Morning spikes / drops in GPP in temperate locations during summer?

[jan_ivar_korsbakken@cicero_oslo_no]

I am encountering some strange-looking behavior in the diurnal profile of GPP in temperate locations during the summer. Basically, GPP starts from zero at dawn as you would expect, then has a steep increase to a peak in the mid-morning, but then drops steeply to a lower level, and continues on a regular-looking diurnal curve from that lower level. It's consistent every day for most of the summer season, so it's hard to see that it would be weather-related (from increasing cloud cover or reduced humidity in the late morning, say). The example below is from 52N, 10E, in northern Germany. But we find the same pattern quite consistently in most places with a temperate seasonal climate and a decent amount of vegetation, including in the southern hemisphere (like southern Argentina/Chile, far southeastern Australia, displaced half a year relative to the northern hemisphere location obviously). In the example location, the phenomenon starts in the beginning of May and continues until mid-September. The plot below is in May, showing the onset. As you can see, before the phenomenon starts, the diurnal curve is relatively smooth. After it starts, the height of the "spike" is roughly the same as the maximum value of the curve before, but it for some reason drops down to a much lower level sometime before noon. This is consistent across all locations we checked. It only appears to happen in temperate vegetated locations in the summer. We have not found it in the winter half of the year, or in the tropics. NPP shows an almost identical profile. Heterotrophic respiration also shows similar behavior, though with a different profile. The case used here was ICLM45BGC with CRUNCEP forcing, with a 0.5 degree grid (hcru) and hourly history files. Has anyone seen this kind of behavior, or do you have any theories about what might be causing it? In particular, could this be a real phenomenon, or more likely an artifact or a bug? I can provide more details if needed, but just wanted to check first whether this is a known phenomenon.