tag:blogger.com,1999:blog-7729093380675162051.post3847139867303544667..comments2024-03-28T13:56:47.604+11:00Comments on moyhu: TempLS: new V3Nick Stokeshttp://www.blogger.com/profile/06377413236983002873noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-7729093380675162051.post-26581856345273161212016-12-09T11:49:33.423+11:002016-12-09T11:49:33.423+11:00Peter,
The least squares method I use does not nee...Peter,<br />The least squares method I use does not need anything special for missing records. It fits a linear model to the monthly means provided by GHCN/ERSST by least squares. I don't handle the process of deriving monthly from daily. There should in principle be a test of whether a station has sufficient data, but GHCN has handled that.<br /><br />There is one small exception - SST near ice. ERSST is optimal interpolation, so it provides data without gaps. But some of them are ice, which I treat as missing. That leaves a fringe area which is sometimes ice, sometimes not, and a bias because the missing months are predominantly winter. I haven't found a good way of dealing with that yet.<br />Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.com