Premium
A Forecast Evaluation of Planetary Boundary Layer Height Over the Ocean
Author(s) -
Lavers David A.,
Beljaars Anton,
Richardson David S.,
Rodwell Mark J.,
Pappenberger Florian
Publication year - 2019
Publication title -
journal of geophysical research: atmospheres
Language(s) - English
Resource type - Journals
eISSN - 2169-8996
pISSN - 2169-897X
DOI - 10.1029/2019jd030454
Subject(s) - forecast skill , quantitative precipitation forecast , global forecast system , meteorology , environmental science , data assimilation , planetary boundary layer , dropsonde , climatology , middle latitudes , geopotential height , depth sounding , boundary layer , numerical weather prediction , atmospheric sciences , precipitation , geology , geography , physics , tropical cyclone , oceanography , turbulence , thermodynamics
The planetary boundary layer (PBL) is the layer closest to the Earth's surface in which heat, moisture, and momentum fluxes occur between the surface and free atmosphere. As these fluxes affect the atmospheric flow, it is important for weather forecast systems to accurately characterize them to provide skillful forecasts. A key PBL property is its height, and this study assesses the PBL height forecast skill over the oceans in the European Centre for Medium‐Range Weather Forecasts (ECMWF) Integrated Forecasting System (IFS). Using the ECMWF IFS and 1,959 dropsonde profiles from multiple flight missions, the average observed PBL height was 744.2 m and the short‐range (3–15 hr) forecast root‐mean‐square error was 242.5 m; the data assimilation step reduced the analysis root‐mean‐square error to 180.4 m. These data were also analyzed based on their location, the observed 925‐hPa wind speed, and the PBL stability, and results showed that the IFS had larger errors in the midlatitudes, under weaker winds on forecast days 1–4, and in unstable PBL conditions. The medium‐range forecasts are also underdispersive, which is considered to be largely due to representativeness errors, whereby the forecast model attempts to represent the grid box average rather than subgrid atmospheric variability. These results highlight an area of forecast uncertainty in current forecasting systems.