Data

The SACO tool is based around the Water Resources GIS (WRGIS) conceptualisation and tables. Here we provide some further background on WRGIS and its use in the SACO tool.

Note

WRGIS data may be made available under licence following a data request to the Environment Agency. Please contact sustainable.abstraction@environment-agency.gov.uk for further information.

WRGIS

WRGIS summarises information on natural flows and the impacts of artificial influences on these natural flows. WRGIS provides this information for selected percentiles of the flow duration curve (FDC) for all Water Framework Directive (WFD) surface waterbodies in England. These percentiles are (lowest flow to highest flow): 95, 70, 50, 30. The WRGIS database does not contain time series information.

Impacts are provided for three artificial influence scenarios, which may be broadly defined as:

  • Recent actual: mean impacts over a recent ~6-year period

  • Fully licensed: hypothetical impacts if all abstractions were to take place at their maximum rates (according to licence conditions)

  • Future predicted: recent actual impacts multiplied by a growth factor to estimate potential changes

Artificial influences are primarily surface water abstractions, groundwater abstractions, and (surface water) discharges. However, a fourth class of “complex” influences is also delineated in WRGIS. One example of a complex impact might be the impounding effect of a reservoir on a natural flow regime. This impact does not necessarily fall easily into the simpler categories of abstractions and discharges. More discussion on this class of influences is given below.

WRGIS is constructed by a set of tools that process information from the Catchment Management Strategy (CAMS) “ledgers”. These ledgers describe the impacts of artificial influences on the FDC at so-called CAMS Assessment Points (APs). The WRGIS tools translate the information on natural flows and artificial influences from the ledgers to the WFD waterbody scale. This processing results in a (geo)database. When referring to WRGIS throughout the SACO package/documentation, we are referring to the database rather than the processing toolset.

Tables

The data tables in the WRGIS database that are relevant to the SACO tool are listed in the summary table below:

Table Name

Notes

AbsSensBands_NBB

Abstraction sensitivity bands (ASBs) per waterbody

ASBPercentages

Permitted deviations (fractional) from natural flow by ASB and flow percentile

Discharges_NBB

Discharges

GWABs_NBB

Groundwater abstractions

IntegratedWBs_NBB

Defines waterbody network and waterbody properties

QNaturalFlows_NBB

Natural flows per waterbody

REFS_NBB

Reference flows (typically environmental flow indicator)

Seasonal_Lookup

Percentile impact factors for surface water abstractions

SupResGW_NBB

Complex impacts

SWABS_NBB

Surface water abstractions

This summary indicates that the SACO tool uses four tables that are indexed by waterbody (i.e. one row per waterbody): AbsSensBands_NBB, IntegratedWBs_NBB, QNaturalFlows_NBB and REFS_NBB. In contrast, the four tables of artificial influences are indexed by “point”. For example, a given abstraction might have multiple point locations associated with it. The impact of each of these points is entered on one row in the relevant table.

The remaining two tables are essentially supporting metadata. ASBPercentages facilitates calculation of the reference flow in REFS_NBB (typically the environmental flow indicator, EFI) for each waterbody. Seasonal_Lookup contains factors that relate long-term average surface water abstraction to impacts at a given flow percentile based on the seasonality of abstraction.

All tables except ASBPercentages have a “wide” format, i.e. they may contain multiple value columns and no “factor” columns. For example, the abstractions tables contain a separate value column for the impact on flow at each combination of scenario and percentile.

Note

An abstraction licence may also include multiple “purpose codes” that define the permitted reason for abstraction. The abstraction tables are therefore technically given on a licence-point-purpose basis.

Note

Some artificial influences are highly complex, so care (and often local expertise) is needed when using and interpreting the various tables in WRGIS.

Conceptualisation

Here we summarise a few relevant aspects of the WRGIS conceptual model:

  • Individual surface water abstraction points, (surface water) discharge points and complex influence points can impact one waterbody.

  • Discharges do not vary as a function of percentile or scenario (RA and FL only) in general. Values tend to represent dry weather flows or consented discharges, but this may vary between discharges and areas.

  • Complex impacts may be positive (increasing flow) or negative (decreasing flow).

  • Groundwater abstractions may impact between one and five waterbodies according to a fixed set of proportions (summing to 100%). These proportions are constant across flow percentiles and abstraction scenarios.

  • Total artificial influence impacts are disaggregated across the FDC (selected percentiles) with reference to known/modelled seasonality.

  • Scenario flows are defined as the flows resulting after artificial influences have been applied to natural flows. These are effectively denaturalised flows.

  • Scenario flows are calculated through a simple water balance for a given combination of artificial influences scenario (RA, FL or FP) and percentile (95, 70, 50, 30).

  • This is equivalent to an instantaneous propagation of impacts through the network of waterbodies. I.e. there are no time-dependent considerations in the “routing” of flows and impacts. Implicitly, all waterbodies thus experience the same flow percentile at the same time.

Limitations

The SACO tool operates in “WRGIS world” and explicitly seeks to preserve the “rules” of this world. Yet, as a national data product, WRGIS encodes a number of simplifications and assumptions, including those outlined in the previous section. These features are inherited by the SACO tool. It is also bound by limitations of the data, i.e. the uncertainties, approximations and errors involved in translating observed or estimated quantities into the WRGIS database.

It is important to note that the SACO tool is “downstream” of WRGIS. This means that it takes WRGIS tables as inputs. One implication of this is that the SACO tool is currently “downstream” of the process that assigns an impact to each artificial influence at each flow percentile to create the “base” WRGIS. (As noted above, this assignment is undertaken by the WRGIS toolset using information from the CAMS ledgers.) The SACO tool can explore the implications of changing specific numbers in the tables (via the Calculator and Optimiser components outlined in the Overview), but it does not contain all the logic, data and justification through which “base” impacts are assigned in the ledgers and WRGIS.

Note

Functionality has been added to allow estimation of (changes in) long-term average abstraction based on (changes) in impacts at a given percentile. See infer_mean_abstraction method entry in Dataset. Additional functionality is being added to make the reverse translation (long-term average to impacts at a given percentile) and also to reproduce in general how WRGIS derives the impacts of surface water abstractions with hands-off flow conditions / reservoir support flows (complex impacts of type QMIN).

Processing

Functions for processing the raw data tables from the WRGIS database (currently in Access format) into the format required by SACO are not included within the SACO package currently. To summarise this functionality, the main processing steps are:

  • Extract tables and perform basic checks of indexes and important columns. Columns not required in the SACO tool are filtered out.

  • Convert waterbody relationships into a directed graph (networkx.DiGraph). This provides useful helper functions/methods for working with the network.

  • Calculate other derived quantities for convenience, including reference flows (typically EFIs) for each waterbody and flow percentile (as a function of abstraction sensitivity band, which defines a permitted fractional deviation from the natural flow).

The processing code writes a set of output files to a specified directory. The data tables are generally saved in parquet format, the network graph is saved in graphml format, and a numpy helper array (routing matrix) is saved in compressed npz format. A directory containing these files forms the main input to the SACO tool.

Note

EFIs are calculated as per WRGIS. This is consistent with the CAMS ledgers at the 95th flow percentile, but some (typically small) divergence is possible at higher flow percentiles. The calculations in the CAMS ledgers use additional data that are not yet readily available at the waterbody scale. However, work is underway to harmonise the WRGIS/SACO method of EFI calculation with the CAMS ledger method for full consistency above the 95th percentile.

Further Details

See the Table Fields and Tutorial pages for more explanation of each of the tables involved in SACO input/output. Synthetic examples of the tables/data (with no relationship to any real waterbodies or artificial influences) are available in the repository (under the examples directory, as well as in tests/data). These examples can be read in and viewed using the notebook in the examples folder of the repository.