Weather Surveillance Radar -- Low-level
This page has information abour how weather radar operates. This described current operation with WSR-88D. MPAR will not use the same system.
Level I (L1) Data: Physical Data #
Each individual pixel on a weather radar is telling the story. The radar isn’t sending just one pulse into a radical, but many. For each pulse sent out, we know how it was sent:
And when we listen for the response, we know the same details about the return signal.3
This process of sending a ping and listening for the response it repeated many times for every radical. Meanwhile4, we are also binning the results. We know the time difference, and the speed of light, so we can know the distance the signal traveled (this is not necessarily the true distance, because of phenomenon that can distort the path of the radar beam).
Now, for a given “pixel” on radar, we can normalize data based on the distance and difference from the transmit state, leaving us with distributions for each polarization of:
- return signal strengths
- return signal phases
Range Optimization #
To get the most out of the radar, it doesn’t actually do this all at once. This is because the different waveforms have tradeoffs with aliasing and range. In reality, , the “phase-based” and “strength-based” data are collected seperately.
- The strength (reflectivty) waveform is “Contiguous Surveillance” (CS), and use a low PRF (Pulse Repetition Frequency, ie, a longer time between pulses).
- The phase (velocity) waveform is “Contiguous Doppler” (CD), and uses a high PRF (short time between pulses).
- In the mid-levels, rather than performing 2 entire seperate sweeps, “Batch Mode” (“B”) interleaves a burst of ‘Surveillance’ pulses while performing the ‘Doppler’ scan. Some VCPs are phasing this out in favor of seperate scans, which can yield higher-resolution data.
Level II (L2) Data: Statistics #
Now, we can get some data out of this. Each item in this table is based on a type of averaging to get the final value. Items marked with * make use of both polarizations. The technology of using these polarizations is refered to as “Duel Pol”.
| Operation/Source | Strength | Phase |
|---|---|---|
| Base Moments | BR | BV5 |
| Statistical Distribution | CC* | SW5 |
| Polarization Diff | ZDR* | PHI* |
Heres what each is. The description can be thought of “how much stuff”
| Abbrev. | Name | Units | Description |
|---|---|---|---|
| BR | Reflectivity | dBZ | Strength of reflected signal |
| BV | Radical Velocity | kts | Speed strictly towards/away from the radar |
| CC | Correlation Coefficent | % | How similar the strengths of the return pulses in a bin were |
| SW | Spectrum Width | kts | How similar the velocities in the return bin were (horizontal pol only) |
| ZDR | Differential Reflectivity | dB | The ratio of the horizontal power return to vertical power return |
| PHI | Differential Phase | ° | The difference in phase between the horizontal and vertical pulses (cumulative) |
This is all of the “Level II” Radar products, which are available in high resolution. For each radical, every 250 meters is binned together (maximum range of 460km for reflectivity). Each radical is a 0.5 degree arc segment (azamidth) from the antenna. This is often refered to as “Super Resolution”, as it is the result of upgrades made when Duel Pol was introduced, at the cost of a slight increase in noise. This upgrade also extended the maximum range for from 230km to 300km for velocity-based products. Unlike Level I data, which is only disseminated in realtime internally at the NWS, this data is available for free to the public in near-realtime. As the radar rotates, it collects this data for each radical, then visits another elevation (it’s tilt angle) and repeats the process. The exact elevations the radar visits depend on the site. How frequently and in what order elevations are scanned at are defined by the Volume Coverage Pattern (VCP) and it’s accompanying parameters, all of which are switch based on weather conditions.
-
My understanding is that this is essentially a constant. ↩︎ ↩︎
-
My understanding is that this polarization is always horizontal or vertical, and there is no measurement of a “twist”. It may be the case that the sent signal isn’t polarized, and only the receiver ‘cares’, but I’m unsure. ↩︎
-
You can find all the details about how L1 data files are structued in the Level I ICD (Doc #2620076) ↩︎
-
I’m not sure the exact ordering of this process. ↩︎
-
Velocity is actually found from the shift in the phase. ↩︎ ↩︎