Water use maps and water use related information

This chapter describes the main features of time variant (transient) sectoral water demand maps for domestic, livestock, industrial, energy (cooling) sectors. This chapter then focuses on the description of the ancillary maps required by OS LISFLOOD for the modelling of water abstraction from groundwater, surface water (channels, lakes, reservoirs), and non-conventional sources (e.g. desalination plants). In this documentation, water demand is the amount of water required to meet the needs of the various uses. Water abstraction or water withdrawal (considered synonyms) is the amount of water to be abstracted from surface or groundwater resources to meet water demand requirements. Water abstraction/withdrawal is generally larger than water demand to account for leakages and other losses in the water supply system. However, if the demand cannot be met, withdrawals are lower than the demand. Consumptive use is the actual water volume used and removed from the hydrological cycle.

Sectoral water demand maps

Sectoral water demand maps indicate, for each pixel, the time-varying water demand value to supply for domestic, livestock, industrial, and thermoelectric water consumption. The segregation of the total water demand for anthropogenic use into four main sectors, namely domestic, energy, industrial and livestock water demand, enables a more accurate representation of the processes and follows the Food and Agriculture Organisation of the United Nations (FAO) terminology (Kohli et al., 2012). Domestic water demand represents indoor and outdoor household water use as well as other uses (e.g. industrial and urban agriculture) connected to the municipal system (e.g. water use by shops, schools and public buildings). Electricity (energy) water demand is the water use for the cooling of thermoelectric and nuclear power plants. Water demand for industry is the water used for fabricating, processing, washing, cooling or transporting products and also includes water within the final products and water used for sanitation within the manufacturing facility. Livestock demand is the water used for drinking and cleaning purposes of livestock (Choulga et al. 2024).

The temporal discretization of these maps (e.g. daily, monthly, yearly update frequency) can be chosen by the modeller, mainly depending on the modelling purposes and on the available input data. Values must be expressed in $[\frac{mm}{day}]$, for any update frequency and for any modelling computational step. In case of frequencies larger than one day (e.g. monthly demand updates) is described in this section of the model documentation. Similarly to the meteorological forcings, OS LISFLOOD internally adjusts daily input values to the sub-daily modelling step.

European 1 arcmin and global 3 arcmin sectoral water demand maps were generated using the OS LISFLOOD utility water-demand-historical. This user guide provides an overview of the protocol implemented to produce domestic and energy demand maps with values updated monthly, as well as industrial and livestock demand maps with values updated yearly. The generation of the maps relies on a number of external datasets: the complete list of external datasets and step-wise instructions are provided in the readme of the OS LISFLOOD utility water-demand-historical. FAO AQUASTAT constitutes the main source of information: specifically, country level data of water withdrawal. Therefore, strictly speaking, these maps represent water withdrawal rather than water demand. As explained in this page, this discrepancy can be accounted for by adequately setting ancillary OS LISFLOOD inputs such as $leakage fraction$.

Domestic water withdrawal estimates rely on FAO AQUASTAT municipal water withdrawal estimate per country and per year. Spatial disaggregation is achieved based on population data (Global Human Settlement Layer). Temporal downscaling to monthly frequency relies on monthly air temperature grids (e.g. MSWX, Beck et al., 2022) and literature parameters (Huang et al., 2018)

Industrial (yearly) and energy (monthly) water withdrawal estimates are based on country-scale FAO AQUASTAT industrial withdrawal data, country-scale World Bank manufacturing value added (MVA) data, Global Change Analysis Model, GCAM regional industry and thermoelectric withdrawals, datasets available in the literature (e.g. Global-scale gridded estimates of thermoelectric power and manufacturing water use by Vassolo & Doll, 2005). Spatial downscaling is based on population grids (due to the lack of more precise data sources). Temporal downscaling of energy water withdrawal estimates rely on temperature information.

Country-scale livestock withdrawals are based on country-level FAO AQUASTAT data (difference between agricolture and irrigation withdrawals, where available) and on regional GCAM livestock withdrawals (where FAO AQUASTAT data are not available). Spatial downscaling of regional GCAM livestock withdrawals to country level is performed based on country values provided in Gridded Livestock of the World.

The protocol for the generation of water withdrawal maps relied on a number of assumptions due to the lack of homogenous and granular data at the continental and global scale. Even at the country scale, information is not available for all countries, and for all years. Gaps in space and time were filled using nearest-neighbor interpolation and linear interpolation, respectively.

Groundwater bodies

The map of groundwater bodies is a boolean map indicating the spatial distribution of groundwater exploitable resources: OS LISFLOOD allows abstraction from the lower groundwater zone only in pixels with values set to 1.

This map can be generated using local, regional, continental, or global scale source data, depending on the specific OS LISFLOOD application.

Examples of continental scale source data are (BGR & UNESCO) and Africa Groundwater Atlas (BGS).

The Groundwater Resources of the World of the World-wide Hydrogeological Mapping and Assessment Programme (WHYMAP) is an example of source data for the global domain.

Datasets such as Groundwater Resources of the World and the global map of bedrock elevation GDEMM2024 could be used as source data of the Global 3 arcmin groundwater bodies map. GDEMM2024 was selected as source of the current version of the Global 3 arcmin groundwater bodies map to ensure the consistency between the global map of groundwater bodies and the available information on country-level groundwater abstraction (i.e. all countries with non-zero values of groundwater abstraction included at least one pixel classified as groundwater body)

The European 1 arcmin groundwater bodies map was generated using the International Hydrogeological Map of Europe, IHME1500 v 1.2 as main data source. Specifically, groundwater resources were deemed available for the areas belonging to the following classes: I. Predominantly porous rocks; II. Predominantly fissured rocks, including karstified rocks; IIIa. Locally aquiferous rocks. This approach satisfied the cross-check with available information on country level water abstraction from groundwater (for details, please see Sectoral water demand maps). Groundwater bodies of areas of the European extended domain (as defined in the introduction to the static maps) not covered by IHME1500 were derived from the global map of groundwater bodies. To avoid abrupt discontinuities, the two datasets were merged at the country level.

Fraction of water abstraction from groundwater, surface water, non-conventional resources

OS LISFLOOD allows water abstraction from groundwater, surface water, non-conventional sources (e.g. desalinization plants). The maps fraction of water abstraction from groundwater (fracgroundwateruse) and fraction of water abstraction from non-conventional sources (fracnonconventionalwateruse) provide information on the proportion of the total water demand that must be provided by groundwater resources and non-conventional water sources, respectively.

Information is provided to the OS LISFLOOD code at the pixel level. However, the actual granularity of the map depends on the spatial resolution of the model exercise and on the available data. Generally speaking, spatial allocation of water demand to the three sources should be defined at the water region level.

The fraction of water to be abstracted from groundwater, $fracgroundwateruse$, is computed in two subsequent steps.

The first step requires the computation of $fracgroundwateruse$ based on the ratio of water withdrawal from groundwater to total water withdrawal, both quantities are aggregated values over the same spatial domain (e.g. water region, country).

\[fracgroundwateruse_1 = \frac{water withdrawal from groundwater}{total water withdrawal}\]

$fracgroundwateruse_1$ represents then the average value for the chosen spatial domain. Pixel values of $fracgroundwateruse$ must account for the proportion of exploitable groundwater resources within the spatial domain: $fracgroundwateruse$ must be zero in pixels with no exploitable groundwater resources, and fraction values in the remaining pixels must be adjusted accordingly. The second step is then defined as follows:

\[fracgroundwateruse = min(( fracgroundwateruse_1 * \frac{area_T}{area_G} * groundwaterbodies), 1-fracnonconventionalwateruse)\]

$groundwaterbodies$ is the boolean with 1 values where groundwater is available. $area_T$ and $area_G$ are the total area of the spatial domain and the area with available groundwater within the spatial domain, respectively.

$fracnonconventionalwateruse$ is the fraction of water derived from non-conventional water sources (e.g. desalination) within the spatial domain. It is computed as the ratio of water withdrawal from groundwater and total water withdrawal, as before, both quantities are aggregated values over the same spatial domain (e.g. water region, country).

\[fracnonconventionalwateruse = \frac{water withdrawal from non conventional sources}{total water withdrawal}\]

Finally, the proportion of water demand to be satisfied by surface water resources is computed internally by OS LISFLOOD as:

\[fracsurfacewateruse = 1 - (fracgroundwateruse + fracnonconventionalwateruse)\]

Information of water withdrawal from the three sources must be retrieved from external datasets.

European 1 arcmin and global 3 arcmin maps were generated leveraging on country level information made available from FAO AQUASTAT. Data were retrieved in 2024: the most recent information at that time referred to the year 2021. The database provides information at the country level, nevertheless, information is not available for all countries, and different variables are available for different countries. The implemented methdology aimed to maximize the exploitation of available information. Total water withdrawal was either directly provided or defined as the sum of the other available information (surface water abstraction, groundwater abstraction, desalinated water, treated waste water, reuse of drained water from agricluture, https://www.fao.org/aquastat/en/databases/glossary/). Groundwater withdrawal was either directly provided or computed as the complementary of the sum of the other variables. Despite the attempt to maximize the exploitation of FAO AQUASTAT information, fraction of groundwater use could not be computed for some countries. The filling was done based on a nearest neighbour approach. To avoid enforcing “extreme” scenarios to countries with no info, only values in the interval [0.15, 0.85] were transferred from the closest neighbors. If the three closest countries did not have information or had values outside of the interval [0.15, 0.85], the fill value was 0.5. When using FAO AQUASTAT database, water withdrawal from non conventional sources was defined equal to the variable ‘desalinated water produced’. Countries where the latter variable was not available were allocated 0 value of $fractionofnonconventional$ water use. It is here noted that this water source is “outside” of the hydrological cycle modelled by LISFLOOD (oceans are not modelled!). Desalination plants processing water of lakes in landlocked countries are, on the contrary, retrieving water from the hydrological cycle modelled by LISFLOOD: to be consistent with the model implementation, the latter quantity did not contribute to the computation of $fractionofnonconventional$ water use, but were added to the abstraction from surface water bodies.

Clearly, leveraging on country level information leads to changes along the country borders which are generally not consistent with the physical groundwater and surface water basins. Where possible, it is recommended to use information at the basin or water region scale.

Fraction of consumptive water use

The fraction of consumptive water use defines the portion of water abstraction which is consumed and leaves the hydrological cycle. It applies to domestic, livestock, energy (cooling), and industrial uses; each fraction can be set independently as a constant for the entire modelling domain (by introducing the desired values in the OS LISFLOOD settings file) or as a map including more granular information, ideally segregated according to water regions.

Fractions of consumptive water use maps are currently not available for the European 1 arcmin and global 3 arcmin domains. Constant value were retrieved from the report of Bisselink et al, 2018: 0.2 domestic consumptive use, 0.15 livestock consumptive use, 0.17 energy consumptive use, 0.15 industrial consumptive use.

Irrigation efficiency, irrigation conveyance efficiency, irrigation multiplier

Water demand for irrigation is computed internally by the code according to soil moisture deficit and crop type. Irrigation efficiency, irrigation conveyance efficiency, irrigation multiplier are used to adjust the water demand to compute the volume to be abstracted for irrigation. This value, included between 0 and 1, can be a constant over the entire modelling domain (by introducing the desired values in the OS LISFLOOD settings file) or a map. The European 1 arcmin domain makes use of a map with data at the level of NUTS2 regions and information retrieved from De Roo et al, 2020 (Figure 3, Benitez et al, 2018). The global 3 arcmin domain currently makes use of 0.75 constant value, as indicated in Bisselink et al, 2018. Conveyance efficiency depends on how water is delivered to the fields, it is always included between 0 and 1, the lower the value, the larger the water abstraction for irrigation. It can be a constant value or a map. European 1 arcmin and global 3arcmin domains currently use 0.8 constant value, as indicated in Bisselink et al, 2018. Irrigation multiplier is a factor larger than 1 applied to increase water demand for irrigation, to mimic the additional water abstraction required to prevent soil salinitization. The currently implemented value is 1.2.

Domestic leakage fraction and domestic water saving fraction

Domestic leakage fraction is used to account for the leakages from pipes in urban water supply systems: this value must be lower than 1 and 0 indicates absence of leakages. Within the code, domestic leakage fraction values larger than 0 are used to compute a multiplier (>1) of water demand for domestic use to compute water abstraction for domestic use. Conversely, domestic water saving fraction reduces water demand, and, consequently, water abstraction. For both inputs, OS LISFLOOD accepts a constant value or a map. The reports Bisselink et al, 2018 and De Roo et al, 2020 provide information on domestic leakage and water saving fraction, respectively. In the current European 1 arcmin and global 3 arcmin set-ups,domestic leakage fraction is set to 0: water demand maps are computed leveraging on water withdrawal values reported by FAO AQUASTAT[https://www.fao.org/aquastat/en/], which should, by definition, already account for leakages.

Environmental flow

Environmental flow is defined as the amount of water which should be always present in a river to ensure the survival (or well-being) of the aquatic ecosystem. In OS LISFLOOD, Environmental Flow is a lower threshold: water abstraction for the channel stops when discharge is lower than such a threshold. OS LISFLOOD accepts either a constant value or a map as input. These values could be defined by water management plans of by statistical analysis (e.g. 10th percentile of a naturalized simulation, i.e. a simulation without water use, lakes, reservoirs). It is here noted that OS LISFLOOD equally ensures a minimum water volume in lakes and reservoirs: these minimum values are set internally by the code, as explained in the Water Use chapter.

Water regions map

As the spatial resolution of the model increases, the assumption of coincidence between demand and abstraction locations within the same model grid cell becomes increasingly invalid. To address this limitation, the concept of water regions is introduced. A water region is defined as a subcatchment where demand and abstraction activities occur, allowing for a more accurate representation of the spatial relationships between these processes. Water regions are generally defined by sub-river-basins within a Country. In order to mimic reality, it is advisable to avoid cross-Country-border abstractions. Whenever information is available, it is strongly recommended to align the water regions with the actual areas managed by water management authorities, such as regional water boards. In Europe, the River Basin Districts, as defined in the Water Framework Directive and subdivided by country, can be used.

Consistency between water region map and model calibration protocol

Water resources (surface water bodies and groundwater) are shared inside the water region in order to meet the cumulative requirements of the entire water region area. For this reason, it is strongly recommended to include the entire water region(s) in the modelled area. If a portion of the water region is not included in the modelled area, then LISFLOOD cannot adequately compute the water demand and abstraction (it is important to notice that LISFLOOD will not crash but the results will be affected by this discrepancy).

The inclusion of the complete water region in the computational domain becomes compulsory when performing catchment-based calibration, where parameters are optimized separately for each catchment inside the larger computational domain. In this case, calibrated parameters are optimised for a specific model domain. Each calibration domain must include a finite number of water regions (and each water region must be entirely included in one catchment). If and only if this condition is met, the calibrated parameters can be correctly optimised. Conversely, when a water region belongs to one or more calibration catchments, the water resources are allocated and abstracted in different quantities during calibration as opposed to the modelling of the entire basin.

The utility waterregions can be used to 1) verify the consistency between calibration catchments and water regions or 2) create a water region map which is consistent with a set of calibration points.

Current European 1arcmin and global 3arcmin water regions map were generated to meet the requirement explained above. The list of calibration points, a basemap of the major surface water basins, and the map of country borders were used as input to the OS LISFLOOD utility in the case of the European 1arcmin domain. The list of calibration points and the map of country borders were used as input in the case of the global 3arcmin domain.