Model Concentrations & Tracer Experiment Verification Instructions
Model concentrations & Tracer Experiment Verification (McTEV) is designed primarily for HYSPLIT users to compare model simulations, using different configuration parameters, with measured concentrations from the following tracer experiments:
Details of an experiment can be found by clicking on the experiment name (Home Page).
A) New-run for a Tracer Experiment (Home Page):
From "New-run", select a tracer experiment from the scroll-down list and click "Next". The North American Regional Reanalysis (NARR) meteorological fields are used for New-runs with the exception of ETEX and SRPTEX New-runs which use the Global Reanalysis meteorological fields.
Select the desired "Release Location" and "Release Date" from the scroll-down and click "Next".
A Base-run has already been made by ARL for your selection. For your New-run, you will change the Base-run configuration parameters; "Particles or Puffs", "Initial Number of Particles or Puffs", or "Averaged Concentration Layer Top" from the scroll-downs and click "New-run" at the bottom of the screen. No more than 2 parameter changes per New-run is recommended.
After the New-run is completed on the HYSPLIT Model DATEM Simulation screen, as indicated in the "Model Status" box, HYSPLIT Model Results for your New-run are given with a unique Job Number that must be applied in using B) (below). New-run Results include:
"Measured vs New-run Concentrations & Statistics Tables" with the concentration table showing sampling site location information, sampling dates and times, and concentration values, and the statistics table summarizing the New-run statistics. It should be noted that the percentiles in this statistics table include measured and calculated 0 values, where as they are excluded in boxplot distributions and boxplots described below.
"Measured vs Base-run & New-run Concentrations - Scatter Plot" with the Base-run parameter change(s) for the New-run at the lower left (all plots can be viewed in both GIF or PS formats).
"Base-run vs New-run Concentrations - Normalized Standard Deviation Plot" (including numeric values). Also included is a Final Index (FX) at the top for the Base-run (B) and New-run (N), Base-run parameter change(s) for the New-run at the lower left, and a log ratio, log{ FX(N)/FX(B)}, at the upper right. (This log ratio is a robust number for run comparisons.) FX combines the normalized standard deviation (NSD) and boxplot characteristics as defined by a Boxplot Index (BX) (see 4.4 below) where:
NSD=SQRT[SUM(M-C)**2/(n * SUM(M)/n * SUM(C)/n)] for
M=measured concentration
C=calculated concentration
n=total number of MC comparison pairs
and
FX = (NSD)*(BX)
"Measured, Base-run & New-run Concentrations - Boxplots" with a key boxplot at the left, Base-run parameter change(s) for the New-run at the lower left, and the Boxplot Index (BX) at the top where:
BX = SUM(i)[(abs(Mi - Ci) * (Pi - P(i-1))/100]
i=0,10,25,50,75,90
Mi=measured concentration at percentile i
Ci=calculated concentration (Base-run or New-run) at percentile i
Pi=boxplot percentile value i and P(i-1)=next lowest boxplot percentile value
B) Results from a Previous New-run (Home Page):
From "Results from a Previous New-run" select "Click Here to Access".
Select the desired "(Experiment Name) New-run Results" and from the desired job number scroll-down select "Submit". The results for that job number are displayed similar to 4 above.
C) Results from All New-runs (compiled by ARL) (Home Page):
From "Results from All New-runs" select "Click Here to Access".
"Select the Parameters" from the first 2 scroll-downs (the parameter from the second scroll-down is optional). "Select the Statistic" from the scroll-down. Select "Get Results". The displayed plot shows the chosen statistic for the chosen New-run parameter(s) for all tracer experiments, which are defined along the vertical axes at the right (experiment) and left (release during the experiment). The horizontal solid bars on the plot reflect the values of selected parameter #1 (black) and optional selected parameter #2 (red). The zero value (red) vertical line on the plot can be used as an aid to separate more robust New-run(N) vs Base-run(B) values.
