Ashok et al. (2025) Impact of improved initial condition in a medium range forecast of super cyclonic storm Kyarr over Arabian sea using high resolution WRF and its data assimilation technique

Identification

• Journal: Theoretical and Applied Climatology
• Year: 2025
• Date: 2025-11-13
• Authors: Rohini Ashok, Kuvar Satya Singh
• DOI: 10.1007/s00704-025-05871-8

Research Groups

• Department of Mathematics, School of Advanced Sciences (SAS), Vellore Institute of Technology, Vellore, Tamil Nadu, India
• Centre for Disaster Mitigation and Management (CDMM), Vellore Institute of Technology, Vellore, Tamil Nadu, India

Short Summary

This study investigates the impact of assimilating conventional and non-conventional observations on improving the initial conditions and medium-range forecasts of Super Cyclonic Storm (Su-CS) Kyarr over the Arabian Sea using the WRF-3DVAR technique. The results demonstrate that data assimilation significantly enhances the accuracy of track, intensity, and rainfall predictions, with a cold start assimilation mode proving most beneficial.

Objective

• To evaluate the impact of data assimilation, incorporating both conventional and non-conventional observations, on improving the initial condition and subsequent medium-range (beyond 7 days) forecast of Super Cyclonic Storm (Su-CS) Kyarr's structure and intensity over the Arabian Sea using the WRF-3DVAR technique.

Study Configuration

• Spatial Scale: Double-nested domain with an outer domain (D1) of 15 km horizontal resolution (451 × 377 grid points) and an inner moving nested domain (D2) of 3 km horizontal resolution (221 × 221 grid points), both with 51 vertical levels.
• Temporal Scale: Medium-range forecasts extending beyond 7 days (up to 8-day simulations). Simulations were initialized at 00 UTC on 24, 25, 26 October and 12 UTC on 25 October 2019, and performed up to 18 UTC on 02 November 2019. Background Error Statistics (BES) were calculated for 2 months (01 October 2019 to 30 November 2019). Cyclic assimilation experiments included 1-cycle (every 6 hours) and 2-cycle (every 12 hours).

Methodology and Data

• Models used:
  • Advanced Research version of WRF (ARW; V4.0) model.
  • Three-dimensional Variational (3DVAR) data assimilation system.
  • Physical parameterization schemes: Kain-Fritsch (KF) Cumulus, Yonsei University (YSU) Planetary Boundary Layer, Goddard microphysics, Rapid Radiative Transfer Model (RRTM) long-wave, Dudhia short-wave radiation, and Noah Land Surface Model (Noah LSM).
  • ISFTCFLX option 1 for air-sea flux algorithms.
• Data sources:
  • Initial and Boundary Conditions: National Centers for Environmental Prediction (NCEP) Final analysis datasets (1° × 1° horizontal resolution), updated every 6 hours.
  • Assimilated Observations: NCEP PREPBUFR and satellite radiances from Advanced Technology Microwave Sounder (ATMS), Advanced Microwave Sounding Unit A (AMSU-A), and Microwave Humidity Sounding (MHS).
  • Validation Data: India Meteorological Department (IMD) best-fit track data for track and intensity; Cooperative Institute for Research on Atmosphere (CIRA) multi-platform satellite data for temperature anomalies; IMERG (Integrated Multi-satellite Retrievals for GPM) for rainfall observations.

Main Results

• 3DVAR data assimilation significantly improved the accuracy of the analysis and initial conditions.
• The mean track error for an 8-day simulation was reduced from 298.4 km in the control run (CNTL) to 198.1 km in the data assimilation (DA) experiment. Track errors for days 1-7 improved by 12.6% to 55% in the DA experiment compared to CNTL.
• Predicted intensity, in terms of Maximum Surface Wind (MSW) and Minimum Central Pressure (MCP), was slightly better in the DA experiment due to improved initial conditions. Kyarr reached a maximum intensity of approximately 66.7 m/s (240 km/h) with a minimum central pressure of 922 hPa.
• Accumulated rainfall from 00 UTC on October 25 to 00 UTC on November 2 showed a Root Mean Square Error (RMSE) of approximately 96.5 mm and a higher Correlation Coefficient (CC) of about 0.7 in the DA2400 experiment, compared to CNTL2400 (RMSE 127.8 mm, CC 0.46). The DA experiment also exhibited higher Probability of Detection (POD) and Post Agreement (PAG) values, and lower False Alarm Ratio (FAR) values for rainfall.
• The storm structure, including temperature anomaly, reflectivity, and horizontal wind (e.g., peak horizontal winds of 43.7 m/s in DA2400 vs. 33.4 m/s in DA2500), was well enhanced by the DA experiment, particularly DA2400, showing a more defined and vertically consistent warm-core structure.
• Simulations comparing cold start and cyclic mode assimilation for Su-CS Kyarr suggested that the cold start mode (DA-00h) was more beneficial for capturing the cyclone track and intensity, outperforming 6-hour and 12-hour cyclic assimilation.
• The best-performing cold start mode (DA-00h) was tested on other cyclones (ESCS Maha2019, ESCS Chapala2015, VSCS Luban2018, and VSCS Phet2010), yielding mean track errors from 43 km (Day 1) to 197 km (Day 5) and MSW errors from 10 m/s (Day 1) to 14 m/s (Day 5).

Contributions

• This study provides a comprehensive evaluation of the impact of data assimilation on medium-range (beyond 7 days) forecasts of a super cyclonic storm (Kyarr) over the Arabian Sea, addressing a gap in literature often focused on shorter forecast ranges.
• It demonstrates the significant improvements in track, intensity, and rainfall forecasts achieved by assimilating both conventional and non-conventional observations using the WRF-3DVAR system with regional background error statistics and bias correction.
• The research offers insights into the comparative performance of cold start versus cyclic data assimilation modes for tropical cyclone prediction, identifying the cold start mode as more effective for this specific case.
• The findings are validated against IMD best-fit track data, CIRA satellite observations, and GPM rainfall estimates, providing robust evidence for the enhanced forecast skill.
• The study extends its evaluation to multiple other cyclones, reinforcing the general applicability and benefits of the improved initial conditions.

Funding

• DST-ANRF (Project file no. CRG/2022/006068)

Citation

@article{Ashok2025Impact,
  author = {Ashok, Rohini and Singh, Kuvar Satya},
  title = {Impact of improved initial condition in a medium range forecast of super cyclonic storm Kyarr over Arabian sea using high resolution WRF and its data assimilation technique},
  journal = {Theoretical and Applied Climatology},
  year = {2025},
  doi = {10.1007/s00704-025-05871-8},
  url = {https://doi.org/10.1007/s00704-025-05871-8}
}