Correlated Observation Error of GPM DPR Spectral Latent Heating Depending on Precipitation Type in Data Assimilation System
Abstract
The observation error covariance matrix is often approximated with a diagonal matrix when assimilating observation data. However, observations about the vertical distribution of precipitation such as space-borne radars have an observation error correlation which cannot be ignored in the vertical. In addition, the structure of the correlation matrix depends on the environment of precipitation such as deep convection and stable stratification.
To incorporate this correlation and dependency, we directly calculated the Kalman gain including the correlated observation error using the Moore-Penrose inverse matrix for each precipitation type, and investigated the impact of full or diagonal observation error covariance matrix in a data assimilation system. In this study, we investigated GPM DPR Spectral Latent Heating (GPM-SLH). The assimilation method of GPM-SLH is a nudging using the Kalman gain, and the data assimilation creates initial conditions for the Local Forecast Model which is a high-resolution model for short range precipitation forecasts and aviation weather forecasts at JMA. As a result of the assimilation experiments, we found that the observation error covariance matrix of SLH has a characteristic structure depending on the precipitation type, and plays an important role in the assimilation of dense observation data without vertical thinning. In the presentation, we will show the characteristics of the SLH observation error covariance matrix and demonstrate the impact on assimilation and prediction.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.A24E..06I
- Keywords:
-
- 3315 Data assimilation;
- ATMOSPHERIC PROCESSES;
- 3360 Remote sensing;
- ATMOSPHERIC PROCESSES;
- 0520 Data analysis: algorithms and implementation;
- COMPUTATIONAL GEOPHYSICS;
- 1855 Remote sensing;
- HYDROLOGY