Setting up a Grape Forecast

After setting up all Sampling Data which can be made up of either Bunch or Segment samples as well as Historical Samplings, go to the Forecasting Analysis tab and click the Add button.

Next ensure you have the appropriate vintage and lot selected on the Forecast screen and also enter a description that makes sense or is relevant.

Grape Forecaster can calculate forecasts of grape deliveries using a range of different formulae. The particular formula that Grape Forecaster eventually uses to make a forecast is built from “components” that you choose step by step.

For example, this relatively simple formula consists of three components, which are multiplied to make a forecast: 

Metres of row in the Lot X Weight per metre of row X Harvest efficiency = Weight delivered to the winery from the Lot

The number of metres of row in the selected Lot is displayed in the top left corner of the “Forecasts” window. Grape Forecaster calculates this for you from the Patch and Row information that you entered when you set up your Vineyards under Configuration / Vineyards, Patches and Rows …

To calculate an estimate of the total weight of grapes in the Lot at the time of harvest, Grape Forecaster needs to know how you want it to calculate an estimate of mean weight per metre of row. If you have picked segments near harvest and entered “Kilograms” data either in the Add/Edit data window or by using Grape Forecaster Module on agCommander Mobile App, Grape Forecaster can calculate this directly, allowing for the segment length. Alternatively, you may need to build an estimate of weight per metre at harvest from other components. This would be the case at any stage earlier in the season.

Building a forecast

Select the drop-down “Base yield estimation on” in the “Weight per row metre” component area. You will see a drop down list of options for calculating weight/metre.

The option at the top of the list will be “Yield components”. Building an estimate of weight/metre from yield components is dealt with below.

If you have picked segments near harvest and entered the “Kilograms” data, Grape Forecaster can calculate weight/metre directly, allowing for the segment length, and will present a list of sampling dates. Commonly there would only be one date, but there could be more if “Kilograms” data from more than one sampling is available. 

Estimating Bunches Per Metre

Select thew drop-down. You will see a list of Segment samplings for which “Bunches” data is available. 

Estimating Bunch Gain or Loss

When you have selected a sampling to provide an estimate of Bunches/metre, the next component area, “Bunch Gain/Loss Factor”, will appear. This offers a list of options like the ones described for Harvest Efficiency (see above).

When you have selected an option for estimating the Bunch Gain/Loss Factor the Value will be displayed, together with the upper and lower limits of Doubt. 

Estimating Weight Per Bunch

When you have estimated a Bunch Gain/Loss Factor, the next component area, “Weight/Bunch”, will appear. This offers a list of options. 

Estimates of Weight/bunch at harvest can be be based on:

• A manual estimate
• A value from a previous year
• An historical statistic
• An estimate of mean weight/bunch derived from a sampling in the current season adjusted for weight gain between the time of sampling and harvest
• An estimate of mean berries/bunch derived from a sampling in the current season adjusted for berry loss between the time of sampling and harvest and multiplied by an estimate of harvest weight/berry. 

Estimating harvest weight per bunch from in-season weight per bunch

Grape Forecaster allows you to estimate harvest weight per bunch from a mean weight per bunch of a bunch sample. To do this you also need to predict a “Weight gain” factor.

The options in the “Estimate from” list in the “Weight/bunch” component area that start with “Bun.Wt.…” refer to samplings from which a mean weight per bunch can be calculated. A mean weight per bunch can be calculated in two ways:

1. from a sample of segments if both “Bunches” and “Kilograms” data is available

2. from a sample of bunches for which “Bunch grams” data is available 

Berry Loss Factor

This is estimated number of berries lost via factors such as:

• Low crop-loads which cause vines to become unbalanced due to heavier fruit.
• Severe Water stress that can cause emergency ripening leading to lower quality in the fruit.
• Lack of adequate nutrition to enable optimal synthesis of enzymes causing fruit quality to drop.
• High heat loads that can cause berries to shrivel up due to water loss and evaporation.

Estimating Harvest Efficiency

Harvest Efficiency is always the last component to be estimated in a forecasting formula. When a value for it has been set, Grape Forecaster will calculate a forecast and display it in the “Most likely Yield and Delivery” area at the bottom of the “Forecasts” window.

The Delivery is the weight (tonnes) that will be delivered to the winery from the Lot. The Yield is the Delivery divided by the area of the Lot, expressed as tonnes/hectare.

Click on the symbol at the right of the “Estimate from” box in the “Harvest Efficiency” component area. You will see a list of options for sources of an estimate of harvest efficiency. The composition of this list will vary according to the availability of historical data for the selected Lot.

The “Manual” option will always be at the top of the list, and may be the only option available if there is no historical data. 

Estimating Harvest Efficiency manually

If you select the “Manual” option to set Harvest Efficiency, you will see this window:

Defaults are initially displayed for the Value that is to be used and the Limits to the range of Doubt above and below the Value. All of the numbers in boxes with “spin buttons” next to them can be set manually. Grape Forecaster will not allow you to set an upper limit lower than the Value, or a lower limit higher than the value. As you adjust the Value and the Limits to the range of Doubt, the Doubt % will recalculate. You can have a different Doubt % above and below the Value.

For other options:

For each component of a forecasting formula that is a statistic derived from historical data, the following options may be available:

If 2 or more years of historical data are available, these statistics will be available:

• Historical mean
• Historical maximum
• Historical minimum
• Above average (mean + standard deviation)
• Below average (mean – standard deviation)

The mean, maximum, minimum and standard deviation are calculated from the sample means for all available years. 

Historical Sampling

To make historical data available for forecasting, you must go to the Sampling Data -> Historical Samplings tabs and click add to create new Historical samplings and designate which samples are to be used for calculation of historical statistics for each Lot in each Vintage. 

With Historical Types, there are five:

• Bunches per segment
• Grams per bunch
• Berries per bunch
• Grams per berry
• Kilograms per segment