How to use ”time“ in kernel

[lzhucheng]

hi， @erikvansebille
I want to know the meaning of "time" in the below code and how to use time in my kernel. Can you give some advices？

`def AdvectionRK4(particle, fieldset, time):
"""Advection of particles using fourth-order Runge-Kutta integration.

Function needs to be converted to Kernel object before execution.
"""
(u1, v1) = fieldset.UV[particle]
lon1, lat1 = (particle.lon + u1*.5*particle.dt, particle.lat + v1*.5*particle.dt)
(u2, v2) = fieldset.UV[time + .5 * particle.dt, particle.depth, lat1, lon1, particle]
lon2, lat2 = (particle.lon + u2*.5*particle.dt, particle.lat + v2*.5*particle.dt)
(u3, v3) = fieldset.UV[time + .5 * particle.dt, particle.depth, lat2, lon2, particle]
lon3, lat3 = (particle.lon + u3*particle.dt, particle.lat + v3*particle.dt)
(u4, v4) = fieldset.UV[time + particle.dt, particle.depth, lat3, lon3, particle]
particle.lon += (u1 + 2*u2 + 2*u3 + u4) / 6. * particle.dt
particle.lat += (v1 + 2*v2 + 2*v3 + v4) / 6. * particle.dt`

[erikvansebille]

Hi @lzhucheng. Thanks for asking. I agree it may be a bit confusing what time is, but it is the 'clock' of the loop. It is in seconds, relative to the first day of the data of the fieldset. So after a day of simulation, you can expect time=86400. This is independent of the calendar used because parcels internally uses seconds since start of fieldset.

Does that help?

[lzhucheng]

Thanks for your answer.
Does "time" influence the trajectory of particles? I got different results when I use (u1, v1) = fieldset.UV[particle] and (u1, v1) = fieldset.UV[time, particle.depth, particle.lat, particle.lon] in AdvectionRK4 kernel.
so I want to know how to use "time" in field sample fieldset in a kernel.
In addition, why do some use (u1, v1) = fieldset.UV[particle] while others use (u1, v1) = fieldset.UV[time, particle.depth, particle.lat, particle.lon] in AdvectionRK4 kernel？

[erikvansebille]

Oh surprising that you get different results; that shouldn't happen. Can you send some more details of your experiment? A minimal breaking example would be useful

The reason why sometimes we use fieldset.UV[particle] is because this is faster in Scipy-mode, see this part of the JIT-vs-Scipy tutorial. For JIT mode, it shouldn't make any difference...

[lzhucheng]

Thanks for your answer.
I wonder about that i got diffent result between particle.lon(next loop) and particle.lon_nextloop. I want to kwon what happen about that？
Here is the code
`
class CustomParticle(JITParticle):
age = Variable('age', dtype=np.float32, initial=0.)

lon_nextloop = Variable('lon_nextloop', dtype=np.float32, to_write=True, initial=attrgetter('lon'))
lat_nextloop = Variable('lat_nextloop', dtype=np.float32, to_write=True, initial=attrgetter('lat'))
time_nextloop = Variable('time_nextloop', dtype=np.float32, to_write=True, initial=0.)   

# particle_state: 
particle_state = Variable('particle_state', dtype=np.int32, initial=0.) 
beached_count = Variable('beached_count', dtype=np.int32, initial=0.)

def Advection(particle, fieldset, time):

particle.lon = particle.lon_nextloop
particle.lat = particle.lat_nextloop

(u1, v1) = fieldset.UV[time, particle.depth, particle.lat, particle.lon]
# (u1, v1) = fieldset.UV[particle]
lon1_adv, lat1_adv = (particle.lon + u1*.5*particle.dt, particle.lat + v1*.5*particle.dt)

(u2, v2) = fieldset.UV[time + .5 * particle.dt, particle.depth, lat1_adv, lon1_adv]
lon2_adv, lat2_adv = (particle.lon + u2*.5*particle.dt, particle.lat + v2*.5*particle.dt)

(u3, v3) = fieldset.UV[time + .5 * particle.dt, particle.depth, lat2_adv, lon2_adv]
lon3_adv, lat3_adv = (particle.lon + u3*particle.dt, particle.lat + v3*particle.dt)

(u4, v4) = fieldset.UV[time + particle.dt, particle.depth, lat3_adv, lon3_adv]

particle_dlon = (u1 + 2*u2 + 2*u3 + u4) / 6. * particle.dt
particle_dlat = (v1 + 2*v2 + 2*v3 + v4) / 6. * particle.dt     
particle.lon_nextloop = particle.lon + particle_dlon
particle.lat_nextloop = particle.lat + particle_dlat 

`

[erikvansebille]

Please have a look at the kernel loop tutorial, which is exactly about this topic. But this is not the code that breaks the time (point above), right?