Improvement in cloud predictions using satellite data assimilation for real-time forecasting
Abstract
The accuracy of quantitative forecasting of low-level operational cloud products such as the cloud top height, cloud top pressure and cloud thickness is rather low. Reliable forecasting of the low-level clouds (cloud top altitudes below 2-3 km) such as fog, stratus or stratocumulus is essential for aviation safety purposes. With the advent of an increased number of spectral channels and high-resolution imagers on the Geostationary Operational Environmental Satellite, cloud products can be diagnostically extracted and, furthermore, these cloud products can be used to modify the initial conditions for numerical weather prediction. Although operational methods are relatively successful in determining the cloud top altitudes for deep clouds and high clouds (usually above 5 km), there is no unique way of inferring the cloud top heights for low-level clouds due to their optical properties and low-level inversions. An algorithm has been developed in this study to classify the low-level cloud types using the brightness temperatures extracted from the GOES satellite visible and infrared channels. Cloud top temperatures above 8° C characterize low-level clouds. The brightness temperature differences between the window channel (11 ìm) and the shortwave infrared channel (4 ìm) are used to segregate the optically thin and thick clouds, and the relative humidity obtained from the surface stations is used to distinguish the fog or clouds formed by fog lifting. The infrared satellite imagery on 29 June 2006 is considered for this study with domain coverage of 400 x 400 km2 . The ground-truth observations were obtained from the surface weather station located at the Naval Air Station, Fallon (NASF), Nevada. Upon classification of low-level clouds in the satellite imagery, (a) the first step is to compute the cloud base temperature in the low-level cloudy pixels using the surface temperature and cloud base height obtained from the ceilometer measurements (at NASF) following a dry adiabatic lapse rate; (b) the second step is to compute the cloud top height using cloud base temperature, and the satellite- derived cloud top temperature following the wet adiabatic lapse rate in the cloud layer; (c) the third step is to obtain a representative lapse rate for the computing domain; (d) the fourth step is to compute the cloud top heights for the individual satellite pixels in the entire domain. The information on cloud top height and cloud top temperature obtained from the cloudy pixels is then dynamically assimilated into the model analysis using Cressman's objective analysis. Using the improved model analyses, a deterministic forecast will be carried out with an option of four-dimensional data assimilation of model winds and thermodynamic variables for a pre- forecast period of one complete diurnal cycle. Verification will be carried out using the hourly surface observations and cloud base measurements, and also using the satellite cloud imagery against the simulated cloud imagery and associated cloud products. The data assimilation of the derived cloud products is being tested in modeling systems such as the Mesoscale Model 5 (MM5) and the Weather Research Forecasting Model (WRF). The data assimilation of cloud products and verification is intended for the pre-processing module in a real-time forecasting system using various objective analysis procedures such as the Cressman-type, multi-quadric and 3DVAR. This study is to develop an efficient forecasting system to support naval aircraft and rotorcraft operations at the Fallon Naval Air Station, Fallon, Nevada.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2006
- Bibcode:
- 2006AGUFM.A31A0842V
- Keywords:
-
- 3315 Data assimilation