TOMO4D Operator for Assimilation of GNSS Tomography‐Derived Water Vapor Fields Into the WRFDA 4DVAR System to Improve Regional Rainfall Forecasting

Abstract Accurate representation of atmospheric water vapor is crucial for improving numerical weather prediction, particularly over regions with complex topography and sparse observation networks. Although assimilation of Global Navigation Satellite System (GNSS)‐derived integrated products such as zenith total delay or precipitable water vapor can improve humidity analyses, these approaches are limited by their lack of vertical resolution. This study introduces TOMO4D, a new four‐dimensional (4D) observation operator developed to assimilate GNSS tomography‐derived voxel‐based wet refractivity ( N w ) fields directly into the WRFDA four‐dimensional variational (4DVAR) system. Performance is evaluated for two heavy rainfall events over northern and northwestern Iran (23–24 October 2022). Three experiments are conducted: CTRL (no assimilation), TOMO3DVAR (3DVAR tomography assimilation), and TOMO4D (4DVAR tomography assimilation). Radiosonde (RS) validation at Tehran and Tabriz shows that TOMO4D improves the thermodynamic structure of the troposphere, reducing relative humidity RMSE by up to 38% and temperature RMSE by 9%–11% compared to CTRL, while decreasing temperature bias to within ±0.3K. Verification against 16 SYNOP stations further indicates that TOMO4D reduces accumulated precipitation RMSE by ∼17% and increases correlation by 35.3%. TOMO3DVAR results consistently fall between CTRL and TOMO4D, confirming that voxel‐based tomography improves moisture initialization even in 3DVAR, while the additional TOMO4D gains arise from time‐consistent assimilation within the 4DVAR window. Overall, GNSS tomography assimilation provides a promising pathway for improving humidity analyses and short‐range precipitation forecasting in data‐sparse regions such as Iran.