Statistical Downscaling of Meteorological Forecasts
In this project, I temporally and spatially downscaled forecasts for five different meteorological metrics for use in a reservoir hydrodynamics forecasting project. I temporally downscaled the 21 climate model ensemble member forecasts from 6-hr to hourly resolution and spatially downscaled to our specific site, the Falling Creek Reservoir in Vinton, VA. Standard error in the statistical model fits was recorded and used to generate ensemble members with random noise proportional to model error. This allowed meteorological uncertainty to be propagated through further steps in the hydrodynamic forecasting project.
This work contributed to the publication, A Near-Term Iterative Forecasting System Successfully Predicts Reservoir Hydrodynamics and Partitions Uncertainty in Real Time.
 |
|---|
| Observed and 1-day Downscaled Forecasts over a 14-day fall period |