equation of T is:
In formula, include the tendency item of secondary grid turbulence parameterized procedureIt is non-with Microphysical scheme and irradiation protocol
The tendency item of adiabatic processWhereinInclude, but are not limited to temperature tendency item and the radiation of Microphysical scheme output
The temperature tendency item of scheme output.At least this two are disturbed.Determined according to Mode EquationWithIn journey
Position in sequence.
The prognostic equation of qv is:
Include the tendency item of secondary grid turbulence parameterized procedureThe tendency changed with steam caused by Microphysical scheme
DetermineWithPosition in a program.
(I.3) determine it is above-mentionedWith's
The disturbance scheme of these tendency items is determined behind position again.Determine disturbance form be:Xp=(1+rX)Xc, wherein XcRepresent above-mentioned 7
A time is inclined to item, XpRepresent the time tendency item after disturbance, rXRepresent increased disturbance.Xp=(1+rX)XcIt will be used in every
On one step integration.
(I.4)rXIt is different on each step integration, in order to ensure r at the same timeXRandomness and continuity, use from
R when regression function determines to integrate in next stepXNumerical value, the form of Autoregressive Functions is:
Wherein, ε (t) is the random number for meeting uniform distributions, secondary when each to be generated using different random seeds.
The maximum and minimum value of uniform distributions are set as 0.5 and -0.5.σ is the variance function of disturbance, with height change, distribution
The homomorphosis of form and tendency of above-mentioned time.
(I.5) two-dimension fourier decomposition is carried out to ε (t) in the horizontal direction.Then by wherein 1-4 times of lattice away from wavelength
The coefficient of ripple is assigned a value of 0, for filtering out high-frequency noise.The coefficient for contributing small wavelength is also assigned a value of 0 again.After finally adjusting
Coefficient new ε (t) is synthesized by inverse Fourier transform.
(I.6) analysis of (I.4) and (I.5) is carried out to each variable, obtains the coefficient of disturbance r of each variable oneselfX。
(I.7) r of each variable (I.6) obtained is usedX, substitute into perturbation equation (I.3), by this perturbation equation
Each time on integration of different variables is reapplied, just completes the random physical process disturbance thing to be done of scheme.