The Overseer engine does the model calculations to produce the results. The following information describing the engine has been sourced from the published Technical Manual Chapters available in the user portal. The Technical Manual Chapters describe the modelling methods used in the Overseer engine and set out the underlying principles, model parameters and sources of data used by the model. The chapters are primarily authored by David Wheeler and Mark Shepard (AgResearch Limited).
The engine uses a series of sub-models to estimate nutrient movement. The interaction of these sub-models produces the nutrient budget1. There are three levels of sub-models:
Much of Overseer modelling is about tracking the flow of nutrients - either as ‘transfers’ or ‘fates’. Some sub-models calculate nutrient transfers which focus, on the spatial (and often temporal) movement of nutrients within a farm system. Other sub-models focus on the fate of redistributed nutrients2.
For example, grazing animals consume pasture containing N. The amount of pasture dry matter (DM) consumed depends on animal production (animal energy requirement sub-model), and the N content of the consumed pasture (pasture growth and pasture nutrient sub-models). Grazing animals also consume supplements containing N (supplement sub-model). Of the total N consumed, some is used to produce products (e.g. live weight, milk, wool, velvet) and the rest is excreted as urine and dung (excreta sub-model). The excreted N can be deposited onto blocks, laneways or in the milking parlour, entering the effluent system and is applied to the effluent block. The excreta N that is returned to the pastoral blocks can be taken up by the pasture again and/or a proportion can be lost to the air (for example, by volatilisation and denitrification) or into drainage water (e.g. via leaching when the soil is draining).
A nutrient budget at block level captures the transfer of N in and out of the block. A nutrient budget at farm level captures the transfer of N in and out of the farm. So, internal transfers that happen within a block, (for example effluent movement, or supplements grown and fed on the farm) will be shown in the block-level nutrient budgets, but not in the farm-level nutrient budget - as the nutrients do not cross the farm boundary.
Understanding the design assumptions for Overseer (like all models) is critical for correct use and interpretation of results. Overseer has six key assumptions arising from its design, development and overall purpose.
The following assumptions give rise to the Overseer nutrient budget definition: The nutrient budget is a summation of annual inputs and outputs. The outputs are the average annual losses, assuming the inputs (management) are constant over time, for the given site characteristics. Click on the + for an explanation and example of each assumption.
An original design objective for Overseer was that it would be able to model information that is readily available on farm, or based on suitable defaults.
When Overseer was created, farmers (or their advisers) did not (and largely still do not) have access to daily climate data for a given location in New Zealand. Default climate data is supplied with the model, including:
Quasi-equilibrium means “inputs and site characteristics are in equilibrium with farm production and stock policy (stock numbers, breeding performance, crop yields etc.)”.
For organic systems, the model assumes that the farm production achieved is commensurate with fertiliser inputs.
Overseer assumes the input data describing the farm system is either actual or sensible. This is a consequence of the quasi-equilibrium assumption. It does not check if the system that has been entered is viable. It assumes that it is for the given inputs and farm production.
Overseer will not identify if input data is unrealistic, e.g. dairy cows on high country blocks, Kikuyu pastures in Southland.
The model estimates annual average outputs, assuming that management, inputs and farm production are constant for the given site-specific characteristics.
A research trial will likely produce different N leaching measurements in each year of a five-year study (due to site-specific conditions and management). The average N leaching for the trial duration is used (alongside other research trial data) to calibrate the model.
The model assumes that the farm is being managed in such a way that additional effects that cannot be captured by the model are having minimal impact on model outputs.
It is not possible to identify or quantify all possible situations, so good practice must be assumed. The intention of this assumption is to capture those unseen and unknown variations in on-farm practice that occur, not to define whether or not a practice is ‘good’ or ‘bad’.
The model assumes the effluent irrigator is applying nutrients in the correct place, not over a drain one day and a paddock the next. The model assumes that dairy cows use laneways to move from the paddock to the milking shed.
Given this assumption, when doing ‘what if’ analysis, farm production data needs to be adjusted if any input data is changed.
Reducing N fertiliser inputs might result in reduced feed supply and consequently productivity. Overseer is a reporting tool and records the farm as set-up, so the user needs to adjust the input information accordingly to ensure the farm system remains feasible.
All models are constrained by their design in some way. As with assumptions, understanding the scale, scope and approach of constraints on the model is critical to interpreting the results.
The Overseer model engine includes the following constraints:
The Overseer boundary is the farm boundary (which does not need to be spatially contiguous), and the bottom of the root zone (60 cm). Exceptions to this include greenhouse gas (GHG) emissions, such as embodied carbon dioxide (CO2) and indirect nitrous oxide (N2O) emissions.
Overseer does not interpret the effect of model outputs or the fate of lost nutrients. For example Overseer, does not report the environmental impact on water quality of nutrient loss from the farm system, nor does it produce a fertiliser plan based on fertiliser nutrient maintenance requirements.
Overseer does not account for the transformations of N that occur between when a nutrient leaves the farm or root zone and enters a receiving water body (vadose zone).
Overseer does not model within-stream processes that occur within the farm boundary. For example within stream attenuation, or stream-bed erosion. It does however account for attenuation due to denitrification from wetlands - and removal of P in overland flow by riparian strips.
Overseer does not model transition periods from one farm system to another. For example, if the farm starts using a slow-release fertiliser, Overseer assumes the slow-release fertiliser is operating effectively at the time it is entered into Overseer– meaning it assumes the production achieved by the fertiliser is commensurate with the fertiliser inputs. The same principles apply to organic systems.
Overseer is not a spatially explicit model and therefore does not take account of specific critical source areas (CSA) on a farm. CSAs are defined as landscape features (e.g. gully, depression) that accumulate overland flow from adjacent areas and delivers that flow to surface water bodies.
Overseer does not model loss of sediment due to mass flow, stream bank or stream bed erosion or loss of Escherichia coli (E.coli) or other microbes and pathogens.
While it accounts for many management practices and farm systems, Overseer does not represent ALL farm systems, management practices or combinations occuring in New Zealand. This means not ALL farms in New Zealand can generate an Overseer nutrient budget.
Overseer assumes effluent is applied in the month it is produced or selected as being applied (deferred effluent applications), and that within the month, effluent is applied in a way that minimises loss, including the use of storage facilities.
Overseer has not been validated against every combination of environment and farm enterprise (this is a limitation of every model). Overseer attempts to represent the various combinations based on scientific principles - so that estimates can be made.
Overseer does not provide an economic analysis of different management practices.
These examples focus on measuring the loss of Nitrogen (N) and Phosphorus (P) on a pastoral farm. For simplicity, these examples only consider the main factors influencing loss for each nutrient. On many New Zealand farms, other factors are also important.
On New Zealand pastoral farms, excreta from grazing animals (sheep, cattle and deer) is the primary source of N leaching. Urine and dung patches contain large N loads (up to 1000 kg N/ha). The N in urine is rapidly converted to nitrate, a highly mobile form of N, so is more prone to leaching.
While there are several ways that N loss occurs and is modelled by Overseer (see diagram below) Nitrate-N is the main form of N calculated, with some allowance for dissolved organic N (DON). Click on the letters to get an explanation of the different types of N Loss.
Description: Leaching of N from urine patches. N moved beyond 60 cm root zone (defined as farm boundary). Does not account for any transformations of N after this point (e.g., between leaving the root zone and receiving water body, collectively termed ‘attenuation’)
Description: Leaching of N from inter-urine areas and incorporates effects of dung, fertiliser, effluent and soil organic matter mineralisation. N moved beyond 60 cm root zone (defined as farm boundary). Does not account for any transformations of N after this point (e.g., between leaving root zone and receiving water body, collectively termed ‘attenuation’).
Description: Removal of nutrients from the land via overland flow.
Description: Nutrient deposited directly by beef or dairy animals into streams and/or drains i.e. when stock are not excluded from waterways and discharge from mole tile drainage systems.
Description: Nutrients discharged directly from effluent ponds into waterways.
Description: Nutrients discharged in irrigation outwash from border-dyke systems (i.e. surface runoff caused by irrigation).
Description: Nutrients discharged from septic tank outflow.
Non-grazed pastoral blocks can also contribute significant N leaching. A monthly N balance estimates available N (source) potentially leached, and transport (drainage). Leaching can be large when a large amount of available N from the monthly N balance corresponds with high drainage.
The primary drivers for N leaching from a non-grazed block are:
The importance of overland flow transport mechanisms means modelling of P loss is quite different to that of N (where drainage is the transport mechanism).
P losses in Overseer are calculated for each block under the following categories:
The model separates P losses into sources: background (soil), and incidental (effluent and fertiliser). Background P losses arise where P has been released from the soil, and is lost in flow events occurring through the year. Incidental P losses are based on an assessment of the probability of a flow event occurring when there is a risk of P losses from incidental inputs.
Background and incidental losses are calculated separately in the model, but rely on the same transport factors to move from the landscape to streams, such as rainfall, overland flow potential and topography. Rainfall is an important factor affecting P loss to streams, in particular when precipitation (rainfall plus irrigation) exceeds the soil infiltration rate and overland transport results.
The potential for overland flow is derived from the soil characteristics (drainage class and a slaking/dispersion index). Overseer estimates the effect of topography (a key driver of P runoff loss) using a subjective weighting to separate slope classes from flat to steep.
Management factors for background losses include Olsen P status and anion storage capacity (ASC), treading by grazing animals, and erosion from fence line pacing and deer wallows, which increase soil susceptibility to P runoff. A ranking for risk of P loss is used to account for the higher probability of overland flow events to occur, based on region and month of the year.
Management factors for incidental losses include the concentration, rate and timing of fertiliser/effluent application, the type of P fertiliser applied, and the speed of effluent application. Generally, P input application rates (kg P/ha) are multiplied by hydrologic and topographic factors that determine the potential for P runoff loss. For effluent application, management system factors (e.g., application method, depth and form of effluent) are also included.
Other P sources include P in animal dung (P intake less P in product), direct addition to streams (based on topography risk categories), lanes, silage stacks, pads and shelters, septic tank discharges. In terms of scaling from block to farm level, block P losses are aggregated using an area weighted average to a farm estimated loss, and then farm scale losses are added in. Farm scale losses include direct deposition of P from farm structures and laneways/races and P loss from ponds.
1 Wheeler and Shepherd, 2013b
2 Selbie et al. 2013
Go to next section: Uncertainty in model results
Go to previous section: What Overseer produces