Experiment 5 (7/25/2016)


This experiment uses the same simulated passive tracer data as Experiment 4. However, here we employed a divergence regularization technique, rather than the Horn and Schunck method employed in Experiments 1 through 4. Though the overall output of Experiment 4 seems to be more accurate (see scatterplots), Typhoon seems to discern some more detailed features using this method.
Typhoon command: ./cutyphoon_server -port 60000 -nM 7 -nD 8 -nE 8 -r 2 -a 0.01



LES Data



This is a large eddy simulation (LES) using the NCAR-LES code on a domain of size \((Lx,Ly,Lz) = (5120,5120,2048)m\). The domain is periodic in the x and y directions and the boundary layer is initialized with a capping inversion at \(z = 1000m\). The flow is forced with a geostrophic wind of \(10 m/s\) and a surface heat flux of \(0.5 W/m^2\). The grid size for the computations is \(256\times 256\times 256\). The horizontal slice shown in the image is of a passive conserved scalar taken at a height of \(z = 20m\). The scalar concentration is set to \(1\) at the bottom surface.

The above video shows the movement of the tracer in the simulation. The two images shown reflect the data passed to Typhoon; they are 15 seconds apart to match the time interval of the REAL hardware.




Speed Analysis



Horizontal wind speed averaged from one hundred LES frames between those passed to Typhoon.




Differences between speeds in above images.





U Component Analysis




Wind u component averaged from one hundred LES frames between those passed to Typhoon.


``


Differences between u components in above images.





V Component Analysis




Wind v component averaged from one hundred LES frames between those passed to Typhoon.


``


Differences between v components in above images.





Spectra and Transfer Functions







Atmospheric Lidar Research Group | California State University, Chico