Simulated Weather Imagery: a Fast 3D Visible Light Radiative Transfer Package
Abstract
High spatial and temporal resolution images of the Earth environment reflecting cloud, precipitating hydrometeor, aerosol, and lower boundary surface conditions abound. Yet the bulk of the information in such observations isn't yet exploited for the initialization of Numerical Weather Prediction (NWP) model forecasts. This is because most information contained in the imagery data is inaccessible via the operationally used 2D radiative transfer models while their 3D versions are computationally prohibitively expensive.
Here we introduce a newly developed fast 3-D visible light radiative transfer package called Simulated Weather Imagery (SWIm). SWIm produces physically and visually realistic displays of sky and landscape conditions from any vantage point in or above the atmosphere, based on digital analysis or forecast fields from NWP systems. The resulting RGB images can be used for subjective visual interpretation, while radiance and reflectance values at various visible light wavelengths permit objective comparison of image data (from ground-, air- or space-borne observations) with simulated conditions (from NWP analysis or forecast fields). Since the propagation of light is influenced by the medium in which it travels, through inverse methods, a comparison of observed and SWIm generated imagery can be used to quantitatively assess and analyze conditions associated with many types of atmospheric and surface weather phenomena, such as convective development, stratiform clouds, smoke from fires, dust storms, aerosol pollution, and snow cover. The addition of visible imagery observations and SWIm (as a new forward operator), along with other existing types of satellite and in-situ observations can thus impart critical details yet missing from current cloud, precipitation, aerosol (e.g. dust, smoke) and surface analyses. In a proof of concept study, simultaneous minimization is used in a variational tomographic 4D scheme to find analysis states that best fit camera and satellite images as well as conventional weather observations. The resulting model initial fields are anticipated to improve forecasts on a variety of time scales.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.A23I2991A
- Keywords:
-
- 3315 Data assimilation;
- ATMOSPHERIC PROCESSESDE: 3336 Numerical approximations and analyses;
- ATMOSPHERIC PROCESSESDE: 3372 Tropical cyclones;
- ATMOSPHERIC PROCESSESDE: 0520 Data analysis: algorithms and implementation;
- COMPUTATIONAL GEOPHYSICS