Catchment-Scale Modelling

In the study of catchment hydrology, it is not possible to measure every part of the hydrological system. This is due to the high heterogeneity of the catchment variables in space and time. Hydrological models are simplified depictions of reality that use mathematical equations to represent hydrological processes that occur in time and space. Models are used to test scientific hypotheses when there is only a small amount or no empirical data.

They can be used to predict or forecast how a catchment will behave in the future or how a catchment will respond under a certain set of conditions.

Models can help further scientific understanding of dominant processes, catchment behaviour at different scales and in the identification of hot spot areas within catchments which can be useful for decision making (Beven, 2012). However, all model predictions are subject to uncertainty that is associated with two main things (Beven, 2012). The first is our limited knowledge of how hydrological systems work and the second is associated with the limitations in our current measurement techniques, which lead to errors in the observed data that are subsequently input into the model.

Models follow either a bottom-up/mechanistic approach which is based on an understanding at a point scale that it upscaled to the catchment, or a top-down approach which looks at catchment scale data and top-downscales it to calculate small scale processes. The bottom-up approach is the most widely used in hydrologic sciences due to its physical basis. However, there are concerns that the equations used in this approach, such as Darcy’s law for flow of fluid through a porous medium are based on small scale observations/laboratory work which may not accurately represent how water

moves over/through the soil at larger scales (Beven, 2012). There are numerous studies on the impacts of land use management on catchment hydrology. Some of the key studies are listed in Table 2.3.

Table 2.3: Modelling of land-use change (LUC) impacts on catchment hydrology.

Reference/Country Model used Purpose LUC scenario Runoff change

(Nandakumar and Mein, 1997) Victoria, south-east Australia

HYDROLOG -

Physical-semi-distributed conceptual, daily version of Monash model

Quantified levels of uncertainty in predictions due to errors in

hydrological/meteorological data and the implications for LUC prediction

- 12-43% deforestation Channel not represented

(Sefton and Howarth, 1998) England and Wales

IHACRES - Lumped conceptual Develop method to estimate model parameters from catchment

characteristics, to eliminate calibration

- 33% change from grassland to woodland - 33% change from grassland to crops

- Reduction in low flows - Increase in flow volume

(Bormann, Diekkrüger and Renschler, 1999) North-west Germany

Coupled SIMULAT

KINematic Runoff and EROSion model (KINEROS) - Physical coupled 1D Soil Vegetation Atmosphere Transfer (SVAT)

Investigate the effects of LUC on runoff, due to European Commission policy in Neuenkirchen catchment

- 3% increase in grassland, 15% winter wheat & 12%

decrease in winter barley - 12% increase in bare fallow

- Minimum tillage practices - Re-meandering channel

- 145% decrease groundwater recharge with winter cover crop

- 30% increase peak discharge - 8 to 34% decrease peak discharge

- 63% reduced peak discharge (Lukey et al., 2000)

South-east France

SHETRAN Study of the impacts of afforestation on

streamflow in Draix catchment (0.86km²)

- Catchment afforestation - 60% decrease annual runoff

HR Wallingford, 2001, and LADEMO land use model

Explore the impacts of climate and LUC on the hydrological regime of the Elbe catchment (80,000km²)

- 10% change urban to

Soil and Water Assessment Tool (SWATmod) - Semi-distributed daily water balance conceptual model

Explore effect of LUC on discharge in Dietzholze catchment (82km²), divided into 218 hydrological response units

- 35% increase in grassland for forest - 27% increase in forest

- 75% increase surface runoff, but 9% increase in discharge - No effect on catchment mass

and increase in arable balance (Crooks and Davies,

2001), South-east UK

Climate and Land use Scenario Simulation In Catchments (CLASSIC) - Semi-distributed conceptual

Explore effect of historical land use changes (1961-1990) on flood frequency in Thames catchment at Kingston (10,000km²)

- 40% increase urban - 24% increase grassland - 30% decline in arable

1. Water Balance Model (WBM) 2. Flash-flood Event-based Spatially-distributed rainfall-runoff

Transformation model (FEST98) 3. Systeme Hydrologique Europeen-TRANsport (SHETRAN)

Evaluate sensitivity of flood risk in flashy systems to the anthropogenic influences on runoff generation mechanisms and climate change using 3 models in 3 catchments

- 12% increase urban - 15% decrease arable

Urbanisation had no impact on flood peaks

European Commission Joint Research Centre - (De Roo et al., 2001)

LISFLOOD – Water balance and flood simulation model at the continental scale – Physical, conceptual, GIS-based distributed

Investigate causes of flooding and influence of LUC, soil characteristics and antecedent moisture conditions in the upper Oder (60,000km²).

- Increase forest and urban - Decrease arable

(Change in land use 1780-present)

0.2% increase peak discharge from 1975-present

(LaMarche and Lettenmaier, 2001) Oregon, north-west USA

Distributed Hydrology Soil Vegetation Model (DHSVM) – Physical grid-based distributed

Examine effects of forest harvesting on flooding in mountainous environments using field observations in Deschutes catchment (149km²).

- Forest removal - Forest roads

- 10% increase annual flood - 10% increase mean annual flood and increased magnitude

(Naef, Scherrer and Weiler, 2002) South-west Germany

Dominant Runoff Processes (DRP) – Binary decision tree

Develop decision support scheme to ascertain likely DRP’s within a temperate catchment and identify appropriate mitigation strategies

No scenarios tested Discharge not produced

(Niehoff, Fritsch and Bronstert, 2002) South-west Germany

Water flow and balance Simulation Model ETH (WaSiM-ETH) with the LUCK toolkit – Physical distributed

Evaluate impact of LUC on flooding by assessing spatial and temporal dynamics of rainfall events in the Lein catchment (115km²).

- Increase in urbanisation by 6%

- 10% land was set-aside

- Increase flood volume &

peak (greater for convective storms)

- Minor increase in runoff for

convective event (Calder, 2003)

Nottinghamshire, England

Hydrology of Land Use Change (HYLUC) – physical daily water balance model

Investigate water use of several vegetation types to derive predictions of the impacts on recharge

Doubling UK woodland Oak woodland reduced flows by half compared to grassland

(Wheater et al., 2008) Flood Risk

Management

Research Consortium (FRMRC) UK

Extended Rainfall-Runoff Modelling Toolkit (RRMT) and Soil-Plant-Water (SPW) model – meta-modelling approach: physical distributed model with conceptual model

Examine LUC on runoff generation and propagation. Used data from multi-scale experiments to inform development/calibration at different scales to make meaningful predictions for Pontbren catchment (16km²)

- Tree strip planting - Catchment afforestation - Deforestation

- 40% reduction overland flow - 9 to 69% decrease peak flow - 6 to 18% increase peak flow

(Pattison, 2010) Thesis, north-west England

Connectivity of RUnoff Model (CRUM) 3 – Physical distributed

Study impacts of LUC on river flows in Eden catchment (2,400km²)

- Afforestation - Soil compaction

- Coniferous forest produced highest peak discharge - Heavy compaction 65%

increase peak discharge - Moderate compaction 3.7%

increase peak discharge (Ewen et al., 2013)

Integrated meta-model – micro-scale physically based hydrological model and lumped model parameterised with HOST soil types and Soil Conservation Service Curve Numbers (SCS-CN)

Explore nature of casual links between land management in rural river

catchments and the flood hydrograph in the Hodder catchment (261km²)

- Blocked gullies and grips - Reduced sheep grazing

Mosaic map outputs show strong spatial patterns in link between land management, soil types, travel distance to outlet and flow rates