5x4to4x3.TXT 2006/02/24 This document describes the significant changes made to the Atmosphere- Ocean Model from its 1998 5x4 version to its 2004 4x3 version. Atmosphere ---------- The recent paper "Atmospheric C-grid Scheme with Variable Number of Layers" by Gary L. Russell [Monthly Weather Review, 2006] describes in more detail several of the changes that were implemented in the 2004 version of the 4x3 Atmosphere-Ocean Model. Use of the linear upstream scheme for linear advection of scalar quantities and improved Coriolis and metric terms were implemented in the 1998 5x4 Model and continued in the 2004 4x3 Model. 1. Horizontal resolution is increased from 5x4 (longitude x latitude) to 4x3. 2. Vertical resolution is increased from 9 layers (with a pure sigma coordinate) to 12 layers (with a combination of constant mass and sigma coordinate layering). 3. Arakawa and Lamb's [1976] second order C-grid scheme for momentum advection is replaced by the simplified advection scheme in which half of the corner fluxes in momentum advection are eliminated and the mass stencil is just two cells. 4. Distances used by the the pressure gradient force were improved. 5. A single horizontal velocity vector is defined at each pole for each layer. The 5x4 Model occasionally diverged at the poles; the 4x3 Model does not. Ocean ----- In both the 1998 5x4 Model and the 2004 4x3 Model, first layer is approximately 12 m thick, variable number of layers is used, bottom topography is adjusted to match layer edges, and simplified advection scheme is used for momentum. 1. Horizontal resolution is increased from 5x4 (longitude x latitude) to 4x3. 2. Vertical resolution is increased from 13 layers (each layer is 1.5 times thicker than the layer above it) to 16 layers (layers are 1.375 times thicker). 3. Distances used by the the pressure gradient force were improved. 4. A single horizontal velocity vector is defined at each pole for each layer. Sea Ice ------- 1. Advection of sea ice cover obeys the linear upstream scheme more rigorously in the 2004 4x3 Model than it did in the 1998 5x4 Model. 2. Acceleration of sea ice velocity is simplified and cleaner. Parallel side stresses of sea ice reduces sea ice shears. Ground ------ 1. Abramopoulos' ground hydrology scheme is replaced with a simpler 4-layer scheme. The ground fraction of a grid cell is subdivided into snow-free and snow-covered parts. Surface radiation responds separately to these partitions. 2. The complex implicit time shceme applied to the surface fluxes over ground are further complicated by taking into consideration the availability of water to evaporate and vapor to condense. Lakes ----- Both the 1998 5x4 Model and the 2004 4x3 Model have a fixed horizontal lake fraction within each grid cell; the ice temperature uses coding nearly identical to sea ice with 2 mass layers and 4 thermal layers. 1. In the 5x4 Model, lake temperature, lake ice cover, and lake ice thickness were specified with a seasonal cycle, but no year to year changes. In the 4x3 Model, lake temperatures and ice are predicted. Liquid lake water has a variable thickness mixed layer of constant temperature and an optional second layer whose vertical temperature structure fits a cubic polynomial. Radiation --------- 1. In the 1998 5x4 Model, non-cloud aerosols were functionally determined from surface type, latitude, and specified vertical profiles. The 2004 4x3 Model uses specified input files of many different non-cloud aerosols.