questions for coupled roms-fennel model

[tony1230]

I am now using coupled ROMS-BIO_FENNEL to do some ideal tests. I ran into a problem with vertical diffusion.

I started from a simple cubic domain with horizontal dimension 32x64 and vertical 10 layers, with constant temp(10) and salt(28), no tide , nor flow. Bio_variables are also initialize with a constant value over the domain in both horizontal and vertical directions. I done some tests as follows.

Test1. With the model configuration aforementioned + no vertical sinking for Phy,LDet and SDet, i got all results reasonable according to biogeochemical processes. Say Phy die and convert into SDet and then LDet, Dets remineralization and produce NH4, and with nitrification NO3 was produced from NH4.

Test2. On the basis of test1, then i added a constant shotwave radiation via ana_srflx.F by giving a value RHO0*Cp*150 to srflx, homogeneous over the domain. Results show, in a short period, Phy grow fast along with NH4 and NO3 was rapidly consumed, and SDet was formed accordingly. Subsequently, NH4 go on increase by remineralization, but the problem is, with nitrification NO3 should have to increase, but to be ran out of approaching 0.

Test3. Based on test2, i actived the cpp for SOLAR_SOURCE to see vertical variation on those variables. Here comes a question, when i plotted out a vertical profile for temp and found unreasonable result. The temp on 5th layer equals initial value (10), from here the temp increase a bit(0.002 one by one layer) upward, but decrease downward at a same rhythm. And there represents almost nothing difference btw surface and bottom layer for bio-variables.

I have test another cases but no forward step was made.

Anyone who gives comments or tips would be greatly appreciated.

BTW: I am running the coupled physic-bio model using fennel module and have many things to learn and some questions to consult with sb.. So if you are doin the same work, i would like to get in touch with you.

Thank you in advance.

- Shou

[kfennel]

I don't see a problem with your Test 2. Nitrification is inhibited by light. If you impose constant light (without a day/night cycle) you effectively shut off nitrification in the sunlit layer, ammonium will accumulate and nitrate will run out. You can activate DIURNAL_SRFLUX to switch on a day/night cycle.

I don't follow you motivation for Test 3. I also don't see why a temperature profile where temperature decreases with depth would be unreasonable.

[tony1230]

Thank you Prof. Fennel

Before start out on my way to run the realcase physical-biological coupled model, i have successfully run the bio_toy test case. I looded into the results and also the input forcing files, to comprehend the relationships btw variables that on different trophic levels and relationships btw dirving forces and variables. Learned from this case and united with textbook knowledge, i thought i could grip the majority features of this coupled model.

Then i took my realcase into practice. Based on the well-calibrated-validated physical model, i put seven, viz. NH4,NO3,Phy,Zoo,LDetN,SDetN and Chlo, state variables in with all forcing drivers (u,v-wind,srflux,airp,airt,airq,lrflux and tides) imposed on. State variables are all initialized constants over the domain in both directions(ini-values:NH4=0.7,NO3=3.55,DetN=0.5,Zoo=0.2,Chl=1.0 and Phy=1.0, some of them are derived from investigation,some are estimated value). But as were outputted, the results show some unreasonable biogeochemical processes. The most remarkable is the process of Nitrification for NH4 to transform into NO3. Then i run a set of tests with values changed from the last case, but all run into a phenomena with which the NO3 consumed up and never incerase a bit even with Nitrification defined. Im now can understand what Prof. Fennel said that with a constant light nitrification effectively shut off, but in my realcase i employed light data from NECP and with diurnal fluctuation indeed. Why it behaves like the one i mentioned in first post of this thread?

I can not faring with my job in relatively a long period for nearly 2 months. So i decided to give it up temporary and to make a ideal test case which i my eyes with something different from bio_toy test case(perhaps a more idealized one with many constants, but i thought this can make me much clear when come out a problem). Here starts the questions in first post, though Fennel offered here instructional suggestion to activate DIURNAL_SRFLUX on to introduce a day/night cycle, but the results seemed to me still a problem. Attached are two plots represent two cases with DIURNAL_SRFLUX on and off repectively. In my file of Exclude/bio_Fennel.h, i have followings(for conveniency and space economy, comments and blank lines were excluded.Besides, as Zoop and sinking were not employed in this test case, variables with values of zeros also were not included.).

Code: Select all

Lbiology == F
     BioIter == 1
       AttSW == 0.04d0
      AttChl == 0.025d0
     PARfrac == 0.43d0
         Vp0 == 1.0d0
     I_thNH4 == 0.0095d0
     D_p5NH4 == 0.036d0
      NitriR == 0.05d0
       K_NO3 == 1.0d0
       K_NH4 == 1.0d0
     Chl2C_m == 0.05350d0
      ChlMin == 0.001d0
       PhyCN == 6.625d0
       PhyIP == 1.5d0
       PhyIS == 0.125d0
      PhyMin == 0.00d0
       PhyMR == 0.07d0
       SDeRRN == 0.03d0
       SDeRRC == 0.0d0
     pCO2air == 370.0d0

I have been wondering if some inportant aspects of the biogeochemistry in nature systems have been missed in my configurations. But i have no answer for this. In definition file i defined these options related.

Code: Select all

#define BIO_FENNEL
#ifdef BIO_FENNEL
#define BIO_SEDIMENT
#undef CARBON
#define DENITRIFICATION
#endif

I need your help in urgent.

Here followed some questions beside the point aforementioned.

1. The biological bottom boundary layer(BBBL). As introduced in Fennel's paper (Nitrogen cycling in ...,2006), appendix A. First i want to ask, are there any biogeochemical processes based on BBBL in ROMS? I can tell i never defined something or assigned values to variables with exception for BIO_SEDIMENT which was explicitly told to activate always to ensure conservation by converting the matter sinking out of the aquatic system. To me, it want to tell that matter/nutrients can be released back into the uplay water system via sinking-remineralization-nitrification processes. But where can i find them in roms?

2. We know processes like mineralization, nitrification and organic matter decomposition need oxygen. Fennel has had some of them represented in that papre. So i also have been thinking if i would add variables like ALK, TIC and OXYGEN into model configuration, or under what circumstances i have to/must include them in. I have no reply for myself.

Um... looks like a savage that needs your help and your patience. Perhaps i should raise them one by one for easy solution, but that would break one's thoughts into pieces.

Anyone can comments on any point of view. Thank you very much indeed.

- Shou

[laanderson]

Hi Shou,

I am not very familiar with ROMS, but I have tried this sort of thing in other models, so hopefully I can help.

1. What are the depths or thicknesses of your 10 vertical levels? If they are e.g. 1 meter each, then even your deepest level will have too much sunlight for net nitrification to occur. So one solution might be to make your levels thicker---so that your deepest level is at 200 meters depth, where it is dark enough for nitrification to occur at all times of day.

But maybe you are interested in a case where the bottom depth actually is 10 meters. River and coastal waters are often mirky, so the solution in this case would be to increase the light attenuation coefficient for water (assuming that it contains alot of silt) to Jerlov Type III or more.

Or maybe you are interested in shallow water that is not mirky but clear---like a beach in the tropics. In that case, net nitrification can only occur in the sediments---you'd need to include a submodel that computes that conversion and a sediment-to-water column nitrate flux. Otherwise, as Katya said, it is perfectly acceptable and realistic to have nitrate disappear in the simulation.

2. I recommend that you initialize the model with some density stratification. I suspect that you used constant T and S in an effort to try to eliminate all physical processes. But it doesn't; constant T and S will allow convective instabilities to occur, and vertical mixing fluxes will be high, not low. To minimize the physical affects, use strong stratification. If T impacts biological rates, and you want those to be constant, then stratify with S.

3. I believe I understand your concern with Test 3 in your original post. You added solar heating, and so are surprised that the deep levels cooled, instead of heating or staying constant. I suspect this cooling is due to a combination of 2 factors. (1) Is domain-averaged T conserved with time in your simulation? Even though you apply solar heating, if your net air-sea heat flux is required to be zero (I.e. T is conserved), then you may be getting cooling at the sea surface and/or at night. (2) Because you initialize with constant T and S---essentially one big mixed layer---this cooled water will penetrate convectively to the bottom levels. This might be as convective plumes with timescales less than one day (if cooling is at night), or in your simulation it might just occur through gradual vertical diffusion---the vertical diffusivity being very high because the stratification is very low.

So make sure your air-sea heat flux is what you want. I am not sure how this is handled in ROMS, but if you want your water column to warm (stratify) due to solar heating only, then your net air-sea heat flux must equal that amount. Actually, running with no net surface heat flux is not unrealistic---in the tropics, solar heating during the day is approximately balanced by surface cooling at night, and the mixed layer depth has a diurnal cycle. If your solar heating does not have a diurnal cycle, and if you start with some stratification (as suggested above), I think the simulation will develop a mixed layer above the remaining stratified water, and the bottom levels will not cool over time.

4. Is domain-averaged total Nitrogen (NO3 + NH4 + SRN + ...) conserved with time in your simulation? That is one way to make sure sediment denitrification is not occuring in your model, which I believe you do not want (yet).

5. In your second post, your first figure shows NO3 increasing with time and all other N pools decreasing. Rather than being the "good" run, this might be the "bad" run---i.e. this is what would happen if there were no light at all. The second figure, where NO3 depletes, is more what I would expect for a tropical mixed layer.

Some of my comments may be off the mark regarding your simulation---but I hope they at least give you some helpful ideas.

All the best,

Larry

[tony1230]

Wow...,big bang.
Thank you Larry for your time and carefulness.

1. What are the depths or thicknesses of your 10 vertical levels? If they are e.g. 1 meter each, then even your deepest level will have too much sunlight for net nitrification to occur. So one solution might be to make your levels thicker---so that your deepest level is at 200 meters depth, where it is dark enough for nitrification to occur at all times of day.

Yes, in my test case it shares one value for depth -- 10 meters -- i.e. just a flat bottom and the thicknesses share also a value 1 meter each layer. The reason why i chose 10 meters is the bathymetry over my real application close to it covers a magnificent portion. There are rivers -- e.g. Yellow R., one of the rivers in the world which have huge values for sediment concentration -- flow into the domain and some are carry huge tons of sediment in flood season. You remind me that my water depth is so small(light can penetrate through the water and reach bottom in some areas) and i may also use a too big value for light attenuation for water(i use default 0.04 1/m embraced within roms), if not the case, your suggestion still worth a trying. Yes, we all expect a submodel to work for sedimental biogeochemistry. But as questioned in my second post, i don't know how to substantially define a Biological Bottom Boundary Layer(BBBL). So, in my test case there are no vertical sinkings for phytoplankton and detritus, once they out of the system(into the sedimental pool) they would never come back again. If that is the case, the system will be unconserved for nitrogen.

I suspect that you used constant T and S in an effort to try to eliminate all physical processes. But it doesn't; constant T and S will allow convective instabilities to occur, and vertical mixing fluxes will be high, not low.

Yes, you hit it. As i more concern with the biogeochemical processes, i don't want noises, such as tides and wind etc., to affect and to modulate the test results. That would be confuse me a lot. When you said, with constant T and S could cause convective instabilities, but why? and how it cause high vertical mixing flux? I konw a bit that using constant T and S with a fluctuant bathymetry, there may induce a tiny depth variation in density then introduce a small velocity.

You added solar heating, and so are surprised that the deep levels cooled, instead of heating or staying constant. .... (1) Is domain-averaged T conserved with time in your simulation? Even though you apply solar heating, if your net air-sea heat flux is required to be zero (I.e. T is conserved), then you may be getting cooling at the sea surface and/or at night. (2) Because you initialize with constant T and S---essentially one big mixed layer---this cooled water will penetrate convectively to the bottom levels. This might be as convective plumes with timescales less than one day (if cooling is at night), or in your simulation it might just occur through gradual vertical diffusion---the vertical diffusivity being very high because the stratification is very low.

Yes, i didn't think i would get a decreasing value in temperature with solar radiation imposed on. I attach here a profile for temperature from which you can find a increase upward and a decrease downward. As the temperature was initialized with constant, and with an equal perform for vertical coordinate, so i simply summed the temperature of 10 layers but only to find that it is a bit smaller than the total temperature(10 layers with 10 Cdeg,i.e. total = 10x10=100, but the model result is 99.9945 even with solar radiation). I dont know if can be said that the T field wasn't conserved. Stratification can be a much important factor that could controls biogeochemistry in many aspects and i should have took this into account.

Is domain-averaged total Nitrogen (NO3 + NH4 + SRN + ...) conserved with time in your simulation? That is one way to make sure sediment denitrification is not occuring in your model, which I believe you do not want (yet).

Yes, i can ensure that the nitrogen is conserved with time, i culculated the nitrogen anomaly (at present in test case is the sum of NO3,NH4,SDN and Phytoplankton) at every timestep, to find a largest value reaches an order of magnitude for -7, i.e. ~ 2.5*10 ^ -7. This can be neglect i think in our modelling.

In your second post, your first figure shows NO3 increasing with time and all other N pools decreasing. Rather than being the "good" run, this might be the "bad" run---i.e. this is what would happen if there were no light at all. The second figure, where NO3 depletes, is more what I would expect for a tropical mixed layer.

Frankly, i can learn from the second figure somewhat that phytoplankton growing rapidly along with nutrients consumed at a hing-speed. What i want to suspect in this case is no NO3 to be produced. As Katya said, i was not expected to gave the system a constant light intensity. But i could not understand from the first figure/case with DIURNAL_SRFLUX defined to introduce a day-night variation. I want to know why Phy decrease from initiation without even a minimum increase with NH4 been uptaken. Would that be the reason for much larger nitrification rate? I will have a try...

One point to notice that the temperature anomaly is only happened in the case with ANA_STFLUX defined and DIURNAL_SRFLUX undefined.

Welcome your comments. Thanks.

- Shou