Cloud and boundary layer forecast evaluations
Abstract
Project Summary #Approved for Public Release#The Navy runs several numerical weather models to aid their operations and has a vested interest in making these predictions as skillful as possible. The methodologies used to assess forecast skill in essence constrainthat skill since model refinements are engineered to improve the chosen forecast skill metrics. The proposed model evaluation work takes an adequacy for purpose perspective. In this view, model evaluation seeks to learn whether a model is adequate or fit for a particular purpose which contrasts with assessing model quality solely based on how accurately and completely a model represents a target. The proposed work will A) address the adequacy for purpose of reanalysis, the current model evaluation standard, by location and weather condition and B) prototype new methods to assess forecasts of cloud characteristics and improve understanding of low clouddissipation. Both portions of the project will be done in close cooperationwith scientists at NRL-Monterey Marine Meteorology Division.Reanalysis blends forecast model output with assimilated observations and is currently being used by NRL and others in the atmospheric science community as a substitute for observations to assess model prediction skill. The assimilation details have a key weakness in that observations that are in larger disagreement with first guess model output fields are weighted less than those that arein closer agreement. As a result of this #perverse feedback#, reanalysis fields are likely of lower quality compared to observations in the very situations where the forecasts have the most error. Our goal is to document the relative quality of several reanalysisoutputs (ERA5, MERRA, and NCEP CFS-R) versus actual observations as a function of location and weather conditions. We will utilize weather observations from airports around the world as well as information from buoys and focus on locations of higher relevance to the Navy including coastal, island, and open ocean locations. Key outcomes will include detailed information on the variables, locations, and weather situations where reanalysis works well and can be used as a reasonable substitute for observations and where it has poor performance and should not be used as part of model skill assessment and tuning.The expected locations, durations, and vertical extent of cloud fields are relevant for several types of naval operations. Current work at NRL has focused on evaluation of modelcloud fields compared to #top-down# observations from satellites in combination with field project data sets from research aircraftand #bottom-up# observations from ground-based sensors. Satellite measurements do not provide direct observations of cloud base norinformation on the advection and dissipation of low-level clouds that are obscured by higher level cloud layers. We will use several profiling lidar ceilometers as well as shortwave and longwave radiation measurements at Plymouth, NC near the Atlantic coast to compare observed cloud characteristics with cloud fields from reanalysis and forecasts. Ground-based data at <10 min temporal samplingin combination with the hourly satellite updates will help to address processes and forecasts of low cloud dissipation. As part of this work, we would design and deploy a new instrument, an IR all sky camera that would provide information on the bottom edges of clouds over the site both day and night. If this initial prototyping work looks promising it could lay the groundwork for a multi-year cloud monitoring site that would provide a large sample size for cloud forecast evaluation combining satellite (top-down) and ground-based (bottom up) measurements.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 15, 2024
- Source ID
- N000142412216
Entities
People
- Sandra E. Yuter
Organizations
- North Carolina State University
- Office of Naval Research
- United States Navy