A draft for a manuscript entitled
Multiple linear regression as a tool in oceanography
The use of multiple linear regression (MLR) for data interpolation is presented and put in relation to oceanic process as for example mixing. Using data from the South Atlantic error estimates are given for various parameters and different MLR schemes. Finally further applications of the MLR method are presented: inferring missing parameters, deducing temporal changes like the increase of CO2 in the ocean, and calculating parameters on a world wide regular grid.
Some further Descriptions and programs are also available: