Carbon model cons
As with any model, simplification has both advantages and disadvantages. This article describes some of the silicon-phosphorus model cons.
Types of Models
In computer modelling there is a continuum between two caricature classes of models:
- conceptual-type models that aim to simplify a system down to include only its most important elements, in particular those that govern its behaviour
- simulation-type models that are highly detailed and try whenever possible to avoid making assumptions about what is important; instead they try to include as much detail as possible in the expectation that nothing important will be omitted
By simplifying to only the most important fluxes and state variables, this model is very much a conceptual-type model.
Potential weaknesses of a conceptual approach include the likelihood of omitting something important. Every assumption runs the risk of being incorrect in an important way and thereby divorcing the model results from reality. It only takes one unreasonable assumption or omission, if it relates to a key part of the system, to invalidate the model's results.
Some of the major simplifications made in the silicon-phosphorus model are as follows:
- The representation of biology is extremely simplified. Limitation of growth rate due to light, temperature and other nutrients (for instance, nitrate and iron) are not included at all in the model. The omission of iron, in particular, is a potentially critical deficiency of the model, because observational evidence suggests that diatoms have anomalously high Si:P ratios in iron-deplete regions such as the Southern Ocean.
- There is no day-night cycle and no seasonal cycle in the model.
- Horizontal variations are completely ignored. The model is unable to simulate differences between different ocean basins, between high and low latitudes, or between deep and shallow water environments.
- Vertical variations can only be captured in an approximate way due to only two boxes in the vertical.
- Only the major fluxes are included. Fluxes deemed to be of lesser importance, such as the delivery of phosphorus to the surface ocean within dust, are not included.
- Only two different types of phytoplankton are simulated in the model.
- Only inorganic phosphate is modelled. Dissolved organic phosphate (DOP) is not modelled.
- When under silicon stress in the real world, diatoms can decrease their silicon requirement to below the ideal Si:P ratio. In the model, nutrient stress can only increase this ratio (under conditions where phosphate is more limiting than silicic acid).
The motivation for all of these omissions is the expectation that their addition would not fundamentally change the behaviour of the model. It is predicted that adding these features would not greatly alter the model's general response to, for instance, changes in the amount of phosphorus in the ocean, or changes in the input rate of phosphate down rivers.
For many of the cases above sensitivity analyses have been performed to test these assumptions. The large majority found that the main results were obtained regardless of whether or not we made the assumptions listed above. More information on these, and other, sensitivity analyses can be found in Yool & Tyrrell (2003).
- Silicon-phosphorus model overview
- Silicon-phosphorus model details
- Silicon-phosphorus model pros
- Silica burp hypothesis
- Yool, A. & Tyrrell, T. (2003). Role of diatoms in regulating the ocean's silicon cycle. Global Biogeochemical Cycles 17, 1103, doi:10.1029/2002GB002018.
- Description of the chemical element silicon, Wikipedia
- Diagram of the silicon cycle, Wikipedia
- Description of the chemical element phosphorus, Wikipedia
- Description of the phosphorus cycle, Wikipedia
You wouldn’t use a chain-saw to cut a slice of bread, or a bread-knife to cut down a tree. Similarly in modelling, it’s important to choose the most appropriate model for the particular question being addressed. This particular model is conceptual… Due to the simplifications made, this model is not the most appropriate for many applications.
Lack of spatial resolution: the physical structure of this model consists of three boxes stacked vertically one above the other. It is therefore slightly more complicated than the P, NP and SiP models, which consist of only two boxes. However, there is again no ability to represent any horizontal differences. For this reason this model is inappropriate (unable) to look at regional changes in carbon cycling. It also means that this model will not be the best tool for exploring palaeo events during which the ocean circulation underwent significant reorganisation (e.g. a shift in the location of deep-water formation from the high-latitude North Atlantic to the high-latitude North Pacific).
Because of the lack of horizontal resolution this model cannot simulate the solubility pump in the ocean. The solubility pump arises out of the fact that it is cold waters that sink to the ocean depths, and cold waters usually hold more dissolved gases (such as CO2) than do warm waters. This pump is therefore a secondary cause of the vertical gradients in carbonate chemistry that are observed in the oceans.
Lack of explicit sediments: some ocean biogeochemical models have a separate part devoted purely to sediments, in which the vertical profiles of sediment material and pore-water chemistry are represented. Processes such as burrowing of worms, decay of organic matter, dissolution of calcium carbonate, etc, are explicitly calculated in each vertical millimetre of sediment. This model does not do that. This lack prevents modelling of the process of carbonate “burn-down”. If, as is starting to happen at present, deep-water pH falls over time, previously deposited CaCO3 can potentially be dissolved many years after it originally fell to the seafloor, because the overlying waters are becoming increasingly corrosive over time. Because of this lack, our model is likely to respond too slowly at times when the ocean is becoming more acidic.
On the other hand, CaCO3 burn-down increases the speed of response to acidification, but does not alter the eventual end-point. The model will converge to the same eventual end-point through the carbonate compensation feedback, but without CaCO3 burn-down will reach it more slowly than it should.
The lack of explicit sediments also prevents the model from representing any decoupling between bottom-water chemistry and sediment pore-water chemistry. This could arise if, for instance, organic matter is being rapidly respired into the sediments, which would make the pore-water more acidic. This omission from the model probably doesn’t introduce any large-magnitude errors.
- the model does not represent calcite and aragonite separately. Coccollithophores and foraminifera secrete the calcite polymorph of CaCO3 when building their shells. Pteropods and coral reefs synthesise the aragonite polymorph. The two polymorphs have different solubilities (i.e. different susceptibilities to dissolution). Some organisms also form shells made of high-magnesium calcite, which again has a different solubility from the more usual form. Biological CaCO3 therefore consists of a spectrum of different materials with different dissolution characteristics, but none of this is represented in the model. In the same water-column, for instance, aragonite dissolves much higher up than does calcite (the compensation depths differ). - Related to this last omission, in reality calcite starts to dissolve at the calcite lysocline and then finally disappears altogether from the sediments at the calcite compensation depth. In-between the sediments have intermediate calcite contents. In the model this is all simplified to a sharp transition at one depth. The sediments therefore change from 100% to 0% CaCO3 at the single CCD. - There are uncertainties in the numbers produced by the model due to uncertainties in the carbonate chemistry constants used in the model, although the size of the associated uncertainties is fairly small compared to most other factors. - The production of CaCO3 in the surface layer (proportional to POC production) is extremely crude. Populations of coccolithophores, foraminifera, pteropods, etc are not represented as separate variables in the model. In any case we do not understand at all well how their population densities are controlled, and so it is not yet possible to code this accurately into models. - Shallow-water calcification is extremely simplified. Although in reality it accounts for a large fraction of overall CaCO3 burial in the global ocean, in the model it is represented as just one number. - Only one feedback is included in the model – carbonate compensation. Other potential feedbacks, for instance those associated with silicate or carbonate weathering, are not included. -