A story of whisky and water

The River Spey runs north east from its headwaters west of the Cairngorm mountains towards the Moray Firth. It is home to more than 50 whisky distilleries. The upper part of the Spey catchment contains 29 active distilleries. It is these, and the land surrounding them, which are the focus of this study.

The availability of a steady supply of good quality water is crucial to the process of whisky production. The analysis presented here compares long term average rainfall statistics to current production capacity per distillery in an effort to explore the relative vulnerability of each distillery under current climate conditions. As part of a wider  project  on the vulnerability of mountain value chains to climate change, we considered whether water availability might become a constraint on the increasing production of whisky in Speyside.

The following sections describe the data used, outline the main steps involved, and present some of the maps created as part of the EU Horizon 2020 MOVING project.

The primary data used comes from the following main sources:

1. Distillery locations - OpenStreetMap ^® 
2. Production capacity - a proxy for water demand measured in Litres of Pure Alcohol (LPA) - figures from 2021
3. Digital Terrain Model - OS Terrain 5 product from Ordnance Survey
4. Rainfall data - Met Office 1km gridded rainfall data (Long Term Average 1991-2018)
5. River network - OS Open Rivers

As part of the project the team were asked to apply a "vulnerability matrix" across the catchment. The aim was to compare water demand (using production volumes as a proxy) versus water supply (using long term average rainfall values) for each sub-catchment and to classify them into values of 1-5 following the logic below.

The idea was to generate an indication of the relative pressure of demand vs supply for each sub-catchment.

To achieve this a series of processing steps were followed which are briefly described below.

First the sub-catchments for each distillery were derived from the digital terrain model.

Second the rainfall data was integrated and the total volume for each sub-catchment calculated. Values were classified into high, medium, and low using a quantile classification (where roughly equal counts of sub-catchments appear in each category).

Third the demand per sub-catchment was calculated using production values as a proxy and each sub-catchment classified into high, medium, or low demand again using a quantile classification.

Finally the combination of the rainfall data and proxy demand data enabled the relative vulnerability of each sub-catchment to be calculated.

The following contains a brief technical description of the process of catchment derivation - also known as watershed derivation. General readers may wish to skip over this section and go straight to the  next section  or go straight to the  end result .  

To derive catchment areas, or watersheds, a series of processing steps were undertaken using the hydrology toolset in the Spatial Analyst toolbox in Esri’s ArcGIS for Desktop ArcMap v10.8.1.  

Catchment derivation consists of the following key stages:

1. Processing of the digital terrain model - a stage known as 'hydrologic conditioning'
2. Identification of abstraction points on the stream network
3. Creation of flow accumulation data
4. Watershed delineation

Firstly the digital terrain model was pre-processed to create a depressionless DEM. This is known as ‘hydrologic conditioning’ and removes any sinks which may be present in the elevation model. This ensures that every cell flows into another cell.

Then, where possible, for each distillery, a likely abstraction point was identified on the river network. This was done through interpretation of aerial photography where points were identified as close to the distillery as possible usually at a point slightly upstream of the distillery itself. 

Flow accumulation is the term used to describe the calculation of accumulated weight of all cells flowing into each downslope cell. Its purpose is part of the mechanism to work out how much rainfall has fallen on a catchment area.

With the abstraction points identified and the flow accumulation calculated, the last part of this stage was to create the watershed (or sub-catchment) boundaries. This was achieved successfully for 25 distilleries of the 29 distilleries. The map below shows these derived sub-catchments.

The image below shows the same derived sub-catchments data and is one of the print products generated as part of the work. The design is a little different and is intended to illustrate that there are distinct collections of sub-catchments into higher level groups. These are referred to as 'sets' here and are distinguished by their thicker outline.

With the catchment boundaries derived the next step was to work out the total volume of rainfall available in each catchment. To do this the 1km long term average data was intersected with the derived catchment boundaries.

With the data for supply and demand prepared the next stage was to classify the catchments into low, medium, and high categories. To do this a quantile classification was applied in each case which uses the data values to split the range into 3 classes with equal numbers in each class. This creates break points for low, medium and high driven by the values themselves.

The interactive web map version below shows the rainfall classified into the three categories. Since this is a measure of the total rainfall across each catchment, while there is some variation in rainfall across the catchment (see earlier) generally speaking the larger the catchment the higher the rainfall total.

The image below is a print version of the same data. This includes values for the derived break points.

In similar fashion the interactive webmap below shows the demand for each sub-catchment using the production capacity as a proxy indicator. This is slightly more complex due to the nested nature of the derived catchments - particularly in the River Fiddich - meaning that multiple distilleries may have to be taken into consideration when calculating the total demand in a catchment.

The image below is a print version of the same data including derived values for the breakpoints.

Bringing both supply and demand elements together and applying the classification seen in the earlier vulnerability matrix yields the interactive webmap below.

Finally, the print version of the same is shown below.

This approach illustrates how water supply (in the form of rainfall statistics) and total demand (via production capacity as a proxy) can be brought together to give an indication of relative resilience of distilleries to reductions in water availability. Water Stewardship is an important aspect of sustainability and distilleries are increasingly considering their reliance on natural capital.

It is important to note that the vulnerability matrix approach enables a relative rather than absolute comparison of the catchment demand vs water supply to be assessed. It is in no way meant to indicate that any particular distillery is under resource pressure. However, given the increasing production of whisky in the area, it is important to consider the potential pressures on water availability. The Scottish Environment Protection Agency licenses water abstraction volumes by distilleries and manages access to water under conditions of water scarcity. This screening analysis is designed to help the industry with their planning, rather than substitute for the regulatory and licensing process.

The approach could be improved upon in a number of ways.

From the demand side, detail could be integrated from other sources of water demand in the catchment - whether that is domestic consumption or other industrial extraction. It would also be worth exploring whether any data exists around absolute water quantities needed by each distillery to generate the final product. I.e. how much water does it take to create a litre of whisky?

From the supply side this analysis uses long term average observed rainfall data calculated per catchment. The approach could also be used for future modelled climate data on a similar basis. Alternatively it may be possible to model known observed dry years, or dry seasons, where low flows have already created operational difficulties for production to get a sense of at what level these events occur. That would allow us to answer questions such as "how dry does it need to be before production becomes threatened?" and "how much more likely are those conditions to occur in future climate scenarios?" Later research undertaken by the CREW project  "Future Predictions of Water Scarcity in Scotland: Impacts to Distilleries and Agricultural Abstractors"  starts to answer these questions.

This analysis uses surface water volumes in assessing the vulnerability of sub-catchments. However we know that surface water is not the only source of water used by distilleries. Groundwater is also used. The British Geological Survey record details of  boreholes  including whether or not they are intended for water. We know from these records that some distilleries have taken steps to secure supply via this route such as Macallan, Cardhu and more recently The Cairn. Water quantities, if measured, from these alternative sources could potentially be integrated into the analysis. Though groundwater supply is not well understood, where there is low recharge and low aquifer storage, the ability for groundwater to buffer surface water scarcity will be reduced.

Finally, the focus of this analysis was limited to surface water volume versus demand. We acknowledge this is a simplistic view which does not include a number of other factors critical to whisky production. With regard to other properties, water temperature may be just as important. It would be worth exploring to what degree increasing water temperatures may make impact activities when used as a cooling mechanism in production and whether any longer term water temperature records exist which might be incorporated into any future work.