PGsonda Front/Back Calibration Report

Generated 2026-05-08 00:16 SAST

Introduction

This report compares the PGsonda front sensor (lionsheadlower) against the back sensor (signalhill) over the selected local time window from 2026-05-03 07:00:00+02:00 to 2026-05-03 17:00:00+02:00. Only periods that satisfy the chosen direction sector, speed threshold, and stability filters are accepted as candidate calibration windows.

Aim

Identify stable, unobstructed wind periods where the front and back sensors can be compared directly, then quantify the average direction offset needed for front-sensor calibration while treating the back sensor as the reference.

Inputs

Configuration Help

parameter value meaning
output_dir E:\Project_PGSONDA_CALIB\reports\front_back_calibration Folder where CSV files, plots, and report files are saved.
front_station lionsheadlower PGsonda logging name for the front sensor.
back_station signalhill PGsonda logging name for the back sensor.
front_correction -18.0 Direction correction applied as corrected = raw - correction for the front sensor.
back_correction 22.0 Direction correction applied as corrected = raw - correction for the back sensor.
direction_ranges 292.5 to 337.5 deg Allowed wind-direction sectors. Both sensors must lie inside these sectors to qualify.
min_speed_kph 10.0 Minimum wind speed that both sensors must exceed before a row can be used.
speed_column avg_speed_kph Which speed metric is filtered, usually avg_speed_kph or max_speed_kph.
stability_window 20min Rolling time window used to test directional stability.
max_direction_std_deg 12.0 Maximum circular standard deviation allowed inside the stability window.
min_window_minutes 15.0 Shortest accepted calibration window after filtering.
min_samples_in_window 4 Minimum number of paired rows required inside a valid window.
pair_tolerance_minutes 5 Maximum timestamp mismatch allowed when pairing front and back rows.
max_gap_minutes 10 Largest gap allowed between candidate rows before splitting into a new window.

Method

  1. Download PGsonda rows for the front and back stations and convert raw directions to corrected directions using corrected = raw - correction.
  2. Pair front and back rows within 5 minute(s) based on nearest timestamps.
  3. Keep only rows where both sensors fall inside these direction ranges: 292.5 to 337.5 deg.
  4. Require both sensors to exceed 10.0 kph using avg_speed_kph.
  5. Require directional stability over a 20min rolling window with maximum circular standard deviation of 12.0 deg.
  6. Split candidate rows into windows whenever the gap exceeds 10 minutes and only keep windows longer than 15.0 minutes with at least 4 paired rows.

Results

Candidate Windows

window_id start_local end_local duration_minutes samples front_mean_direction_deg back_mean_direction_deg front_minus_back_deg_from_means front_minus_back_deg_mean_rows front_direction_std_deg back_direction_std_deg front_avg_speed_kph back_avg_speed_kph front_max_speed_kph back_max_speed_kph avg_pair_gap_seconds
1 2026-05-03 12:15:58.500000+02:00 2026-05-03 16:59:20.500000+02:00 283.4 209 325.76 324.03 1.73 1.73 6.51 6.06 20.37 20.46 23.18 22.65 20.3
0 2026-05-03 07:58:17+02:00 2026-05-03 12:05:37+02:00 247.3 192 317.31 316.93 0.37 0.38 9.89 8.90 17.28 17.89 19.74 19.87 19.0

Overview plot

Overview plot

Selected window plot

Selected window plot

Sector difference plot

Sector difference plot

All candidate windows difference plot

All candidate windows difference plot

Conclusion

Best accepted window: 2026-05-03 12:15:58.500000+02:00 to 2026-05-03 16:59:20.500000+02:00 with a front-minus-back difference of 1.73 deg. Recommended front correction with back fixed: -16.90 deg.