We are running some of our RegCM3 runs  over a rather large domain using the Lamber projection.

To compare the results with the CRU data from the ICTP RegCNET site, we use the RegCM3 CRU interpolation tool. Everything goes fine but a discrepancy  in positioning of the boundaries of sea areas in the CRU based patterns with the GrADS geography is evident  - since the CRU data are available over land only. No such problem is found in the case when the interpolation is made using normal Mercator  projection.

Any suggestion please.

  Simon

Krichak