8R.8 Customer profiled polar volume compositing using phenomena related data quality

Thursday, 27 October 2005: 5:15 PM
Alvarado D (Hotel Albuquerque at Old Town)
Harri Hohti, Remote Sensing for Weather Applications, Finnish Meteorological Institute, Helsinki, Finland; and M. Peura, T. Kuitunen, and J. Koistinen

Finnish Meteorological Institute has established a project to detect, measure and correct various error sources related to phenomena commonly known in radar meteorology. The goal of the project is to find quality measures which could be used like weighting factors in algorithms of a polar volume compositing system.

This kind of quality and compositing approach gives a possibility for automatically tailor-made, on-demand radar products for customers having wide scale of different needs for radar data and its quality indicators.

The list of actions due to various phenomena must be organized so that their order in processing chain is physically consistent:

1. 3D analysis of water phase distribution 2. Detection and classification of non-meteorological echoes by pattern recognition 3. Classification of precipitation type affecting Z/R relation, uses pattern recognition, fuzzy logic and neural computing 4. Hail detection for warnings and correction to Z/R relation 5. Rain attenuation using 3D water phase 6. VPR analysis and correction

Every analysis step produces information of effect on data values and/or on data reliability. Typically this information contains correction value proposal and the probability of phenomenon causing the need for correction. The scope of this information can be volume wide, scan wide, ray wide or even each bin.

These correction and quality related data fields are finally combined to fulfil specific needs of each customer and product. This is done by defining user and product profiles which determine the contribution of each phenomenon to each product. The profile can be e.g. a list of weight factors or a list of decision rules. Thus radar data quality is not a fixed constant attached to a bin of measured data (except in the case of completely useless data) but a vector of variable quality measures depending on the application and user requirement thresholds.

Some examples of configuration of the system for different products to specific customer groups like hydrologists, aviation meteorologists and farmers will be presented.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner