1 / 13

GOES-R AWG Product Validation Tool Development

Develop and enhance validation tools for the GOES-R Rainfall Rate algorithm, focusing on evaluating algorithm performance, identifying biases, and improving accuracy. Utilize ground validation datasets from TRMM and Nimrod radar, and analyze statistics, spatial plots, and joint distribution plots. Deep-dive validation tools include comparing calibration MW data, analyzing spatial biases, and examining predictor and target data. Consider GUI for regional plots and predictor fields.

hneal
Download Presentation

GOES-R AWG Product Validation Tool Development

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. GOES-R AWG Product Validation Tool Development Hydrology Application Team Bob Kuligowski (STAR)

  2. Outline • Products • Validation Strategies • Routine Validation Tools • “Deep-Dive” Validation Tools • Ideas for the Further Enhancement and Utility of Validation Tools • Summary

  3. Products • Rainfall Rate / QPE • 4-km resolution • every 15 min • FD equatorward of 60° and with a LZA<70°

  4. Validation Strategies: Overview • Twofold validation strategy: • Determine performance of the algorithm against spec • Identify systematic biases / weaknesses in the algorithm • Ground validation datasets: • TRMM Precipitation Radar (35ºS-35ºN) from NASA • Nimrod radar composite (western Europe) from the British Atmospheric Data Centre Sample TRMM rain rates for a 24-h period Sample Nimrod 3-h accumulation

  5. Validation Strategies • Routine validation tools: • Time series of accuracy and precision • Is the algorithm meeting spec on a consistent basis? • Are there any trends in performance that might need to be addressed even if the algorithm is still meeting spec at this time? • Spatial plots of rainfall rates vs. ground validation • Are the rainfall rate fields physically reasonable? • Do the rainfall rate fields compare reasonably well with ground truth? • Scatterplots vs. ground validation • Are there any anomalous features in the scatterplots that could indicate errors not revealed by the spatial plots?

  6. Validation Strategies • Deep-dive validation tools: • Comparing calibration MW data with ground truth • How much of the error is due to the calibration data rather than the calibration process? • Divide data by algorithm class and analyze • Are errors in the algorithm associated with a particular geographic region or cloud type? • Spatially distributed statistics • Does the algorithm display any spatial biases (e.g., latitudinal, land vs. ocean) that need to be addressed? • Analyze the rainfall rate equations for particular cases • Are there particular predictors or calibration equations associated with errors?

  7. Routine Validation Tools • Capabilities: • Match Rainfall Rate with ground data (TRMM and Nimrod radar pre-launch; GPM and Stage IV / MPE post-launch) • Compute accuracy, precision • Compute basic validation statistics (volume bias, correlation, and threshold-dependent POD, FAR, area bias, and HSS) • Create joint distribution files

  8. Routine Validation Tools • Use GrADS for all visualization: • Spatial plots of Rainfall Rate and ground-truth data • Plots of POD, FAR, area bias, and HSS vs. threshold • Rainfall Rate / ground-truth data joint distribution

  9. “Deep-Dive” Validation Tools • Capabilities: • Match Rainfall Rate with ground data (TRMM and Nimrod radar pre-launch; GPM and Stage IV / MPE post-launch) • Match calibration MW rainfall rates with ground data • Divide matched Rainfall Rate and ground data by algorithm class • Divide matched Rainfall Rate and ground data by location • Compute performance statistics by algorithm class and location • Extract rain/no rain and rate equations and distribution adjustment LUTs

  10. “Deep-Dive” Validation Tools • Use GrADS for all visualization: • Spatial plots and joint distribution plots of both calibration MW rain rates and of GOES-R Rainfall Rates vs. ground data • Spatial plots of performance statistics

  11. “Deep-Dive” Validation Tools • Use GrADS for all visualization: • 2-D plots of rain/no rain and rate values as a function of predictor values • Spatial plots of the training data (predictors & targets)

  12. Ideas for the Further Enhancementand Utility of Validation Tools • A handy tool would be a GUI that would allow the user to select a portion of a Rainfall Rate field and automatically create regional plots of: • Performance statistics vs. ground validation and available calibration data • Joint distribution • Predictor fields • Rain / no rain and rain rate equations and distribution adjustment LUTs • etc.

  13. Summary • The GOES-R Rainfall Rate algorithm will be validated against TRMM and Nimrod radar pre-launch and GPM and Stage IV / MPE post-launch • Validation will focus on evaluating performance and identifying areas for potential improvement • Routine validation will focus on the former using time series of statistics, spatial plots and joint distribution plots • Deep-dive validation will focus on the latter by examining the predictor and target data along with the calibration to determine the reasons for any anomalies in performance

More Related