13-1 PC-HYSPLIT WORKSHOP Workshop Agenda Introduction to HYSPLIT Introduction.ppt Model Overview Model_Overview.ppt Meteorological Data Meteorological_Data.ppt.

13-1PC-HYSPLIT WORKSHOP

Workshop AgendaWorkshop Agenda

Introduction to HYSPLIT Introduction.ppt

Model Overview Model_Overview.ppt

Meteorological Data Meteorological_Data.ppt

ERA_Example.ppt

hands-on_part1.ppt

Particle Trajectory Methods Trajectory_Methods.ppt

Real World Example – Balloon_flights.ppt

hands-on_part2.ppt

Pollutant Plume Simulations Pollutant_Plumes.ppt hands-on_part3.ppt hands-on_part4.ppt Real World Example – Arsenic.ppt

Special Topics Special_topics.ppt Real World Example – Volcanic_ash.ppt hands-on_part5.pptReal World Example – Mercury.ppt

Extra Topics Extra_Topics.ppt

TRAJECTORY CONFIGURATION


Trajectory Model Configuration


Now that you’ve had a brief introduction to running trajectories, a more thorough description of some of the trajectory GUI inputs will be presented. Open the Trajectory Setup menu.

Top of the model (m agl): the height above which the meteorological data are not processed. For calculations within the troposphere, the default of 10 km is a good value. Trajectories are terminated when they reach this height. Processing fewer levels reduces computational times.

Starting Location Setup: Enter a 1 for the Number of Starting Locations and then click the button to the right. A starting location(s) can be entered directly from this menu or a position may be chosen from a predefined list by clicking on the List button. This list is located in the \hysplit4\working directory, is user editable, and named plants.txt. Heights are always entered as meters above ground level (m-AGL) in PC HYSPLIT unless otherwise specified in the Advanced menu (discussed later). Enter the coordinates shown.




One key feature for any simulation is selecting the best meteorological data. HYSPLIT allows up to 12 meteorological files to be defined simultaneously. When multiple files are defined, at each integration step, the model uses the finest spatial resolution file at the location and time of the trajectory end-point. This can be useful if you have several higher resolution grids “nested” within each other.

The Add Meteorology Files button is used to select multiple meteorological data files. The Clear button will erase all selected files from the list. For the next example, Click the Clear button and then choose both the NAM 40 km and the GFS forecast files as shown below.




Now, set the Total run time to 84 hrs and the Vertical Motion Method to isobaric (1:isob, constant pressure) in order to compute an isobaric forward trajectory starting at 1200 UTC on February 17, 2009, from a height of 5 km at 46N, 68W.




The resulting trajectory (right) moves southeast using the NAM 40 km forecast file until running off the NAM domain (~ 50N). The model then uses the GFS for the remainder of the calculation. The CONTROL file can be viewed HERE.

The meteorological file identifier is written with each end-point position in the second column of the ASCII trajectory tdump output file, which can be viewed with the Trajectory / Utilities / Simple Listing menu. The diagnostic MESSAGE file can also be viewed from the Advanced/View MESSAGES menu. In this example the MESSAGE file shows the switch from NAM to GFS occurred at 2300 GMT on February 18, 2009, causing the 2100 and 0000 GMT data to be reloaded from the GFS.


TRAJECTORY COMPUTATION


Trajectory Computational Method


How is a trajectory calculated?

If we assume that a particle passively follows the wind, then its trajectory is just the integration of the particle position vector in space and time. The final position is computed from the average velocity at the initial position (P) and first-guess position (P').

P'(t+ Δ t) = P(t) + V(P,t) Δt

P(t+Δt) = P(t) + 0.5 [ V( P,t) + V(P',t+Δt) ] Δt

The integration time step is variable: Vmax Δt < 0.75 meteo. grid spacing




The meteorological data remain on the native horizontal coordinate system when read by HYSPLIT, but are interpolated to an internal terrain-following (σ) vertical coordinate system during computation:

σ = ( Ztop – Zmsl ) / ( Ztop – Zgl )

Ztop - top of the trajectory model’s coordinate system Zgl - height of the ground level Zmsl - height of the internal coordinate

The model’s internal heights can be chosen at any interval, however a quadratic relationship between height and model level is specified, such that each level’s height with respect to the model’s internal index, k, is defined by

Zagl = ak2 + bk +c

The constants are automatically defined such that the model’s internal resolution has the same or better vertical resolution than the input meteorological data.




To illustrate this process, a simple forward trajectory was calculated with archived NAM 40 km meteorological data from (60N, 110W) at 2500 m-agl for a total of 84 hours using the isobaric vertical motion method, as shown below.

The relationship of the trajectory to the temporal and spatial variations of the 700 hPA height field (approximate height of trajectory) is illustrated in the attached animation which was created using only the standard tools that come with PC HYSPLIT.


TRAJECTORY ERROR


Trajectory ErrorTrajectory Error

A common question often arises when running trajectories....

"What is the error associated with a given trajectory calculation."

From the literature, one can estimate the total error to be anywhere from 15 to 30% of the travel distance. The total error is composed of four error components:

physical error due to the inadequacy of the data’s representation of the atmosphere in space and time,

computational error due to numerical inaccuracies,

measurement errors in creating the model’s meteorological data fields,

forecast error if using forecast meteorology.

Physical Error

The physical component of the error is related to how well the numerical fields estimate the true flow field. There is no way of knowing this error without independent verification data.



Computational Error

The computational component of the error is composed of:

the integration error (part of which is due to truncation) and,

the data resolution error, i.e., trying to represent a continuous function, the atmospheric flow field, with gridded data points of limited resolution in space and time.

The integration error component of computational error can be estimated by computing a backward trajectory from the last forward trajectory endpoint. The error is then 1/2 the distance of the final endpoint from the starting point. Continuing with the last example, set the final endpoint position (36.848N, 38.597W, 5392.9 m AGL) from the last tdump endpoints file as the starting point (at 0000 UTC on the 21st) for a backward trajectory calculation. When running this case, insure that the endpoints file names are different for both the forward and backward calculations by setting this next tdump filename to tdumpback, for example.



To show the differences between the forward and backward trajectories, display both of them on the same plot by entering both file names in the Input Endpoints box of the Trajectory Display GUI using a + symbol

(e.g. ./tdump+tdumpback).

Note how the backward (blue) trajectory returns very close to the initial forward (red) starting location, indicating very little integration error. Most of the time the integration error is very small compared to resolution error. You can also see that the model automatically switched back to the NAM 40 km grid at 2000 UTC on February 18 by viewing the tdumpback endpoints file.



Resolution Error

The resolution error component of computational error can be estimated by starting several trajectories about the initial point. This is done by offsetting them in the horizontal and vertical (ensemble). The divergence of these trajectories will give an estimate of the uncertainty due to divergence in the flow field. An initial offset should be used that is comparable to the estimated integration error. In this ensemble example (right), there is little horizontal and vertical error until near the middle of the calculation when the trajectories encounter a frontal system and begin to diverge.

One component of the resolution error that is difficult to estimate relates to the size and speed of movement of various flow features through the grid. There should be a sufficient number of sampling points (in space and time) to avoid aliasing errors. Typically, a grid resolution of "x" can only represent wavelengths of "4x". This error will be a function of the meteorological conditions. 3-15PC-HYSPLIT WORKSHOP

The figures in the next slide show meteorological model surface wind fields in a high pressure system in the Mississippi / Louisiana area (lower figures) and in a low pressure system in the South Carolina / North Carolina area (upper figures) at a given time. The left figures are from the NAM output on a 40 km horizontal grid; the right from NAM output on a 12 km grid. Note the domains for the left and right plots are very close, but not identical.

In HYSPLIT, trajectory calculations are made using winds that are linearly interpolated in space and time. Part of the trajectory uncertainty is related to this interpolation.

The meteorological horizontal and temporal resolution needed depends on the real wind fields in the region and over the time period of interest.

Trajectory Error Trajectory Error


Resolution Error

SC / NCLow pressure

MS / LAHigh pressure

NAM 12 km

NAM 12 kmNAM 40 km

NAM 40 km



Another method of ascertaining the resolution error is to run trajectories from the same location using several different sources of meteorological data, which is a type of ensemble method that will be explained later.

In this example (right), trajectories have been computed using meteorological data from NAM(40 km)/GFS (red), NAM(12 km)/GFS (blue), and RUC/GFS (green). After the first 24 hours, the RUC/GFS (green) resolves the flow field differently than the NAM/GFS combinations. Note that these trajectories show smaller differences than most similar simulations due to the isobaric assumption that was made for this case.


MULTIPLE TRAJECTORIES


Multiple TrajectoriesMultiple Trajectories

Trajectories can be started at regular time intervals from the same location by setting the restart interval to something other than 0 hrs. Click on the Advanced / Configuration Setup / Trajectory menu tab which produces another menu (top right) for entries to the optional trajectory namelist file SETUP.CFG used by the model to set more options.

Click on the Multiple trajectories in time (3) menu button which produces the menu shown to the right. To demonstrate the restart feature, change the Restart interval to 3 hrs, click Save, click Save again, and leave all other trajectory parameters the same as in the original 5000.0 meter NAM 40/GFS isobaric trajectory previously computed at 46.0N, 68.0W. (To do this you will need to change the starting location and time, forward trajectory, and output filename to tdump.)

Multiple Trajectories from the same Location



After clicking on Run Model, a message box (shown at right) will appear to indicate that a SETUP.CFG configuration file exists and asks if you want to run with it. In this case since we made changes to the advanced settings, choose Run using SETUP file.

The resulting trajectory graphic shows new trajectories starting every 3 hours, terminating at the end of the 84 h computational period of the first trajectory. Trajectories starting after the initial time will have a shorter total duration as they all end at the same time. All trajectories can be set to have the same duration from the Advanced menu tab.

Multiple Trajectories from same Location (cont.)



Trajectory Matrix (grid of starting locations)HYSPLIT supports an unlimited number of starting locations, however the GUI is limited by the user’s screen size (GUI will extend off the bottom with too many selections).

Instead of manually adding many starting points, a matrix of starting locations can be defined by selecting three points, representing the lower left, upper right, and the location increment from the Starting Location Setup menu (see example below).

Once configured, the Matrix option is selected through the Special Runs menu tab of the Trajectory menu instead of Run Model.

This causes the locations defined by the grid (in this case 25) to be calculated and written to the CONTROL file.


Location 2

Location 1

Location 3


Rerun the 5000 m-AGL isobaric case again but change the Total run time to 24 hours and enter the 3 source locations from the previous slide.

From the Advanced / Configuration Setup / Trajectory menu choose Define subgrid and MSL/AGL UNITS (2) menu and set the height unit Relative to mean-sea-level because the terrain height varies across the matrix. Save.

Reset the Restart Interval back to 0. Save.

Save the advanced settings and run the Matrix from the Trajectory / Special Runs menu tab (Run using SETUP file).

When finished set the Time Label Interval (UTC) to 0 and the zoom to 80 in the Trajectory Display GUI before executing the display (this makes a less cluttered display).

Hands-on Trajectory Matrix Example

The resulting graphic should be similar to the one shown above; a matrix of 24-h duration isobaric trajectories. In this case, there is a large spatial difference between trajectories started to the north of the original location versus those to the south. Note that we are using spatial and temporal offsets to ascertain the trajectory error or, in this case, sensitivity to the meteorological data.


TERRAIN HEIGHT ISSUES


Terrain Height IssuesTerrain Height Issues

The trajectory starting height defaults to meters AGL (above ground level), but can be changed to meters MSL (above mean sea-level) from the Advanced / Configuration Setup menu tab as was done in the previous exercise.

Regardless of how the input heights are defined, internally HYSPLIT treats all heights in a terrain following coordinate system based on the chosen meteorological data.

These heights may be quite different from the actual terrain height at a point of interest.

As an example, examine the various model estimated terrain heights (right) for Broomfield (KBJC), Colorado, at 39.92N and 105.12W, which has a surface height of 1724 m above MSL.

Model HorizontalResolution

Terrain Height (MSL)

NAM 12 km 1970 m

NAM 40 km 1840 m

RUC 20 km 1890 m

GFS 1 deg. 2020 m



The terrain heights for the NAM 12 km (left) and GFS (right) are shown below (Bloomfield is indicated by the black star in the center). The terrain in the vicinity of Bloomfield is much smoother in the coarser GFS than the NAM and the terrain gradient is much steeper in the NAM and therefore we would expect to see differences in the terrains. Also, when the model terrain is consistently above the true terrain in all models, as in this case, one might suspect that the station is located in a valley.


NAM 12 km GFS 1 Deg.


• The average terrain value is shown by the solid blue line, which represents the model ground level (Zt)

• The sinusoidal solid black line shows the real terrain within that grid cell

• The dashed lines represent the top of the planetary boundary layer (PBL), the height to which the surface characteristics influence the meteorological profile

• Starting point “o” is within the valley, regardless of whether the height is specified as AGL or MSL, the starting point will always be at point “oo” (negative heights not allowed)

• Although “oo” may have a much higher absolute height (MSL) than “o”, it is correctly represented to be within the model’s PBL

• The starting point “p” is at the highest elevation, specifying the starting height as AGL will start the calculation at “pp” and correctly represent the position within the PBL.

• However, if the height is specified as MSL, the computation starts at “p” which is p-Zt above the model’s ground level and well above the PBL. This may result in unrepresentative initial transport winds.

Q: How do you define your starting height (MSL or AGL) if the model terrain is significantly different than the real terrain?

A: Make sure the resulting model starting height is representative of the initial transport winds at the true starting height.

VALLEY

HILL

1 grid cell

Terrain HeightTerrain Height

In the example to the right, we plotted 3 separate trajectory results on the same map using the NAM 40 km (blue), the RUC (green), and the GFS (red) originating from 10 m above model ground level (AGL).

Even though all the trajectories start out at the same height AGL, they start at different pressure levels, especially noticeable with the GFS, due to differences in elevation between the various datasets.

Some users set the model to start trajectories at a fraction of the mixed layer from the Advanced / Trajectory menu in an attempt to avoid these terrain issues.


METEOROLOGICAL ANALYSIS ALONG A TRAJECTORY


Meteorological Analysis Along a Trajectory


HYSPLIT can provide details on some of the meteorological parameters along the trajectory if the appropriate boxes are selected in the Advanced / Configuration Setup / Trajectory menu.

This information can be useful in diagnosing why a trajectory took the path it did, to show the underlying terrain height, to show the mixed layer depth along the trajectory, or to show whether any precipitation was being produced along the trajectory path by the meteorological model.

Currently only ambient and potential temperature, precipitation, mixing depth, relative humidity, solar radiation, and terrain height are available to output or display.

One or more meteorological fields may be selected and all will be written to the trajectory endpoints file in the rightmost columns, however only the rightmost (or last selected) variable will be plotted at the bottom of the trajectory map if the plotting option is enabled.




Click Reset from the main menu.

Configure the model for a trajectory originating at Broomfield, Colorado, (39.92N and 105.12W) using the NAM 40 km data.

Set the starting height to 1500 m AGL,

the vertical motion data field to 0 (Data),

and the total run time to 36 hours.

In the advanced trajectory menu click on Add METEOROLOGY output along trajectory (6)

Choose Terrain Height (right) and Save.

Save the configuration and choose Run using SETUP file. A flag is set in the SETUP.CFG when meteorological variables are selected.

In order to display the terrain height along the trajectory in the vertical cross-section at the bottom of the trajectory map, choose Meters-agl as the vertical coordinate in the Trajectory Display menu and set the Time Label Interval back to 3.

Hands-on Trajectory Example




For the first 12 hours, the resulting trajectory (right) follows the terrain as the terrain descends from Colorado into Nebraska.

If you choose Pressure as the Vertical Coordinate, the trajectory may look like it is descending in height, but in reality it is following the descending terrain, so care must be exercised when interpreting the up or down movement of trajectories with respect to height above model ground level and pressure vertical height coordinates.

The terrain heights can be viewed as the right-most column of the trajectory endpoints file (tdump)




Rerun the last case, but instead of Terrain Height, select Mixing Depth.

In order to display a meteorological variable other than Terrain Height, select Meteo-varb as the vertical coordinate in the Trajectory Display menu (upper-right).

This plot and the tdump file, show that the mixed layer depth along the trajectory varied from 16 m at 1500 UTC on February 17 to 442 m during the afternoon of the 18th.


VERTICAL MOTION OPTIONS


Vertical Motion OptionsVertical Motion Options

There are six vertical motion options in PC-HYSPLIT.

The suggested default (0:data) uses the vertical velocity field included with most meteorological data.

Other options may be required for special situations such as following the transport of a balloon on a constant density surface, comparing isobaric flow fields between data sets, or situations when the meteorological data’s vertical velocity field may be too noisy compared with the time step at which the data are available (high spatial resolution simulations).

In the (4) sigma option the trajectory remains on its original terrain following sigma surface.

In the (1) isobaric, (2) isentropic, and (3) constant density (isopycnic) options, the vertical velocities are computed from the equation,

W = (- ∂q/∂t – u ∂q/∂x – v ∂q/∂y) / (∂q/∂z)

where W is the velocity required for the trajectory to remain on the q surface (pressure, potential temperature, density). Note that the equation results in only an approximation of the motion and a trajectory may drift from the desired surface.

In the (5) divergence option, the vertical velocities are computed using the vertically-integrated horizontal divergence from the meteorological data.


Vertical Motion OptionsVertical Motion Options

Shown below (left) is the same trajectory of the previous example (with terrain) using the NAM 40 km vertical velocity fields. To the right is the same trajectory computed using the divergence option. This graphic shows that the choice of vertical motion can impact the transport direction as the divergence method introduced more vertical motion near the source and hence the particle entered somewhat more northerly transport winds near the end of the calculation.


Data VV Divergence VV

13-1 PC-HYSPLIT WORKSHOP Workshop Agenda Introduction to HYSPLIT Introduction.ppt Model Overview Model_Overview.ppt Meteorological Data Meteorological_Data.ppt.

Documents

pc hysplit

hysplit introduction

meteorological files

list button

trajectory setup menu

trajectory endpoint

meteorology files button

multiple files