• Intercomparison between IMD ground radar and TRMM PR observations using alignment methodology and artificial neural network

    • Fulltext


        Click here to view fulltext PDF

      Permanent link:

    • Keywords


      Ground radar; space radar; alignment methodology; spaceborne; artificial neural network.

    • Abstract


      An inter-comparison of ground radar reflectivity with space-borne TRMM’s Precipitation Radar using alignment methodology has been presented. For this purpose, reflectivity data from Dual Polarization Ground Radar (GR) maintained by the India Meteorological Department (IMD) at the IMD Delhi site is utilized. IMD Delhi has collected radar data during Continental Tropical Convergence Zone (CTCZ) programme from 2011 to 2013. The present study utilizes monsoon data collected during 4 months, namely, June, July, August, and September (JJAS) from the year 2013. The GR observables are first converted from polar coordinates to Cartesian coordinates and then spatially aligned with TRMM PR data at a near-real-time to a common volume. It was found that in all the overpass cases, IMD’s GR reflectivity has a positive bias when compared with TRMM PR. A methodology is proposed to ‘correct’ the GR reflectivity data by considering TRMM PR data as ‘truth’ using a neural network-based approach. A supervised learning algorithm based on the back-propagation neural network is used for this purpose. Ground radar reflectivity is fed as input to the network, while the TRMM PR reflectivity is the target. The trained network is then tested for its performance against data which is not used as part of the training process. The present methodology demonstrates the match up of uncalibrated ground radar measured reflectivity and a well-calibrated space-borne radar.


      $\bullet$ IMD’s ground radar data from CTCZ campaign during the monsoon of 2013 is utilized for Intercomparison study with TRMM PR observations.

      $\bullet$ IMD’s ground radar and TRMM PR reflectivity observations are spatially aligned within minimum volume required to produce spatially coincident sample.

      $\bullet$ Non-parametric based approach using ANN is used to reduce the difference between the two instruments.

      $\bullet$ With the ANN training, the correlation coefficient (RMSE) between the observations made by the two instruments increased (decreased) from 0.45 (15.77 dBZ) to 0.79 (4.69 dBZ).

    • Author Affiliations



      1. School of Mechanical Sciences, Indian Institute of Technology Bhubaneswar, Argul, Odisha, India.
    • Dates

  • Journal of Earth System Science | News

    • Editorial Note on Continuous Article Publication

      Posted on July 25, 2019

      Click here for Editorial Note on CAP Mode

© 2021-2022 Indian Academy of Sciences, Bengaluru.