Status of the BRAMS project & plans for the future Hervé Lamy & BRAMS team Royal Belgian Institute for Space Aeronomy METRO annual meeting 2016 Brussels – 29 November 2016

Meteor forward scatter

BRAMS network 2 new stations in 2016 : Haacht and Sivry-Rance f = 49.97 MHz P=150 Watts Circularly polarized 32 stations 2 stopped 2 inactive

Typical receiving station AGC off RG213 Spectrum Lab

Typical receiving station

BRAMS data BEOTTI – 01/02/16 – 12H40 Raw WAV data 5 minutes Fs = 5512 Hz Spectrogram : 0 to Fs/2 =2756 Hz nFFT=16384 ; overlap=90%

BRAMS data Filtered raw data between for frequencies > 1200 Hz and < 1000 Hz f  0.33 Hz t  3 sec

BRAMS data BEUCCL – 04/01/16 – 04H55 Quadrantids – More noisy station

BRAMS data

Future plans for hardware 2-3 more stations per year (see strategy during talk on trajectories) Use SDR-like receiver (e.g. Dongle) instead of ICOM-R75 + external sound card. Also use single-board computer (e.g. Raspberry-Pi) instead of local PCs. Internship of a student from EPHEC in early 2017. Goal is to increase reliability, control, decrease cost & keep stability & sensitivity. Answer an important question about using Yagi antenna vertical vs inclined

Antenna vertical vs tilted BEOVER BEMAAS BEGENK

9 meteors Local effects only? Or also influence of the direction of the antenna? 14 meteors 6 meteors

What is at stake? 2 questions : 1) are we detecting the same meteors? 2) can we accurately determine the radiation pattern of an inclined antenna?

Tests in BEUCCL Sensitivity too low for tilted antenna due to use of a set of connectors and/or oxydation

Automatic detection of meteor echoes in BRAMS data

Method using the time signal See Roelandts (2014)

Method using the time signal Band-pass filtered signal to remove noise / possible parasitic signals

Method using the time signal Ratio of energy content in a short window (101 points  101/5512  0.018 sec) and in a large window (30001 points  30001/5512  5.44 sec) Indicator signal 3 parameters : Short window length Large window length Threshold

Test of the method : manual counts Fichier csv with coordinates of all the rectangles

Comparison manual – automatic counts Detection: everything detected by TR/automatic method Manual: everything manually counted (lines of the CSV files) TRUE POSITIVE: TR method detects something which falls into a rectangle FALSE POSITIVE: TR method detects something else than a meteor echo FALSE NEGATIVE: TR method misses a meteor that was manually counted    

Variation of the threshold BEUCCL : 0H00 – 0H55

Variation of the threshold BEUCCL : hour by hour Goal : check if threshold varies during the day

Variation of the threshold BEUCCL – Whole day

Choice of the threshold Varies from station to station Does not seem to vary significantly during one day Difficult to define a simple criterion to select it, so mostly chosen empirically so far.

Interferences Visual inspection of the FP reveals that 30-40 % of them are due to broad-band interference Some may be weak and barely visible Easy to remove a posteriori by filtering the signal inside the 200 Hz range where meteor / airplane echoes occur. Remaining peaks are interferences

Interferences

Origin of some FP ?

Gaussian noise with mean=0.1 and std=0.1 Simulations of IS Gaussian noise with mean=0.1 and std=0.1

Simulations of IS Noise + beacon

Simulations of IS

Future plans Finalise the tests with TR’s method : comparison with manually counted data from several stations, data from the Quadrantids (HL + RMZ) & Perseids (RMZ). Continue simulations of IS with 2 goals : 1) automatically select a range of threshold to search for the optimal value, 2) try to understand some of the false detections Set up this automatic method and apply it to archived data from January/February 2017 in order to produce raw counts per day & per station Continue to investigate other methods