Up: Multiple linear regression as
Previous: Multiple linear regression as
The resolution of bottle data is low compared to CTD data. Whereas
some parameters are normally measured from every bottle
(e.g. nutrients), the measurements of other parameters are to
cumbersome or take to much time so that this parameters can only be
measured every second or third station/bottle and therefore have even
a lower resolution. To get profiles the bottle data can be
interpolated using pressure. Or, to plot a section, be gridded using
distance (or latitude/longitude) and pressure (or density, depth).
Such interpolations do not make use of the information from other
measured parameters. The MLR method presented here, uses this
information and can reduce the interpolation error and can even be
used to infer missing parameters or depict temporal trends.