Title: Robin Hogan and Jon Shonk
1Implementation of multiple regions in
Edwards-Slingo
- Robin Hogan and Jon Shonk
2Shadows!
Natural logarithm of IWC (g m-3)
- Need separate regions even in clear skies?
Downward short-wave flux (W m-2)
Upward short-wave flux (W m-2)
3Anomalous horizontal transport
- Homogenization of clear-sky fluxes
- Reflected radiation has more chance to be
absorbed -gt TOA shortwave bias - Effect is very small in the longwave
- This problem can be solved in a way that makes
the code more efficient
Single cloud
4Two-stream scheme
TOA flux STOA-
F0.5
F0.5-
Layer 1 reflection, transmission and emission
R1, T1, S1, S1-
F1.5
F1.5-
- Consider a two-layer atmosphere
- Solve a tridiagonal matrix problem to obtain the
fluxes
Layer 2 reflection, transmission and emission
R2, T2, S2, S2-
F2.5
F2.5-
Surface emission and albedo Ss, as
5Edwards-Slingo solution
- It is conceptually convenient to solve the system
by - Working up from the surface calculating the
albedo ai and upward emission Gi of the whole
atmosphere below half-level i. - Then working down from TOA, calculating the
upwelling and downwelling fluxes from ai and Gi.
6Two-regions cloud and clear-sky
a
b
- With 2 regions (a b), matrix is denser
- Latest Edwards-Slingo only has approximate
solvers SOLVER_MIX_DIRECT and SOLVER_TRIPLE for
the 3-region version - OK for upright convection but unacceptable errors
introduced with realistic overlap
Layer 1
a
b
Layer 2
Note that the overlap coefficients have been
omitted in this example
7New version
a
b
- But some elements represent unwanted anomalous
horizontal photon transport - Remove them and the problem can be solved
exactly, and in closer agreement with ICA - Enables triple (and quadruple, quintuple)
regions to be implemented quite easily!
Layer 1
Rab
a
b
Layer 2
Rab is the reflection from region a to region b
at the same level
8Performance of new solver
- In calculating upwelling flux at half-level 1.5,
downwelling flux sees albedo of whole atmosphere
below, a1.5, which it reflects back into the same
region - Much closer to the independent column
approximation!
0.5
1.5
a1.5