8 Catchment Modeling-I

Ranjana Ray Chaudhuri

epgp books

 

 

Objectives:

  • To understand the various modeling principles available for simulating catchment characteristics
  • To develop understanding of increasing complexities in modeling
  • Develop an ability to apply modeling principles for assessing catchment problems

Introduction

 

A catchment is an integrated unit of land, soil and water which provides for society, economy and ecosystem. The catchment is an essential unit which must be healthy for a healthy, protected and sustainable environment. A catchment can typically have industries, cities, towns, as many catchments are no longer are purely agricultural, pasture land or covered completely with forests. Similarly, many catchments support lakes, wetlands, are dependent on ground water completely, have many streams, or are a part of a major river system. Each catchment has specific water resources which have been harnessed for agriculture, industry, domestic use, power generation, drinking water, recreation. This in many regions of the world has led to excessive withdrawals from surface water and ground water so that the current water demands cannot be met, and the diverse ecosystems have started showing signs of degradation. Land is degraded due to excessive use of fertilizers, with intensive agricultural practices, heavy nutrient land or suffering from soil erosion. All this needs the catchment characteristics to be studied in detail and holistically in order to provide support to decision makers.

 

Some of the typical concerns in catchments now due to the excessive use of resources are as follows:

 

Excessive withdrawals from surface and ground water

 

Construction of storage reservoirs, ponds at farm level, leading to changes in water received by downstream

 

Deforestation leading to excessive erosion and changes in sediment flow regime in streams

Pollution from agriculture and industry practices

Eutrophication in lakes and ponds

Excessive fish harvesting

Reduction in flood plain zones due to construction activities

Plantation of crops not native to the catchment, leading to undesirable changes in the catchment

 

In addition, forest fires, floods event and droughts further affect the security of available water resources of the catchment. The impacts of all the mentioned factors above are felt on the environment, society and economy. The idea today is to manage the catchment in such a way so that it continues to be in good health for future generations too or is sustainable in the future too. Sharing water resources between competing uses in a sustainable manner is a challenge which has to be addressed. The issues related to water management are multi-dimensional and multidisciplinary, thus many factors need to be considered, so we use the framework of system analysis or modeling to understand the challenges and provide solutions, be it on catchment scale or river basin scale.

 

The systems analysis or modeling has a demarcated boundary within which the functions take place. Every model has inputs, goes through a connection (systems) and outputs are revealed. Many times the entire catchment cannot be understood with the help of one model only, many subsystems or smaller models (related to a few aspects only) are used which are then collated to understand the complete catchment.

 

A model typically is based on the following:

 

Identification of clear objectives

 

Determining the measurable components in the model

 

Alternative courses available to answer the objectives

 

Comparative evaluation of all alternatives to reach the objectives

 

To give an example, input can be climate variables, land use, water use, and population while outputs can be stream discharge, erosion, sediment flow in streams. The alternative systems available to reach these outputs through the given inputs are what is explored in the various modeling systems.

 

A typical model has the following properties:

 

  • A model must have a clear relationship between input and output or relationship between inflow and outflow
  • Identify and address system constraints like effects on the ecosystem
  • Prioritize between short term and long-term objectives
  • Provision for including stakeholders, so that social, economic and environment are addressed

A model is prepared as it is easier to work with a model than the real system, at the same time it is a representation of the bigger actual system. What parts of the real system is being modeled depends on the sub system that we are expecting to study in detail. The most common model used in the water balance model for a catchment.

 

The water balance equation is as follows:

 

Inflow(I)-Outflow(O)=change in storage (∆S)

 

This basic equation is modeled at the catchment scale.

 

This is the basic hydrological model, where the difference between water inflow and water outflow from the catchment is the change in storage. It represents the continuity equation which is the basis for most hydrological models. The input for the system may be the stream inflow, the stream outflow, while the change in lake level is modeled using the inflow and outflow characteristics to understand the change in lake storage or storage in the catchment in general if the same equation is modeled at the catchment outlet.

 

Some of the common catchment models are on:

 

Rainfall runoff relationships (surface water)

 

Flow of water in vadose-saturated groundwater zone (sub surface water)

 

Surface water-groundwater interactions

 

River basin modeling

 

Models relating water quantity and quality, their interrelationships within the catchment, relationship with the ecosystem and the community.

 

These models also represent increasing complexity as the rainfall runoff model has few variables while the model relating quantity to quality may have many variables.

 

Necessity of modeling

 

Many anthropogenic activities can interfere or change the hydrological cycle. Thus, in order to understand the entire catchment holistically, integration of knowledge from within the water management field to across disciplines such as economic, ecological, social factors are considered while analyzing a system. Models help in understanding the system and if a proper interface is developed then through graphs or otherwise scientific results may be communicated to all stakeholders.

 

Models help in optimization of resources and forecasting and evaluation of resources. A simple rainfall runoff relationship can be established using existing observation, once the model is established; it can be used to predict runoff from different rainfall scenarios. It can also help in optimizing water allocation to different water demands using the quantified runoff.

 

Classification of modeling approaches

 

They are classified into three types (Refsgaard, 1996)

 

a.  The model may be empirical, conceptual or based on physical phenomena

b.    Models may be stochastic or deterministic depending on how inputs are taken in the model.

c. Models may follow lumped or distributed approach

 

Empirical models are the simplest models, in which the relationship between variables is based on observed data alone. For example,

 

Runoff = a X (rainfall)b is of an empirical type, which is also called a black box model.

 

This is because only a mathematical relationship is established between the two variables without going into the physical processes involving the catchment. The parameters a and b are derived from regression analysis of the measured rainfall runoff data only (Subramanya, 2008). These models though may be accurate for a catchment, may not be replicated for other catchments, nor extrapolated for other catchments. These models require data for few variables only and they are region specific. Example of fitting regression lines to fit to rainfall and runoff data can be seen from the mentioned source (www.epa.gov).

 

The next in complexities are conceptual models, also called perceptual models. They use the actual physical condition of the catchment to develop the modeling framework. The hydrological cycle of a area is a conceptual model. The conceptual model of the catchment conceptualizes the many water storage capacity structures and uses mathematical functions to understand the linkages between upstream and down stream surface and sub-surface flows. The conceptual model can be lumped, where all the storage capacity is represented as one unit or semi distributed where the catchment is divided into sub catchment storages which are then linked.

 

The next categories of models are physically based, or process based models. They are more complicated that the other two categories and use more variables, more parameters to explain the system. MIKE, SHE for surface water flow and MODFLOW for ground water flow are computer based algorithms which are physically based models. They solve mathematical equations to understand physical processes like conservation of mass or momentum.

 

The next category of models under discussion is deterministic and stochastic models. The deterministic models are more common, where a single set of input variables result in a single output being determined. In stochastic modeling the variables are represented as statistical distributions, so that a range of output values are generated, each derived from a certain set of variable inputs having certain probability of occurrence.

Event based or continuous modeling

 

When a single storm event is analyzed may be lasting for a few hours or few days, produces a single runoff for the event, it is called event modeling. The other form of modeling is continuous catchment modeling in which surface water balance, subsurface water balance and ground water balance is continuously modeled.

 

Process based models are the most complicated, these models are based on solving mathematical equations describing fundamental principles like the law of conservation of mass or momentum. Often, they assume distributed modeling approach, as in the catchment is divided into many smaller segments or grids as shown in figure 1. As the units of data become finer, often they become more complicated and uncertainty in results obtained increases.

 

Geographic information system (GIS) in modeling

 

GIS allows visualization of the catchment characteristics as free data sets from internet are available for land use, digital elevation, soils and climate variables. So, using GIS tools, studying spatial characteristics of the catchment is becoming more common and therefore hydrological modeling is seeing prolific use of GIS. For example, terrain data sets can be used to determine slope of a catchment which is used as input in hydrological modeling. GIS data uses ground surveys, digitize existing maps, recorded aerial photography, satellite imaging data or a combination of all these. The digital elevation maps help in delineating catchment boundaries and the catchment stream network. Thus, hydrological assessment of catchments is using GIS more and more useful now. There are many hydrological modeling packages using GIS interface.

 

Modeling complexity

 

There is no hard and fast rule that the model must be either empirical or conceptual or process based, it can be a hybrid of all these. It is not always necessary that the most complex model gives the best solutions, the balance between data availability and model complexity gives good prediction. The simplest models are lumped in both space and time while more complex models are distributed in space and time. There are models which give an extremely good fit to data to let us say stream flow, yet at sub catchment level the model does not give representative data as the input parameters needed to represent the catchment a finer scale may be different. Predictive performance of a model is governed not just by model complexity, as the figure below shows model B is far more complex than

 

Model A yet, the predictive performance is not that much better than model A. Optimum data availability and use with model complexity gives optimum predictive performance.

 

To explain modeling complexity, typically flood simulation processes can be used:

 

Deterministic modeling-A flood event is generated with the help of storm hydrograph for the given design rainfall event. Traditionally the temporal rainfall distribution is studied from historical records; the extreme events are chosen for records, volume of runoff generated from such events is determined. However, in these the variability of storm pattern remains challenge which is not addressed. The parameters chosen for such models are the rainfall depth with time, losses, idea about the temporal and spatial pattern of rainfall. All these parameters are fed in the catchment model to generate the flood peak with time Probabilistic modeling-For this kind of modeling, sample stochastic rainfall events are chosen, then a flood frequency curve is generated using storm hydrograph. With Monte Carlo simulations, produces likely flood events hence, allows to incorporate uncertainty in flood predictions.

 

Steps in modeling:

 

1.  Establishment of the model

2.  Calibration

3.  Validation

4.  Outcome

 

Once the model is established, then comes the calibration stage. In the calibration stage it is important to understand the control parameters. These parameters are different for gauged and ungauged catchments. In case of gauged catchments both rainfall and runoff data are available, while in case of ungauged catchments, rainfall data may be available, but runoff data is not available. Thus, control parameters are very difficult to establish in case of ungauged catchments. The control parameters in case of ungauged catchments are established based on physical characteristics which are like a gauged catchment having similar characteristics or from regional relationships, if these are already established. For calibration the control parameters must be modified so that the simulated conditions match with the measured results. The calibrated control parameters for one series of storm events may not be valid for a second set of storm events. However, the system needs to be robust if extrapolation is to be allowed.

 

In the validation step model from a separate time is used to understand the catchment system response. This helps us to assess the accuracy and uncertainty of predictions. One of the most commonly practiced validation process in catchment modeling is called split sample procedure. The data is divided into two groups, one set is used for calibration, using that calibrated model real time data is simulated with the adopted parameters. The next set of data is then used for validating the model. If the data set is large then the data is split in three ways, it is advisable to finally get the parameter values to be used in practice by performing a calibration by using the whole data sample, to reduce the uncertainty of the parameter estimates as much as possible.

 

Expert knowledge is used to estimate parameter values, then observed data is not needed. This is used for ungauged basins, often parameter values from similar basins is used for calibration by an expert who has knowledge of a neighbouring catchment with similar characteristics.

 

Catchment modeling must take care of parameters like rainfall, losses in the catchment. While assessing rainfall, the temporal and spatial patterns must be considered. Whether the storm is sub areal in nature, or if it occurs in a large region then areal reduction factors also need to be considered. Difference in storm characteristics alters the catchment’s response to the storm; any model must be able to capture this aspect. In case of catchment rainfall runoff model, it is important to consider antecedent moisture conditions in the catchment in the 5 days prior to the storm event. For example, the storm event considered may be of 24-hour duration, the frequency may be yearly, 2 yearly, 5 yearly, 10 or 50 years return period for that matter. Thus, the time stepping in modeling is important to fix for any type of model.

 

Continuous modeling-In this case a continuous time series of rainfall events and other climatic factors are studied in order to generate complete stream flow time series using the catchment rainfall runoff response. The continuous modeling brings out the variability in flow pattern, for that variability of input patterns must be available or can be simulated. Innovative storm water management techniques need to be based on continuous models so that the constraints in implementation may be understood and only then implementation shall be successful.

 

Conclusion

 

Catchment modeling is used to generate information on real catchments so that catchments are better managed. In today’s world of water scarcity and soil loss in watersheds, it is an important tool for understanding the resources of the catchment. At the outset we must understand that catchment modeling is not an exact replication of the catchment but a simplified version of the same. We often use mathematical processes to understand the complicated physical processes that occur in the catchment.

 

Holistic planning of catchment as more than the sum of the parts, application of systems theory to catchment planning and management can help to meet multifunctional goals. Traditional approaches to river management have been limited due to tendencies to break up and study the individual influencing variables such as storm water runoff, groundwater recharge, sedimentation nutrient cycling etc. as if they were independent, ignoring the simple fact that water, more than anything else is a landscape integrator. Complex river and catchment systems support and are supported in turn by complex social systems. It is important that we choose the appropriate models so that the objectives are answered, we get to work within defined catchment boundaries and the forecast are reasonably accurate so that the decision makers and stakeholders can take informed decisions. The concept of sustainability must be addressed in catchment modeling too, so that resources are used without long term degradation of the ecosystem in general and water resources.

you can view video on Catchment Modeling-I

Readings

  • Beven, K.J., How far can we go with distributed hydrological modelling?, Hydrology and Earth Science Systems, 5(1), 1-12, 2001.
  • Chow,V.T. (ed), Handbook of Applied Hydrology, Mc Graw Hill, New York, 1964
  • Clarke, K.C., Getting started with geographic information systems, Prentice Hall, 338 pp., New Jersey, USA, 1999.
  • Computer Modeling of Water Distribution Systems. AWWA MANUAL M32Third Edition. Publisher: American Water Works Association
  • Grayson, R. and G. Blöschl, Spatial Patterns in Catchment Hydrology: Observations and Modelling, Cambridge University Press, 404 pp., Cambridge, 2000
  • Powell, S., Jakeman, A. J., Norton, J.P., 2008. Model Development and Analysis. In Sven Erik Jørgensen and Brian D. Fath (Editors), Ecological Models. Vol. 3 of Encyclopedia of Ecology, 5 vols. pp. 2402-2410 Oxford: Elsevier.
  • Subramanya, K., Engineering Hydrology, Tata McGraw-Hill, New Delhi, 2008.