S. Zhang et al., Use of differentiable and nondifferentiable optimization algorithms for variational data assimilation with discontinuous cost functions, M WEATH REV, 128(12), 2000, pp. 4031-4044
Cost functions formulated in four-dimensional variational data assimilation
(4DVAR) are nonsmooth in the presence of discontinuous physical processes
(i.e., the presence of "on-off'' switches in NWP models). The adjoint model
integration produces values of subgradients, instead of gradients, of thes
e cost functions with respect to the model's control variables at discontin
uous points. Minimization of these cost functions using conventional differ
entiable optimization algorithms may encounter difficulties. In this paper
an idealized discontinuous model and an actual shallow convection parameter
ization are used, both including on-off switches, to illustrate the perform
ances of differentiable and nondifferentiable optimization algorithms. It w
as found that (i) the differentiable optimization, such as the limited memo
ry quasi-Newton (L-BFGS) algorithm, may still work well for minimizing a no
ndifferentiable cost function, especially when the changes made in the fore
cast model at switching points to the model state are not too large; (ii) f
or a differentiable optimization algorithm to find the true minimum of a no
nsmooth cost function, introducing a local smoothing that removes discontin
uities may lead to more problems than solutions due to the insertion of art
ificial stationary points; and (iii) a nondifferentiable optimization algor
ithm is found to be able to find the true minima in cases where the differe
ntiable minimization failed. For the case of strong smoothing, differentiab
le minimization performance is much improved, as compared to the weak smoot
hing cases.