skip to main content

Explore Research Products in the NSF-PAR

Title: Spectranet: A High Resolution Imaging Radar Deep Neural Network for Autonomous Vehicles
The potentials of automotive radar for autonomous driving have not been fully exploited due to the difficulty of extracting targets' information from the radar signals and the lack of radar datasets. In this paper, a novel signal processing pipeline is proposed to address the max ambiguous velocity reduction issue introduced by staggered time division multiplexing (TDM) scheme of high resolution imaging radar system with a large number of transmit antennas. A dataset of 1,410 synchronized frames (stereo cameras, LiDAR, radar) with three classes, i.e., bus, car, and people, is constructed from field experiments. Next, we implement a vanilla SpectraNet and show its promising performance on moving object detection and classification with a mean average precision (mAP) of 81.9% at an intersection over union (IoU) of 0.5.  more » « less
Award ID(s):
2153386
NSF-PAR ID:
10394714
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
IEEE 12th Sensor Array and Multichannel Signal Processing Workshop (SAM)
Page Range / eLocation ID:
301 to 305
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ( , Dryad)
    This dataset contains over 14,000 hours of regional radar mosaics over the northeast US from 600+ winter storm days between 1996-2023. Winter storm days are defined when at least 2 out of 15 surface stations in the northeast US (see attached map) produced at least 1 inch of snow over the 24 hour period. Sequences of these mosaics aid in analyzing the precipitation area and the structures within winter storms. Radar reflectivity data is combined from the first, lowest (0.5 degree) elevation angle from 12 NEXRAD WSR-88D radars in the northeast US (see attached). The scans occur every 5-10 minutes from each radar depending on the radar scan settings.  The time label of the regional map is based on the scan time central radar, KOKX (Upton, NY). Scans from other radars in the region are used for that time as long as they are within 8 minutes of the KOKX scan. The polar radar data from each radar is interpolated to a regional 1202 km x 1202 km Cartesian grid with 2 km grid spacing covering 35.73-46.8 degN and 66.36-81.85 degW. Where the radar domains overlap, we take the highest reflectivity value. For dates after dual-polarization integration (2012 onwards), files contain the correlation coefficient (RHO_HV) field and a binary field  that can be used to “image mute” the reflectivity which reduces the visual prominence of melting and mixed precipitation commonly mistaken for heavy snow. Image muting is applied where radar reflectivity is ≥ 20 dBZ and RHO_HV is ≤ 0.97. This product is different from other widely used radar mosaics such as the MRMS produced by NOAA since it does not interpolate to a constant altitude and thus preserves the finer scale details in the reflectivity field. Because the data used to create these mosaics are not interpolated to a constant altitude, the altitude varies over the region (altitudes of radar scan used at each grid point are provided as a field for each data file). This data set is specifically designed to analyze fine-scale structures in winter storms. Part 1 contains files pre-dual polarization integration (1996-2012)Part 2 contains files post-dual polarization integration (2012-2023) 
    more » « less
  2. ; ; ; ; ( , Dryad)
    This dataset contains over 14,000 hours of regional radar mosaics over the northeast US from 600+ winter storm days between 1996-2023. Winter storm days are defined when at least 2 out of 15 surface stations in the northeast US (see attached map) produced at least 1 inch of snow over the 24 hour period. Sequences of these mosaics aid in analyzing the precipitation area and the structures within winter storms. Radar reflectivity data is combined from the first, lowest (0.5 degree) elevation angle from 12 NEXRAD WSR-88D radars in the northeast US (see attached). The scans occur every 5-10 minutes from each radar depending on the radar scan settings.  The time label of the regional map is based on the scan time central radar, KOKX (Upton, NY). Scans from other radars in the region are used for that time as long as they are within 8 minutes of the KOKX scan. The polar radar data from each radar is interpolated to a regional 1202 km x 1202 km Cartesian grid with 2 km grid spacing covering 35.73-46.8 degN and 66.36-81.85 degW. Where the radar domains overlap, we take the highest reflectivity value. For dates after dual-polarization integration (2012 onwards), files contain the correlation coefficient (RHO_HV) field and a binary field  that can be used to “image mute” the reflectivity which reduces the visual prominence of melting and mixed precipitation commonly mistaken for heavy snow. Image muting is applied where radar reflectivity is ≥ 20 dBZ and RHO_HV is ≤ 0.97. This product is different from other widely used radar mosaics such as the MRMS produced by NOAA since it does not interpolate to a constant altitude and thus preserves the finer scale details in the reflectivity field. Because the data used to create these mosaics are not interpolated to a constant altitude, the altitude varies over the region (altitudes of radar scan used at each grid point are provided as a field for each data file). This data set is specifically designed to analyze fine-scale structures in winter storms. Part 1 contains files pre-dual polarization integration (1996-2012)Part 2 contains files post-dual polarization integration (2012-2023) 
    more » « less
  3. ; ; ; ; ; ; ; ; ( , Journal of Applied Meteorology and Climatology)
    Abstract This study evaluates ice particle size distribution and aspect ratio φ Multi-Radar Multi-Sensor (MRMS) dual-polarization radar retrievals through a direct comparison with two legs of observational aircraft data obtained during a winter storm case from the Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) campaign. In situ cloud probes, satellite, and MRMS observations illustrate that the often-observed K dp and Z DR enhancement regions in the dendritic growth layer can either indicate a local number concentration increase of dry ice particles or the presence of ice particles mixed with a significant number of supercooled liquid droplets. Relative to in situ measurements, MRMS retrievals on average underestimated mean volume diameters by 50% and overestimated number concentrations by over 100%. IWC retrievals using Z DR and K dp within the dendritic growth layer were minimally biased relative to in situ calculations where retrievals yielded −2% median relative error for the entire aircraft leg. Incorporating φ retrievals decreased both the magnitude and spread of polarimetric retrievals below the dendritic growth layer. While φ radar retrievals suggest that observed dendritic growth layer particles were nonspherical (0.1 ≤ φ ≤ 0.2), in situ projected aspect ratios, idealized numerical simulations, and habit classifications from cloud probe images suggest that the population mean φ was generally much higher. Coordinated aircraft radar reflectivity with in situ observations suggests that the MRMS systematically underestimated reflectivity and could not resolve local peaks in mean volume diameter sizes. These results highlight the need to consider particle assumptions and radar limitations when performing retrievals. significance statement Developing snow is often detectable using weather radars. Meteorologists combine these radar measurements with mathematical equations to study how snow forms in order to determine how much snow will fall. This study evaluates current methods for estimating the total number and mass, sizes, and shapes of snowflakes from radar using images of individual snowflakes taken during two aircraft legs. Radar estimates of snowflake properties were most consistent with aircraft data inside regions with prominent radar signatures. However, radar estimates of snowflake shapes were not consistent with observed shapes estimated from the snowflake images. Although additional research is needed, these results bolster understanding of snow-growth physics and uncertainties between radar measurements and snow production that can improve future snowfall forecasting. 
    more » « less
  4. ; ; ; ; ( , Atmosphere)
    In this study, a polarimetric radar forward model operator was developed for the Weather Research and Forecasting (WRF) model that was based on a scattering algorithm using the T-matrix methodology. Three microphysics schemes—Thompson, Morrison 2-moment, and Milbrandt-Yau 2-moment—were supported in the operator. This radar forward operator used the microphysics, thermodynamic, and wind fields from WRF model forecasts to compute horizontal reflectivity, radial velocity, and polarimetric variables including differential reflectivity (ZDR) and specific differential phase (KDP) for S-band radar. A case study with severe convective storms was used to examine the accuracy of the radar operator. Output from the radar operator was compared to real radar observations from the Weather Surveillance Radar–1988 Doppler (WSR-88D) radar. The results showed that the radar forward operator generated realistic polarimetric signatures. The distribution of polarimetric variables agreed well with the hydrometer properties produced by different microphysics schemes. Similar to the observed polarimetric signatures, radar operator output showed ZDR and KDP columns from low-to-mid troposphere, reflecting the large amount of rain within strong updrafts. The Thompson scheme produced a better simulation for the hail storm with a ZDR hole to indicate the existence of graupel in the low troposphere. 
    more » « less
  5. ; ; ; ; ( , Atmosphere)
    In this study, a polarimetric radar forward model operator was developed for the Weather Research and Forecasting (WRF) model that was based on a scattering algorithm using the T-matrix methodology. Three microphysics schemes—Thompson, Morrison 2-moment, and Milbrandt-Yau 2-moment—were supported in the operator. This radar forward operator used the microphysics, thermodynamic, and wind fields from WRF model forecasts to compute horizontal reflectivity, radial velocity, and polarimetric variables including differential reflectivity (ZDR) and specific differential phase (KDP) for S-band radar. A case study with severe convective storms was used to examine the accuracy of the radar operator. Output from the radar operator was compared to real radar observations from the Weather Surveillance Radar–1988 Doppler (WSR-88D) radar. The results showed that the radar forward operator generated realistic polarimetric signatures. The distribution of polarimetric variables agreed well with the hydrometer properties produced by different microphysics schemes. Similar to the observed polarimetric signatures, radar operator output showed ZDR and KDP columns from low-to-mid troposphere, reflecting the large amount of rain within strong updrafts. The Thompson scheme produced a better simulation for the hail storm with a ZDR hole to indicate the existence of graupel in the low troposphere. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email