Comments to Preliminary Draft Report (First Version) Contribution by Victor M. Ponce, Panelist July 8, 2005 Response to comments by Rene Therrien on July 8, 2005. The earliest reference to implicit schemes was made by O'Brien et al. (1950), which has been cited in the Draft Report. O'Brien et al. state that an implicit scheme is that in which the unknowns occur simultaneously, rather than singly, as would be most convenient for computation (as in an explicit scheme) [see O'Brien's paper online at "http://ponce.sdsu.edu/obrienandhymanandkaplan231.html"]. Thus, the term implicit is taken to refer to a simultaneous, i.e., line-by-line, solution, requiring matrix inversion; while an explicit scheme, being a point-by-point solution, does not require matrix inversion. The value of the weighting factor α per se does not define whether a scheme is implicit or explicit. In general, a scheme having a value of α smaller than 0.5 is still implicit. However, a value of α = 0 converts the scheme into explicit, by definition (of implicit). Typically, the stability of a scheme improves as α increases above 0.5 (with α = 1 being the practical upper limit), and it deteriorates as α decreases below 0.5. Values of α smaller than 0.5 are not used in practice, so there is no need for concern. Therrien is correct in pointing out that in Section 2.3, the word "convergence" is used to mean two different things: (1) iterative convergence, where convergence refers specifically to the performance of the end product of the iterative procedure, and (2) direct convergence, where it refers to the closeness of the (direct) numerical and analytical solution, where there is no iteration involved [see O'Brien's paper online at "http://ponce.sdsu.edu/obrienandhymanandkaplan237.html"]. To clarify the above issue, I suggest the following modification to the Draft Report: Page 7, Section 2, paragraph 2:   Replace the sentence "However, in numerical modeling, stability is usually achieved at the expense of convergence." with "However, in numerical modeling, stability is usually achieved at the expense of convergence, in the sense of O'Brien et al. (1950)." Regarding Therrien's comment on whether ET should be calculated in or out of the model, it is best to let the District judge how heavy, from a computational standpoint, the ET component is in an actual model application. Obviously, if the ET component amounts to more than 30% of the computer time, it should be taken out of the RSM and evaluated as an input. Conversely, if the ET component takes less than 1% of the time, it may be more advantageous, from a modeling standpoint, to specify primitive variables such as solar radiation and/or air temperatures as spatial (and temporal, as ecosystem restoration is implemented) input. As pointed out by the District's modelers (see Irizarry-Ortiz, 2003), due to differences in roughness characteristics, the crop coefficients developed for grass reference ET will need to be modified for use with wet-marsh potential ET.

050708