| An error was recently discovered in the initial sequence step of creating ground validation rainfall products. Specifically, rain gauge range and azimuth values present in the 2A56 header files were incorrectly calculated from gauge latitude and longitude. |
| |
| The program get_radar_data_over_gauge, which extracts radar reflectivity data over the gauge locations to create intermediate radar/gauge merged files, determines the (ix, iy) extraction point by utilizing the range and azimuth present in the 2A56 gauge headers. Therefore, errors in the 2A56 range and azimuth values resulted in the extraction of reflectivity data over incorrect gauge locations. Subsequently, errors in the gauge/radar merged files extended into automated gauge quality control and bulk adjustment statistics. This directly affected the Z-R tables and resulting rainfall products (2A-53, 3A-53, and 3A-54). |
| |
| Following are the list of (ix, iy) gridpoint discrepencies between the incorrect (v1.4) and corrected (v1.5) extraction programs for all gauges at each GV site:
|
| |
| A new subroutine to calculate range and azimuth from the gauge latitude and longitude was implemented in all pertinent GVS and gauge software. New 2A56 gauge files (version 3) were created containing corrected header information. Using the corrected 2A56 files, numerous tests were conducted to verify that merged radar/gauge files contained reflectivity data extracted over the correct gauge locations. |
| |
Specific tests included:
- 2A56 range, azimuth, (ix, iy) comparison: New 2A56 (ix, iy) values from all GV sites match perfectly between 2A56 headers (where these values are now present) and values from two independently developed programs.
- Get_radar_data_over_gauge (ix, iy) comparison: New (ix,iy) values match perfectly with Amitai's independent code for locations over all gauges within 150 km at each site.
- First_intermediate file (1IF) comparison: For a test case (Melb 980714), a small, test 1IF was created with corrected software and compared to 1IF created with old software. Many expected differences were evident in center pixel radar reflectivity extraction between old vs. new 1IF's. Some dBZ differences in the 18-25 dBZ range.
- Range comparison 2A56 header vs. 1IF_v1.5 (new):
- MELB: ranges match for all KSC, STJ, and SFL gauges
- HSTN: ranges match for all HAR gauges
- KWAJ: ranges match for all KWA and RMI gauges
- DARW: ranges match for all CSC gauges
- Effect of corrected software on gauge QC (AQC) and bulk-adjustment statistics: created new merged files and ZR tables from corrected (v1.5) 1IF's and existing gauge databases for one test month for each GV site. R/G statistics and A coefficients, along with the number of gauges which passed AQC, changed slightly.
| Site/Month |
R/G(tot) |
R/G(strat) |
R/G(conv) |
A(tot) |
A(strat) |
A(conv) |
AQC |
| HSTN 9-98 v1.4 |
0.44 |
0.38 |
0.49 |
96 |
78 |
110 |
57
|
| HSTN 9-98 v1.5 |
0.44 |
0.42 |
0.45 |
94 |
90 |
98 |
62 |
|
| MELB 7-98 v1.4 |
1.01 |
0.97 |
1.02 |
306 |
289 |
310 |
36 |
| MELB 7-98 v1.5 |
0.99 |
1.22 |
0.95 |
294 |
395 |
277 |
42 |
|
| KWAJ 9-98 v1.4 |
0.20 |
0.14 |
0.25 |
31 |
20 |
42 |
2 |
| KWAJ 9-98 v1.5 |
0.19 |
0.16 |
0.21 |
29 |
23 |
34 |
2 |
|
| DARW 1-98 v1.4 |
0.64 |
0.63 |
0.65 |
161 |
156 |
165 |
11 |
DARW 1-98 v1.5 |
0.67 |
0.66 |
0.67 |
171 |
169 |
173 |
12 |
|
- Effect of corrected extraction software on 3A-54 rain maps: sample 3A-54 products were created using new ZR tables based on the previous test. New (v1.5) rainmaps spatially very similar to existing (v1.4) rainmaps. Monthly area-averaged rainfall values demonstrated a correction:
| Site/Month |
v1.4 (mm/day) |
v1.5 (mm/day) |
| HSTN 9-98 |
9.66 |
9.75 |
| MELB 7-98 |
4.15 |
4.25 |
| KWAJ 9-98 |
6.21 |
6.44 |
| DARW 1-98 |
10.74 |
10.28 |
|
- Intermediate file center pixel reflectivity vs. 2A-55 reflectivity over gauges: Focusing on one MELB volume scan of a heavy rain event, all gauges within 100 km of the radar demonstrate a PERFECT match between the 2IF center pixel reflectivity and the reflectivity values independently extracted over the gauges from the 1.5 km 2A-55 ascii slice file. All gauges located > 100 km from the radar demonstrate a PERFECT match between the 2IF center pixel reflectivity and the reflectivity values independently extracted over the gauges from the 3.0 km 2A-55 ascii slice file.
|