AMIP II Standard Model Output
Questions concerning sampling, interpolation and model output compression

Perspectives of the AMIP Panel, the PCMDI staff, and many others have formed the basis for the AMIP II Standard Model Output sampling and interpolation recommendations. However, the implications of some of these recommendations are not yet fully understood. Tests are underway to examine the effects of these recommendations (published in the guidelines for AMIP II) on model validation and intercomparison. Here, questions are addressed concerning these recommendations. Questions and comments should be sent to: amip@pcmdi.llnl.gov.


Disclaimer: The responses to the questions below have been prepared or reviewed by the AMIP Panel, which often consults with individuals who are particularly familiar with the issue at hand. To some questions there may be no single right answer. Differing views are welcome, and will be added to the discussion below if they help to clarify the question or provide an important perspective on a contentious issue.
Some questions...

Return to AMIP home page



Question responses

Should 3-d fields be time averaged before or after interpolation to pressure surfaces?

In the AMIP II Guidelines no recommendation was made for vertical interpolation because there is no single method of interpolation that is most appropriate for all fields and models. However, how to interpolate and when to interpolate are very different questions. Some groups have expressed that their job-stream is more practical when quantities are averaged on model surfaces and then interpolated to pressure surfaces. With this approach, the average surface pressure is used for interpolation, despite the temporal pressure fluctuations at each model level. The resulting inaccuracies are not well known, and are likely to be variable dependent. This is the first issue addressed in a sampling/interpolation study. Preliminary results suggest that the effects of the ordering of interpolation for most fields is not critical, and thus if it is not the same for all AMIP simulations the resulting biases should not be too severe. However, when interpolating to pressure surfaces, it is recommended that these calculations be done as correctly as possible (i.e., accumulate each model time step on pressure surfaces). This is particularly important if cloud related fields are being interpolated to pressure surfaces. When the results of the sampling study become available, they may assist modeling groups in making their own judgements. The results of this study will also serve as a reference of potential bias in the output provided by groups who average on model surfaces and then interpolate to pressure.


Should all 3-d fields be interpolated to standard pressure surfaces?

Conceptually there should be no problem with defining most of the AMIP II 3-d quantities on pressure surfaces. Budget calculations with these data will only be approximations, and most features of fields that vary relatively smoothly in the vertical should not be lost when interpolating to standard pressure surfaces. However, fields with sharp vertical gradients or discontinuities such as cloud cover (and other cloud-related fields) may be significantly compromised by the interpolation to standard pressure levels. Preliminary tests help to quantify the severity of this. Nevertheless, the treatment of such fields varies substantially from one model to the next, and thus whether or not to provide cloud related fields (Table 1C) on model surfaces or standard pressure levels is to be decided by each modeling group. Clearly further study of this issue is necessary.


Last update 28 April 1997

LLNL Disclaimers

UCRL-MI-135952