final weather

The Weather ProjectFinal Release Presentation: 

The Weather Project Final Release Presentation Yi Ru Selvakumaran Vadivelmurugan Ayman Abdel-Haleem

Project Goals: 

Project Goals Construct and visualize the wind field from Doppler radial velocity, Compute and visualize the reflectivity field from 3 Doppler radar sites, Detect meteorological phenomenon in Doppler data

Wind Velocity Visualization: 

Wind Velocity Visualization Computation of the vertical velocity component Variation of the data density in the Cartesian space Data points are close to each others Data points are widely separated

Wind Velocity Visualization: 

Wind Velocity Visualization Computation of the vertical velocity component Computing the vertical velocity on the curvilinear grid Smoothing along the radial Using the LS method with a neighborhood of size 7 x 7 x 7

Wind Velocity Visualization: 

Wind Velocity Visualization Computation of the vertical velocity component Computing the vertical velocity on the curvilinear grid Using the LS method with a neighborhood of size 7 x 7 x 7 Smoothing along the radial

Wind Velocity Visualization: 

Wind Velocity Visualization Computation of the vertical velocity component Computing the vertical velocity on the Cartesian grid Computation of the Cartesian grid based on weighted average Using the LS method with a neighborhood of size 7 x 7 x 7 Cartesian grid point Algorithm takes about 30 minutes to execute on the whole data set !

Wind Velocity Visualization: 

Wind Velocity Visualization Visualization using streamlines Southeastern section of the storm Northwestern section of the storm

Wind Velocity Visualization: 

Wind Velocity Visualization Streamlines algorithm Seed greed points near section of interest Trace streamline over the grid points using the following criteria: Closest grid point to wind vector direction Flow direction should not conflict with the average flow direction Flow direction at next end point should maintain a consistent flow Terminate short streamlines or those that leave the grid Western section of the storm

Composition from 3 sites: 

Composition from 3 sites Combining radar data from multiple sites Site3 Site2 d3 d1 d2 Site1 (a) sample computed using Interpolation among 3 sites (b) sample from the nearest site Site2 Site1 Site3 Site3

Time Aliasing: 

Time Aliasing For the given reference time, sample1(v1,t1), sample2(v2,t2), sample3(v3,t3) are from 3 consecutive radar sweeps located in the same place and t1<t2<t3. given a reference time tr, T1<Tr<T2, alpha = (Tr-T1)/(T2-T1); v = (1-alpha)*v1 + alpha*v2; if T2< Tr<T2, alpha = (Tr-T2)/(T2-T3); v = (1-alpha)*v2 + alpha*v3. sample1(v1) sample3(v3) t1 t2 t3 sample2(v2) Interpolation among 3 consecutive radar sweeps

Result: 

Result Also we created the animation for the reflectivity data for a whole day. The animation can be downloaded from http://web.ics.purdue.edu/~yru/courses/ECE595E/doppler_03122006_3sites.avi. Without interpolation With interpolation

De-aliasing Velocity: 

De-aliasing Velocity In case of distant storms, -absence of continuous data between the radar and the storm -true velocity exceeds unambiguous or Nyquist velocity. -This could give relative Root-Mean-Square errors in the mean vertical wind profile. -In extreme weather condition the data are with increased complexity. The reference checking might suffer from degradation when the horizontal shear is strong. In case of continuity checking, -the errors in the first range gate might propagate -along the radial and azimuthal directions. RVn = cλ /8 V=Vm + nVn

Overview of Michael D. Eilts and Steven D. Smith’s Algorithm: 

Overview of Michael D. Eilts and Steven D. Smith’s Algorithm If the Gate to Gate check passes, then an Azimuthal error check is made. If not, then a check on Gate-to-Average of 9 surrounding points check is made. If that is not passed then the corresponding velocity is removed. If there are no points then a local search for valid velocity is made. If it is found then a comparison with the velocity is made. If that passes then an azimuthal error check is made. The Radial error check is carried out. If it is the last gate then the dealiased radial is set. If the radial error check ever failed, then the old radial for comparison is used. If the same radial is used for 5 times, then a flag to radial for comparison is set to missing. If this is not the last radial then again the procedure is carried out.

De-aliased Velocity Dataset: 

De-aliased Velocity Dataset

Tornado Detection: 

Tornado Detection The Tornadic Vortex Signature (TVS) is used to determinewhether the given dataset is that of a tornado. If the shear between the velocity extrema exceeds a minimum shear criterion at 2 or more elevation angles, then a TVS is declared. 2.08491 2.04744 2.05766 2.04879 2.08491 2.05583 2.05766 2.04879 Number of exceeds: 8 Result: possible TVS

Future Work: 

Future Work Test the wind construction on an ideal data set like a hurricane, Study the possibility of integrating a more sophisticated algorithm in the visualization system, Improve the streamlines mechanism for a more smooth and continuous flow, Design an algorithm for automatic adjustment of the streamlines parameters according to flow, Implement stream volumes using PRIME, Solve ghosting problem when interpolating data from multiple sites, Find better interpolation techniques for Doppler data from multiple sites, Improve the tornado detection algorithm

Thank you: 

Thank you