Goal: Have data for the right spatial scale, corresponding to the system boundary. (The overall goal of downscaling data is to fill and close data gaps, so that ultimately the Sankey diagram can be filled and indicators calculated.)

Problem: Very frequently, it is difficult to find (all) material flow data for a city or sector.


One common way to bridge this gap is to downscale, which means to scale it down from a higher to a lower (spatial) level. This can be done with three components:

  • the (top-down) data from larger territorial scales (region, province, country),
  • a proxy (see Proxies)
  • and a downscaling formula for the smaller territorial scale (city, neighbourhood).

Tip 1: We encourage you to process your local data on the city scale first. It will help you to understand what you already have locally and downscaling calculations will not be required for those.

Tip 2: Focus on data that will fill your data gaps, so for parts of your data collection where you do not have any data yet at all.

Tip 3: As a rule of thumb, downscaling from the smallest (spatial) scale possible to the city should always be prioritised. For example, NUTS2 or 3 are better than the country level for downscaling because those levels are already low, which makes the data more relevant and precise.

Outline of the video

  • Downscaling is the way to get data if you don’t have data on your own local scale and you calculate data from a higher spatial scale to approximate it for your city.
  • From the data gap analysis, you probably have identified a number of gaps, in a number of places, lifecycle stages or years. If there is no local data for stocks and flows, then we need to look to higher or larger spatial scales, like region, NUTS, country etc. and downscale, which means to scale it down to the city level with proxies;
  • The approach of this module is a bit different: It is not very sensible that we present a number of good downscaling means, if you won’t actually use or need them. Therefore, we ask you to let us know which data downscaling you need help with and we will support you in that. (Note: While we had stated that during the course where fewer information was available, we have now aimed to better explain means and proxies to you here and believe that you will be able to make first calculations without help.)
  • In this video we are giving you some broad tips and tricks. We’ve also summarised them in the handbook.
  • M3:30, Process for downscaling
  • M6:14, Overview of proxies that can be used by lifecycle stage and in order of preference.
  • M7:27, the general formula to downscale data.
  • M11:00, examples and data sources

Process for downscaling

After identifying the data gaps (in the case of the SCA with the Data gap analysis, other data needs to be used to bridge the gaps. Ideally, in the data gap analysis, this additional data was identified, e.g. Eurostat data can be used to fill the gaps.

  1. Determine for which lifecycle stage (LCS) or material (group) the downscaling calculation needs to be made.
  2. Refer to the tables in the Proxies part to determine which proxy should be used, depending on the nodes / lifecycle stages.
  3. Collect the respective proxy data, for example on “persons employed”.
  4. Calculate with the proxies and the downscaling formula (see below). This does NOT happen on the Data Hub. Instead, it is recommended that you have different excel sheets or tabs per lifecycle stage for the calculations.
  5. Perform a sanity check, in order to be able to gauge whether results are logical. For example, you can compare the data aggregated to the groups (MF1-4) to the results of the downscaled national data, multiplied by the local population. See here the results for every CityLoops city
  6. (Document this (in a single online spreadsheet), like a master document. This can later on be expanded to include the indicator calculations and Sankey diagram as well.)
    • Have a column for years (for instance, 1995, 2000, 2005, 2010, 2015, 2016, 2017, 2018, 2019, 2020).
    • Tabs for calculations of each LCS (workplace with notes, assumptions etc.)
  7. Upload this document to “Layer 3.1 Extraction/Harvesting” on the Data Hub. There is no need to process it. Single (added or otherwise calculated) values of them will be taken from the document for the indicators and Sankey diagram.

Downscaling formula

The general formula to downscale data is always:

Quantity per city for the year ABC = Quantity of country / country proxy year X * city proxy year Y

(Country can be replaced by NUTS3 or NUTS2)


If there is no city level data for imports and exports there are different ways to extract this data.

Example 1: Using employment by industries to downscale national data.

This approach was used for Apeldoorn and Bodo, where we used studies using economic activity by sector as a downscaling proxy. Employees by sectors with “NACE 2” levels (*Second level is necessary for Mayer et al. calculations) are used for downscaling.

We gathered information of employees by industries with SBI, SIC codes (similar to NACE codes) for city level and the national level and performed matchmaking.

You can also use the chart we have created to match employees by industries with NACE-CPA codes to MF 2nd level (Source > go to the tab Matching NACE - CPA - MF):

Example 2: Using freight by group of goods.

This approach should include every transportation mode (maritime+rail+road+aviation), not just road freight information. Please see below the conversion table for NST07 codes to MF codes (Source > go to the tab Matching codes_NST). However this will only give you the main group values, while second level MF information is needed to perform Mayer et al. calculations. If you have this information on a different scale, you can use the economic activity again to downscale it.