Assimilation of Water Vapor Mixing Ratio Profiles into the WRFDA System with a New Forward Operator

Assimilation of humidity data from ground-based profilers has not yet been fully explored in the field of numerical weather prediction. Apart from radiosondes and radiometers, which provide bottom-top profile humidity scans, lidars provide datasets with high temporal and spatial resolution as well as high accuracy. A suitable forward operator for the direct assimilation of water vapor mixing ratio (WVMR), a primary variable in the prognostic equations, has not yet been developed in the Weather Research and Forecasting data assimilation (WRFDA) system. Here we present an exclusive forward operator for WVMR that has been developed and tested by the modification of the atmospheric infrared sounding retrieval (AIRSRET) operator. The WVMR and temperature from the lidars are assimilated into the WRFDA system with the 3 dimensional variational (3DVAR) DA system in rapid update cycle (RUC) mode. The forward operator ingests the profile information from the temperature rotational Raman lidar (TRRL) and WVMR from water vapor differential absorption lidar (WV-DIAL), both from the University of Hohenheim (UHOH).

The impact of assimilating temperature and WVMR observations from the lidars using the modified forward operator was investigated from the HD(CP)² Observation Prototype Experiment (HOPE) of the project High Definition of Clouds and Precipitation for advancing Climate Prediction (HD(CP)²). The WRF model was configured for the central Europe domain in a convection permitting scale of 2.5 km spatial resolution and 100 levels in the vertical with fine resolution in the PBL gradient region. Three assimilation experiments were conducted to evaluate the performance of the forward operator in a RUC mode with 1 hourly assimilations cycles. In the first experiment (CONV_DA), or the control run, only assimilation of the conventional observations were carried out. The second experiment (T_DA_TEMP) assimilated temperature data additionally with the conventional data where temperature is ingested with the conventional radiosonde forward operator (TEMP operator in WRF-DA). Finally the third experiment (QT_DA_AIRSRET) with the WVMR and temperature together in addition to the conventional dataset were assimilated with the modified AIRSRET operator. The WVMR RMSE of the QT_DA_AIRSRET showed a 20 percentage reduction compared to CONV_DA with respect to WV-DIAL observations. The T_DA_TEMP showed a 60 percentage decrease in temperature RMSE compared to CONV_DA with respect to TRRL observations. But the temperature RMSE of QT_DA_AIRSRET was almost same as that of CONV_DA. The reason for this behavior is the correlation between the temperature and WVMR variables in the background error covariance matrix of the 3DVAR which is static and not flow-dependent. The initial attempt to develop an exclusive forward operator for the WVMR gives a promising result for assimilating humidity observations directly into the WRFDA system.

The work has been extended in collaboration using the Raman lidar for meteorological observation (RALMO) from MeteoSwiss. The objective is to compare the performance of the WRFDA system using advanced DA strategies with the forward operator versus the COSMO-KENDA model adopted by the MeteoSwiss. The results from the HOPE campaign and initial results from the MeteoSwiss will be presented and discussed.