Strange output when hitting the ground

Topics about the HYSPLIT trajectory model.
Post Reply
khartig
Posts: 16
Joined: June 16th, 2020, 4:22 pm
Registered HYSPLIT User: Yes

Strange output when hitting the ground

Post by khartig »

I'm running HYSPLIT trajectories on Linux (compiled from source code, Last Changed Rev 1011), using data I've converted from the Community Atmosphere Model v4 to ARL with MeteoInfo. Many of my trajectories are close to the ground, so I expect some to hit ground, but I am confused by the output produced in the trajectory files and would like to know more about how "grounding" is determined. I've seen three different behaviors that I would like to understand:

1. Switch to NaN: sometimes, when the height is very low, all columns from the latitude column on will become NaN:

Code: Select all

     1     1    11     1    18     5     0    99   -25.0   44.440 -111.338      0.3    786.2    248.2   1864.3
     2     1    11     1    18     5     0    99   -25.0   46.052 -109.776      0.0    824.1    243.8   1517.5
     3     1    11     1    18     5     0    99   -25.0   46.920 -108.695      0.0    865.8    243.8   1154.2
     4     1    11     1    18     5     0    99   -25.0   38.079 -109.228    144.2    768.0    267.3   2018.2
     5     1    11     1    18     5     0    99   -25.0   40.467 -120.049   4855.3    433.6    234.7   1437.5
     1     1    11     1    18     4     0    99   -26.0   44.442 -111.374      0.8    786.3    248.6   1862.8
     2     1    11     1    18     4     0    99   -26.0   46.056 -109.825      0.0    823.5    244.4   1522.2
     3     1    11     1    18     4     0    99   -26.0   46.928 -108.785      0.3    864.7    244.3   1161.7
     4     1    11     1    18     4     0    99   -26.0   37.715 -109.474    109.2    773.8    267.9   1999.5
     5     1    11     1    18     4     0    99   -26.0   41.374 -120.266   4947.3    435.0    234.9   1324.2
     1     1    11     1    18     3     0    99   -27.0      NaN      NaN      NaN      NaN      NaN      NaN
     2     1    11     1    18     3     0    99   -27.0      NaN      NaN      NaN      NaN      NaN      NaN
     3     1    11     1    18     3     0    99   -27.0      NaN      NaN      NaN      NaN      NaN      NaN
     4     1    11     1    18     3     0    99   -27.0   37.374 -109.690     74.4    779.8    268.6   1982.3
     5     1    11     1    18     3     0    99   -27.0   42.310 -120.455   5040.4    436.7    234.9   1202.4
     1     1    11     1    18     2     0    99   -28.0      NaN      NaN      NaN      NaN      NaN      NaN
     2     1    11     1    18     2     0    99   -28.0      NaN      NaN      NaN      NaN      NaN      NaN
     3     1    11     1    18     2     0    99   -28.0      NaN      NaN      NaN      NaN      NaN      NaN
     4     1    11     1    18     2     0    99   -28.0   37.096 -109.923     43.7    784.9    269.2   1957.8
     5     1    11     1    18     2     0    99   -28.0   43.262 -120.624   5140.0    438.2    235.1   1093.9
2. Recovery from height of zero: in other cases, trajectories will sit at a height of 0.0 meters for many time steps without becoming NaN, and then recover and increase in height again

3. Row of asterisks: sometimes, the switch to NaN as the trajectory presumably hits the ground will be preceded by a row of asterisks (see trajectory 1 in sample_traj_asterisks.txt)

Code: Select all

     1     1    15    12    30    15     0    99   -21.0   28.243  -82.265     15.0   1024.5    285.9   -112.1
     2     1    15    12    30    15     0    99   -21.0   33.475  -84.965    111.1    988.4    274.9     82.0
     3     1    15    12    30    15     0    99   -21.0   33.758  -86.949    318.6    965.4    277.5     65.5
     4     1    15    12    30    15     0    99   -21.0   33.949  -88.400    672.7    928.3    277.8     33.7
     5     1    15    12    30    15     0    99   -21.0   44.433 -110.504   5582.3    377.4    237.2   1812.3
     1     1    15    12    30    14     0    99   -22.0*********  -82.368 ******** ******** ******** ********
     2     1    15    12    30    14     0    99   -22.0   33.505  -85.006    118.9    986.2    275.2     98.4
     3     1    15    12    30    14     0    99   -22.0   33.778  -87.046    324.1    964.5    277.5     67.0
     4     1    15    12    30    14     0    99   -22.0   33.954  -88.617    674.6    928.6    277.9     27.0
     5     1    15    12    30    14     0    99   -22.0   44.862 -112.152   5663.4    380.9    237.6   1696.1
     1     1    15    12    30    13     0    99   -23.0      NaN      NaN      NaN      NaN      NaN      NaN
     2     1    15    12    30    13     0    99   -23.0   33.547  -85.068    127.6    984.3    275.5    100.1
     3     1    15    12    30    13     0    99   -23.0   33.819  -87.158    329.9    963.8    277.6     66.3
     4     1    15    12    30    13     0    99   -23.0   33.970  -88.828    677.2    929.1    278.0     20.6
     5     1    15    12    30    13     0    99   -23.0   45.156 -113.749   5758.5    386.9    238.4   1501.9
     1     1    15    12    30    12     0    99   -24.0      NaN      NaN      NaN      NaN      NaN      NaN
     2     1    15    12    30    12     0    99   -24.0   33.603  -85.154    137.3    982.4    275.7    102.3
     3     1    15    12    30    12     0    99   -24.0   33.881  -87.284    335.7    963.1    277.6     66.3
     4     1    15    12    30    12     0    99   -24.0   33.995  -89.033    680.2    929.4    278.1     14.6
     5     1    15    12    30    12     0    99   -24.0   45.332 -115.290   5822.3    392.6    239.3   1337.5
What leads to these three different output behaviors? I would attach files with examples, but the post interface says the board attachment quota has been reached and won't let me.
domanton
Posts: 2
Joined: January 19th, 2021, 1:00 pm
Registered HYSPLIT User: No

Re: Strange output when hitting the ground

Post by domanton »

Interesting. I got the same results when I tried. anyone know how to correct it.
Fantine
Posts: 150
Joined: November 8th, 2012, 3:41 pm

Re: Strange output when hitting the ground

Post by Fantine »

About trajectory hitting the ground, I would recommend looking at these posts on the forum where the issue has been discussed.
viewtopic.php?t=295
viewtopic.php?t=1523

The "NaN" and asterisks in the trajectory output are more like a numeric error. For example, something is divided by zero or the values are too (unreasonable) large exceeding the printing format. Have you checked the meteorological data (the ARL file) used to run the trajectories? You can use the contour utility to generate spatial plots and profile utility to print out vertical profiles. I will suggest especially look at the vertical velocity because the trajectory endpoints stayed at zero and then jumped up to almost 5000 m.
khartig
Posts: 16
Joined: June 16th, 2020, 4:22 pm
Registered HYSPLIT User: Yes

Re: Strange output when hitting the ground

Post by khartig »

Thanks for the suggestions, I think I figured it out. I'm working with CESM/CAM data, and for a while I provided the 10-m zonal wind U10 but not the meridional wind V10. I have since discovered that the 'U10' variable provided in CAM4 is actually the total 10-m wind speed, not the zonal velocity, and it was removed for CAM5+. Once I stopped providing just U10, I stopped getting these odd errors. Figured it out by plotting the trajectories on a map; sometimes when the trajectory dropped below the lowest data level, into the region where wind is extrapolated down to the surface, the meridional wind component seemed to adopt huge values and the trajectory would suddenly jump many 10s of degrees in latitude. I no longer provide either U10 or V10 and have not gotten NaNs in my trajectories since; providing only one but not both seems to do strange things without always raising an error.
Post Reply

Return to “Trajectory Model”