Overseer should address the real issue here. Year End nutrient budgets representing actual management practices in combination with a synthetic daily climate model based on long term average annual rainfall and regional parameters are nonsensical. This is the case for dryland as well as irrigated landuse. Stocking rates and crop production for drought years is being matched to average rainfall with drainage being over estimated, while the effects of wet years on leaching and runoff are being ignored. As noted in the discussion it is also nonsensical to match the current Overseer irrigation predictions to actual irrigation.

While not 100% accurate, use of the Virtual Climate Station Network historical predictions for rainfall, PET and temperature would be a major step forward. The NZ government should use some of the funding that it is providing to improve Overseer to make the VCSN service free for all users (not just for Overseer).

For predictive nutrient budgets the use of multiple years of VCSN data would help give some indication of wet year and dry year variability, although management adaptation would still not be accounted for.

Whilst using VCSN climate data would help, it does not address the issue of variability in soil properties impacting on irrigation requirements and efficiency, but would result in predicted historic irrigation having a closer alignment with actual than is the case with the current fixed climate model.

Thanks Alister, whilst what you propose is an interesting hypothesis it's not really relevant to this discussion.

Overseer and OverseerFM are strategic management tools. We continue to believe that the use of the long term climate data is the most appropriate method of including climatic conditions in the model. It not only supports OverseerFM's use as a predictive tool, but it also allows farmers, rural professionals and regulators to look back at Year Ends - and understand the effects of management practice changes, rather than the impacts of weather events.

OverseerFM should be used to understand a farm system and as regularly submitted on by organisations such as FANZ and Ravensdown, it should not be used as a "calculator" for annual N losses. It is farm systems performance over a period of time which is important. So if a "baseline" is needed, it should be taken over a number of years - as we see with the ECAN 4 year period. This provides the ability to understand the impact of mitigations in the long term.

It does of course mean that some skill is needed in interpreting "actual" farm data, and entering that into OverseerFM in a way that supports the strategic farm management approach - which of course is recognised by most regulators in their recommendations or requirements for the involvement of ASNM and/or CNMA advisors in either the production or auditing of OverseerFM analyses.

Agree OverseerFM should not be used to monitor water use. For NB audits, management method would be the preferred option. For an additional level of checks, monthly use in FM against Irricalc or Irrimap would be enough to raise outliers.

I would rather keep monthly water applied for quick checking after modelling. Soil moisture and drainage visibility would be nice. No to 1 management option, lots of irrigators are managed differently in 1 yr (shoulders and peak). Yes to FF limit. Some warnings would be great (long term model-don't use actual data).

Apologies seems I was the only one who could see the document - now attached above as the proposal.

I agree with the general direction of the document. It is important that all data is always entered based on consideration of the long term climate data used by Overseer.

I believe Overseer should continue to show the amount of water supplied as it does provide a useful data entry sense check, it should just not be expected that it is an obsolete sense check for any given year (I would generally consider +/- 100mm/ha to an ok guide).

Any review of modelled accuracy should focus on the methodology modelled relative to that undertaken on farm, not absolute volume. Although it is not Overseer’s problem, increasingly farmers have soil moisture monitoring which displays graphically with two lines, which they attempt to stay between. Often these systems are not calibrated, and the farmer does not fully understand what the lines mean in terms of mm deficit in the soil, or what the soil PAW is relative to the % scale displayed.

I see benefits and risks (regulatory) in displaying of soil moisture and drainage over time. The benefit is it may hep improve understanding of the irrigation sub model. The risk is that farmers / modellers in a restrictive regulatory environment may start to groom triggers and targets. Given the way Overseer calculates its daily data using an algorithm, I currently see more downside than upside to displaying this data over time.

I agree with Marcleo that restricting an irrigator to a single management type is undesirable as farmers will often use more than one decision rule – a good example of this is fixed irrigation activities for crop establishment in conjunction with soil moisture monitoring once established.

I currently do not support the concept of limiting of FF inputs as this would make some of the existing regulatory monthly inputting such as using Irricalc as a proxy for GMP impossible.  Additionally, this flexibility also needs to exist so to regulate how Overseer currently interprets data monthly e.g A farmer says on average I irrigate three time per month applying 20mm/pass Oct - March. This would roughly translate to 20mm on a 10 day return, but this data would create 4 applications (80mm) in Oct (31 day month), but then 60mm (3 applications) in other months.

With regards to warnings and help options - it cannot be stressed enough this is a long-term model. Increasingly care should be advised if using PAW triggers or targets on blocks containing grouped siblings as the output of this modelling approach is almost never repeatable on farm without VRI. Increasingly the use mm deficit triggers and targets are advisable to best represent the more common soil moisture installation in a predominate sibling being used to advise an irrigator/s across the lesser siblings. The old flow chart in the BPDIS was quite good and could likely be improved upon to provide better guidance.

Earlier this year I attended a meeting where we were presented with the results of three well-known and respected Overseer modellers, all modelling the same irrigation system on the same property. The irrigation system was not complicated yet all three modelled the system differently, confirming what I suspect many of us already know, that there is little consistency in how irrigation is modelled. The fact that some users “.. are trying to adjust these inputs on a monthly basis to get the same amount of water applied (estimated by the model) as was measured on the actual irrigator for that year …” is concerning when the guidance clearly states that irrigation inputs should reflect long-term use not use over any one year.  

To promote consistency I would prefer to see irrigation inputs based on observable and verifiable evidence:

1. The hardware in use and whether it’s delivering to specs – e.g. evidence of commissioning certification, maintenance records, bucket tests etc.;

2. Whether active steps are taken to minimise overfill – e.g. whether application depth is known and matched with monitored deficit; and

3. Whether active steps are taken to maximise rainfall capture – e.g. links or reference to nearby weather station, system in place to actively incorporate weather forecasts.

Each of these is answered yes/no (or yes/no/partially) and links to a proxy hard-wired into Overseer that takes into account matters such as soil depth and climate pattern.

This would result in several outcomes:

1. Inputs that everyone understands, particularly those managing the irrigation system;

2. Inputs that can be verified on-site;

3. Consistency between modellers; and

4. Reductions in N loss are linked to real improvements in the field.

Selection of proxies that reflect each combination of system performance and soil/climate will be challenging but has less significance when Overseer is used correctly to compare rather than estimate absolute N loss.

Leo

Hi Leo, don't see an issue with your example, as long as the 3 modelers arrived to a similar outcome. In saying that, I'm not the regulator. :-)

I think your preference around irrigation inputs is being addressed at an auditor level, and shouldn't be necessary when doing a Year end NB yearly. When we do a NB, we trust the information provided by the farmer is accurate and represents the system. If not the case, there are consequences for the farmer during the audit.

Agree 100% with "1. Inputs that everyone understands..". FF, FV, VF and VV are not easy to understand or to read in Overseer. Ideally we need translation to common terminology (e.g. deficit irrigation, conservative irrigation, little and often, when water is available). Not very scientific though..

I'm a simple person and don't like to overly complicate any of my modelling when the outcome is likely to be the same so am of the opinion we need to keep this as simple as possible. If their is still confusion out there on how to model irrigation and how people are using the data from Overseer, then would it be best to provide more education as opposed to consider changes? Irrigation is a relatively simple concept but the complexity and variations in management on some farms makes it sometimes difficult to accurately represent the actual management within the model .

I totally agree with Leo's last comment 'less significance when Overseer is used correctly to compare rather than estimate absolute N loss.' I do realise that this is not necessarily happening, but at the very least this should be in the back of modellers/Council minds.

I agree with Marcelo - the three points in Leo's email are being addressed at an auditor level and shouldn't be necessary to complete an YE budget. These points are the justification for the modelling, as opposed to the data behind the modelling.

It is vital that all information is always entered based on consideration of the long term climate data used by Overseer and frightening to hear that some are trying to match irrigation outputs to actual water use - I thought this issue had been addressed many years ago.

I would also like to note that we have a Regional Council that is using the Overseer modelling to 'check' that water use within the model is close to or matches the consented annual volume, and where it doesn't the applicant is being asked to amend the modelling to reflect the annual volume!!. So perhaps Education for all users and Councils on the irrigation model is necessary, regardless of what changes are deemed necessary.

I find the four management options simple and easy to use: FF, VF, FV, VV and would not like to see limits on the number of changes within a year. I agree with Marcelo that perhaps some common terminology around these aspects would be beneficial.

In answer to the five key points that Alastair wants from this discussion -below are my thoughts and comment

- No longer show amount of water applied through irrigation - I find having the amount of water applied through irrigation is a good point for discussions with farmers on efficiency e.g. difference between K line and pivot based on their management. I would like to see this stay but once again perhaps another round of education would be useful.

- Could have some representation of soil moisture and drainage over time - I don't see this as necessary

- Only allow one management type per irrigator for the year - I don't agree

- For FF, limit number of changes that can be made in a year (maybe 4) - I don't agree

- Add more help and warnings to input screens - I definitely think more help and education is required based on some of the comments, along with perhaps some common terminology added to help the user better understand.