Question about subdaily boolean in &meteorology block

[Arnaud PUJOL]

Hi everyone,

I want to do simulations of primary production for synthetic lake. For the exercise, I suppose I won't have any hourly meteorological data, and therefore, I would like to use a daily approach. In order to be assured that the daily approach was consistent with the hourly one, I made a benchmark. This is where I encounter something peculiar.

Here is the first lines of the meteorological file I used,  with an hourly discretization but where I have averaged the values on a daily basis:
time,ShortWave,Cloud,AirTemp,RelHum,WindSpeed,Rain
2007-01-01 00:00:00,43.94,0.7955566666666667,9.625,87.55556666666666,4.25,0
2007-01-01 01:00:00,43.94,0.7955566666666667,9.625,87.55556666666666,4.25,0

...

2007-01-01 22:00:00,43.94,0.7955566666666667,9.625,87.55556666666666,4.25,0
2007-01-01 23:00:00,43.94,0.7955566666666667,9.625,87.55556666666666,4.25,0
2007-01-02 00:00:00,55.69,0.6688866666666667,4.95,85.88889999999999,4.25,0
2007-01-02 01:00:00,55.69,0.6688866666666667,4.95,85.88889999999999,4.25,0

...

2007-01-02 22:00:00,55.69,0.6688866666666667,4.95,85.88889999999999,4.25,0
2007-01-02 23:00:00,55.69,0.6688866666666667,4.95,85.88889999999999,4.25,0
2007-01-03 00:00:00,55.913333333333334,0.6555566666666667,5.6499999999999995,88.33333333333333,2.0,0
2007-01-03 01:00:00,55.913333333333334,0.6555566666666667,5.6499999999999995,88.33333333333333,2.0,0

...

Here is the first lines of my second meteorological file with a daily discretization:

 time,ShortWave,Cloud,AirTemp,RelHum,WindSpeed,Rain

2007-01-01,43.94,0.7955566666666667,9.625,87.55556666666666,4.25,0.0
2007-01-02,55.69,0.6688866666666667,4.95,85.88889999999999,4.25,0.0
2007-01-03,55.913333333333334,0.6555566666666667,5.6499999999999995,88.33333333333333,2.0,0.0

...

As you can see, both files has the same values. I used a timestep of 3600 for each simulation.

I expected to have the same results. However, when I compare the net primary productivity (derived from the NCP output property) in both simulations, I found quite different results. Curves have the same shape, but the amplitude is very different (up to 50 %). 

I have tried something else: I changed the timestep from 3600 to 86400 for my "daily" simulation. There was a lot of ugly numerical artefacts but I got the same amplitude as for my hourly simulation.

Could anyone tell me why, with the same data, daily and subdaily simulations do not give the same results?

Why changing the timestep modifies so much my results?

Best regards,  

Arnaud

[Jordi Prats]

Hi Arnaud,

When using a daily meteorological file and a 3600 timestep, GLM estimates a daily (sinusoidal) cycle for solar radiation and solar radiation is zero during the night. Instead, in your hourly file you assume solar radiation is constant all the day round (including the night). The forcing is different in each of the two cases.

Best,

Jordi

El dia dijous, 6 de juliol de 2023 a les 16:49:09 UTC+2, pujol....@gmail.com va escriure: