This may be too late for you, but I’ve only just joined this community so I thought I’d send you reply anyway! I’ve posted about long-term monitoring to SD cards using CT/VT and meter pulse outputs. As long as you calibrate the CT and (more important) the VT at the outset, I found that this method would typically correspond within 5% of the meter data over long periods – if you then have the meter readings you can adjust the logged result. This was achieved using the recommended low cost 100A clip-on CT, so you could probably improve on the result by using a lower-rated CT commensurate with the maximum current you expect to see. I also compared profiles between this method of logging and logging of meter pulse data and found surprisingly good correspondence, even over lightly loaded periods.
Just to be clear, I assume you are using a CT and voltage source (via Emon products?) to get true power rather than relying only on a current reading and assumed voltage? I find that the voltage varies quite a lot, particularly in rural areas with lots of PV, and you may have phase variations between the voltage and current (non-unity power factor), so it is important to have the voltage profile for accuracy.
All the best
Simon (North Wales, UK)