Up to Specific issues in real coupled models
Hi, We have an apparent problem when allowing SCRIP/OASIS3-MCT to generate weights at run time. It seems that if we have fields which require BICUBIC weights followed by fields which require BILINEAR weights, then the BILINEAR weights are corrupted. However if we do things in a different order: BILINEAR first followed by BICUBIC, then things are OK. In a basic example we regrid from a high res to low res grid where the ratio of source to target grid cell sizes is uniformly 0.25 : 1.0. So the BILINEAR weights would normally be a uniform value of 0.25. However it seems that performing the BICUBIC weights generation first results in the subsequently produced BILINEAR weights file containing values which look more like those associated with the actual BICUBIC weights. e.g. instead of 0.25, 0.25, 0.25, 0.25, ....etc we have a repeating pattern of 0.25, 0.0625, 0.015625, 0.0625, 0.25, 0.0625, 0.015625, 0.0625, 0.25, 0.0625, 0.015625, 0.0625..... etc It kind of suggests something not being reset between generating different types of weights. We're using OASIS3-MCT vn4.0. Is this something seen elsewhere? Thanks, Richard
In lib/scrip/src/remap_bi_interp.F90, the logical variables must be initialised in the program and not during declaration (see ticket #2725) Best regards, Eric
