Comparison of Hydrological Models (HBV and LANDPINE) for Inflow Computation in the Modi River, Nepal

TABLE OF CONTENTS

TITLE PAGE

ACKNOWLEDGEMENT

EXECUTIVE SUMMERY

LIST OF FIGURES

LIST OF TABLES

LIST OF ABBREVIATIONS

1 INTRODUCTION
1.1 BACKGROUND
1.2 MOTIVATION
1.3 OBJECTIVE OF THE STUDY
1.4 RESEARCH METHODOLOGY
1.5 THESIS ORGANIZATION

2 DESCRIPTION OF PROJECT AREA
2.1 LOCATION AND COVERAGE
2.2 TOPOGRAPHY
2.3 GEOLOGY, SOIL, VEGETATION AND LAND USE
2.4 HYDROMETEOROLOGY
2.4.1 General Climatic Condition of Nepal
2.4.2 Meteorological Networks in Nepal
2.4.3 Precipitation
2.4.4 Temperature
2.4.5 Evaporation

3 LITERATURE REVIEW ON HYDROLOGICAL MODELLING
3.1 HYDROLOGICAL MODELS
3.2 THE MODELLING PROCESS
3.3 CLASSIFICATION OF HYDROLOGICAL MODELS
3.4 THE HBV MODEL
3.4.1 Background and Introduction
3.4.2 Properties and Applications of HBV model
3.4.3 HBV Model Structure
3.4.4 Catchment Description
3.4.5 The Snow Routine
3.4.6 The Soil Moisture Routine
3.4.7 The Runoff Response Routine
3.5 LANDPINE
3.5.1 Introduction
3.5.2 Model Description
3.5.3 Meteorological Input
3.5.4 High Vegetation
3.5.5 Interception Capacity
3.5.6 Potential Evaporation
3.5.7 Actual Interception and Evaporation
3.5.8 Snow
3.5.9 Low Vegetation and Land Surface
3.5.10 Lake
3.5.11 Root Zone/ Soil Moisture Zone
3.5.12 Surface Flow and Ground Water

4 DATA COLLECTION AND PROCESSING
4.1 GENERAL
4.2 CATCHMENT CHARACTERISTICS
4.2.1 CLIMATIC AND HYDROLOGICAL DATA
4.3 PRINCIPLES OF DATA ANALYSIS
4.3.1 Corrections to Point Measurements
4.3.2 Estimation of Missing Data
4.3.3 Checking the Consistency of Point Measurements
4.4 QUALITY CHECK OF AVAILABLE DATA FOR THE STUDY
4.4.1 Precipitation
4.4.2 Temperature
4.4.3 River Flow
4.4.4 Evaporation

5 METEOROLOGICAL PARAMETERS
5.1 PRECIPITATION GRADIENT ANALYSIS OVER MODI CATCHMENT
5.2 TEMPERATURE GRADIENT
5.3 AREAL PRECIPITATION DISTRIBUTION

6 HBV MODEL CALIBRATION AND VALIDATION
6.1 INPUT DATA PREPARATION
6.1.1 Confine Parameters
6.1.2 Air Temperature and Lapse Rate
6.1.3 Precipitation and Lapse Rate
6.1.4 Potential Evapotranspiration
6.1.5 Runoff
6.2 CALIBRATION OF THE HBV MODEL
6.3 MODEL VERIFICATION/ SPLIT SAMPLE TEST
6.4 RESULTS AND DISCUSSION

7 CALIBRATION AND VALIDATION OF LANDPINE MODEL
7.1 INPUT DATA PREPARATION
7.1.1 Digital Elevation Model (DEM) and Land Use Data Sets
7.1.2 Vegetation Type
7.1.3 Vegetation Cover and Height
7.1.4 Leaf Area Indices
7.1.5 Infiltration Capacity
7.1.6 Field Capacity
7.1.7 Surface Storage
7.1.8 Initial Soil Saturation
7.1.9 Snow Parameters
7.2 CALIBRATION OF LAND PINE MODEL
7.2.1 Model Calibration
7.2.2 The Calibration Process
7.2.3 Hydro-Metrological Input Data Processing
7.2.4 Simulation With Different Parameters
7.3 MODEL VERIFICATION
7.4 DISCUSSION AND EVALUATION OF RESULTS

8 COMPARISON OF COMPUTED RESULTS FROM HBV AND LANDPINE MODEL APPROACHES
8.1 RUNOFF
8.2 AREAL PRECIPITATION DISTRIBUTION
8.3 AVERAGE TEMPERATURE OVER THE CATCHMENT
8.4 SNOW STORAGE
8.5 EVAPORATION

9 COMPARISON OF MODEL RESULTS BETWEEN MODI AND UPPER TRISHULI 3A CATCHMENT

10 CONCLUSION AND RECOMMENDATION
10.1 CONCLUSION
10.2 RECOMMENDATION FOR FURTHER STUDIES

LIST OF REFERENCES

APPENDICES

APPENDIX-A :GRAPHICAL REPRESENTATION OF SIMULATED RESULTS FROM HBV MODEL

APPENDIX-B: REGRESSION ANALYSIS OF COMPUTED RESULTS FROM HBV AND LANDPINE

APPENDIX -C : PHOTOGRAPHS

LIST OF FIGURES

Figure 2.1:Location of Modi Catchment

Figure 2.2 :Physiography of the Nepal Himalaya (after Dahal and Hasegawa)

Figure 2.3 : Generalized geological cross section of the Nepal Himalaya (modified after Dahal 2006)

Figure 2.4 :Seasonal Precipitation Distribution in Nepal

Figure 2.5 :Mean annual precipitation in Nepal ( source: Department of Hydrology and Meteorology)

Figure 2.6 :Temperature variation trend in different region of Nepal*

Figure 3.1 :Hydrological cycle concept (Chow V T, 1989)

Figure 3.2 :Precipitation runoff modeling concept

Figure 3.3 :The different steps in hydrological model application- a modeling protocol, (modified after Anderson and Woessnor 1992)

Figure 3.4 :Classification of hydrological models according to process description

Figure 3.5 :Lumped and distributed Models

Figure 3.6 :Main structure of the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

Figure 3.7 :Catchment area division and hypsographic curve

Figure 3.8 :The snow routine in the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

Figure 3.9 :Soil moisture routine in the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

Figure 3.10 :The runoff response routine in the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

Figure 3.11 :Model structure of LANDPINE (After Rinde T, 1998)

Figure 3.12 :Model Structure for grid cell process in LANDPINE (After Rinde T., 1998)

Figure 3.13 :Snow Distribution within area-element (After Rinde T., 1998)

Figure 3.14 :SOLMST / FLDCAP vrs SOILOUT / INFILTR curve with different β

Figure 3.15 :Response routine for ground water and surface water runoff (After Rinde T., 1998)

Figure 4.1 : Area Elevation curve with DEM and Catchment Delineation of Modi Catchment.

Figure 4.2 :Availability of hydro meteorological data in time scale

Figure 4.3 :Annual precipitation of Lumle, Baglung and Ghandruk stations

Figure 4.4 : Different average annual precipitations and their comparisons

Figure 4.5 :Monthly average precipitations from 1987 to 2000 for three stations

Figure 4.6 :Seasonal distribution of total precipitation from three stations

Figure 4.7 :Double mass curve of annual precipitation of three stations

Figure 4.8 :Mean monthly temperature from 1987 to 2000

Figure 4.9 :Flow regime of Modi River

Figure 5.1 :Precipitation gradient over Modi Region from 835masl to 2742masl

Figure 5.2 :Precipitation gradient from 1740masl to 3705masl

Figure 5.3 :Weightage factor for different precipitation stations

Figure 5.4 :Thissen area for different meteorological stations for Modi catchment

Figure 6.1 :Hypsographic curve, Modi catchment

Figure 6.2 :Temperature lapse rate at different conditions

Figure 6.3 : Model calibration process

Figure 6.4 :Graphical representation of comparison data

Figure 6.5 :Hydrograph observed and computed from HBV from 1991 to 1995

Figure 6.6 :Hydrographs observed and computed from split sample test from 1997 to 2000

Figure 6.7 :Observed and simulated hydrograph, 1991

Figure 6.8 :Observed and simulated hydrograph, 1992

Figure 6.9 :Observed and simulated hydrograph, 1993

Figure 6.10 :Observed and simulated hydrograph, 1994

Figure 6.11 :Observed and simulated hydrograph, 1995

Figure 7.1 :Digital Elevation Model, Modi catchment

Figure 7.2 :Land use and vegetation type according to LANDPINE

Figure 7.3:Vegetation cover and height Modi Catchment

Figure 7.4 :Leaf Area Index, Modi catchment

Figure 7.5: Infiltration and field capacity, Modi catchment

Figure 7.6: Water relations and soil texture (From Forest Service Hand Book)

Figure 7.7 :Surface storage and Initial soil saturation of Modi catchment

Figure 7.8 :Assumed initial snow pack in Modi catchment

Figure 7.9 :Hydrograph from LANDPINE simulation with common Params to HBV

Figure 7.10 :Hydrograph from LANDPINE calibration

Figure 7.11 :Observed and simulated hydrograph

Figure 7.12 :Observed and simulated hydrographs

Figure 7.13 :Plot of Accumulation Difference for calibration period, 1991- 1995

Figure 7.14 :Average snow pack from 1991 to 1995

Figure 7.15 :Observed and simulated hydrograph, Validation period, 1997 -2000

Figure 7.16 :Average snowpack, Validation period, 1997 – 2000

Figure 8.1 :Simulated hydrographs from HBV and LANDPINE

Figure 8.2 :Simulated flow regime from HBV and LANDPINE in Modi River

Figure 8.3 :Accumulated water volume difference from HBV and LANDPINE

Figure 8.4 :Areal precipitation distribution from HBV and LANDPINE from 1991 to 1995

Figure 8.5 :Computed average temperature from HBV and LANDPINE

Figure 8.6 :Initial Snow Pack for HBV and LANDPINE

Figure 8.7 :Computed snow pack from HBV and LANDPINE during calibration

Figure 8.8 :Computed snowpack from HBV and LANDPINE, 1997 to 2000

Figure 8.9 :Computed snow out from HBV and LANDPINE

Figure 8.10 :Actual evaporation calculated from HBV and LANDPINE for a typical year

Figure 8.11 :Potential evaporation from HBV and LANDPINE for a typical year

Figure 9.1 :Area elevation curve for Modi catchment

Figure 9.2 :Area elevation curve for Upper Trishuli 3A catchment

Figure 10.1 :Observed and simulated hydrographs from HBV and LANDPINE

Figure 10.2 :Observed and simulated flow regime

LIST OF TABLES

Table 2.1:Physiographical division of the Nepal Himalaya (modified after Upreti, 1999)

Table 2.2 : Location and annual rainfall of different stations

Table 3.1 :Distributed input parameters for LANDPINE

Table 4.1 : Details of meteorological stations (source: Hydrological Estimations in Nepal)

Table 4.2 :Annual precipitation in mm from 1987 to 2000 for different stations

Table 4.3 :Monthly average precipitation, from 1987 to 2000.

Table 4.4:Percentage distribution of precipitation from 1987 to 2000

Table 4.5 :Calculated Potential evapotranspiration at different locations

Table 5.1 :Monthly average temperature gradient 0C/100m

Table 5.2 :Optimum weightage calculation for different precipitation stations

Table 5.3 :Areal precipitation distribution over the catchment

Table 5.4 :Snow storage over the catchment

Table 6.1 :Confine parameters for Modi catchment

Table 6.2 :Average temperature over the catchment

Table 6.3 :Comparison of precipitation and runoff data

Table 6.4 :Optimized free parameter’s values for Modi catchment

Table 6.5 :Results of Model calibration

Table 6.6 :Results of model verification

Table 7.1 :USGS and LANDPINE land use/ Land cover system legend (Modified level 2)

Table 7.2 :LANDPINE/ land use classification

Table 7.3 :Assumed vegetation cover and height in Modi catchment

Table 7.4 :Vegetation type with leaf area index

Table 7.5 :Infiltration capacity of different soils

Table 7.6 :Surface storage for different land surfaces

Table 7.7 :Model calibration parameters

Table 7.8 :Optimized Free Parameter in LANDPINE

Table 8.1 :Observed and simulated monthly average daily discharge from HBV and LANDPINE in m3/sec

Table 9.1 :Optimized free parameters for Modi and Upper Trishuli 3A catchment

Table 9.2:Model simulation results for Upper Trishuli and Modi catchment

LIST OF ABBREVIATIONS

illustration not visible in this excerpt

ACKNOWLEDGEMENT

I am deeply indebted to my supervisor, Associate Professor Knut Alfredsen, for his valuable and continuous guidance throughout the work. His cooperation, stimulating suggestions and encouragement helped me to finalize the work on time.

I would also like to acknowledge Prof. Haakon Støle and Prof. Ånund Killingtveit, Professor In charge for M.Sc. programme in Hydropower Development, Mrs Hilbjørg Sandvik, course coordinator and other staff members for their academic, technical and administrative support during my stay in Norway.

My appreciation also goes to Mulugeta Bereded Zelelew and Yisak Sultan Abdella, Ph D fellows for their co operation during my thesis work. I am very much thankful to Mr Jagadishwor Man Singh Pradhan, Manager, Project Development Department, Nepal Electricity Authority for his kind co operation with stimulating suggestions from the beginning of my carrier and my colleague Padam Kunwar, Ananda Dhungel and, Parmardhir Sen for their co operation during my data collection work.

I would like to thank Norwegian State Educational Loan fund for providing financial support for my study. I am very indebted to my parents and family members who always insist me for study and hard working. Finally, I would like to thank all my friends who helped and supported me to accomplish this task on time and my special thanks goes to my colleague Subarna Shrestha and Kiran K Shrestha for their cooperation in every part of my life during my stay in Trondheim.

EXECUTIVE SUMMERY

Hydrological system is a very complex process having great contribution in the control of water balance in the world. The hydrological cycle is a major part of hydrological system and its components are precipitation, evaporation, snow melt, infiltration, run off, ground water movement etc. The hydrological modeling is a system concept comprising the effects of all the components involving in the hydrological cycle.

In this study, comparison of hydrological models for inflow computation in the Modi river, a tributary of Kali Gandaki river, has been carried out by using HBV a lumped and LANDPINE a distributed model although, at first both models were developed for Scandinavian catchments. The HBV model is a conceptual precipitation run off model which is used to simulate the run off process in a catchment based on observed data of precipitation, air temperature and potential evapotranspiration. LANDPINE is a distributed hydrological rainfall runoff model to study the effects of land use changes in runoff from catchment. It operates in integration with a geological information system, usually used for preparation of input data and analysis and presentation of simulation results.

The hydro meteorological data from stations Lumle, Ghandruk and Baglung with daily discharge data at Nayapool, have been used for the study of the catchment. The two stations, Lumle and Ghandruk lie within the catchment in its southern part and the Baglung lies about 12 km outside in south west direction from the boarder of the catchment.

Available precipitation data from stations within the region are examined to determine the trend of precipitation with respect to elevation. The long term average annual precipitations for all the stations from DHM have been used. The precipitation records show that the precipitation gradient is not only dependent on the elevation and seems to be difficult to define only as a single function of the elevation. It is observed that the precipitation increases up to certain elevation and decreases thereafter.

Because of the complex nature of precipitation distributions and limited number of gauging stations, the areal precipitation distribution for this study in the case of HBV model is calculated by the combination of different weightage for different precipitation stations according to the best result obtained among them. In order to find the best combination, the stations have given different possible weightage factors to calculate areal precipitation over the catchment and HBV has run in each combination from 1991 to 1995. Based on the R2 and percentage of deviation of runoff value obtained from each run, the best weightage combination for precipitation stations has been found with 0.43 to Lumle, 0.35 to Baglung and 0.22 to Ghandruk which gives R2, 86.0% and 0.24% runoff deviation. The LANDPINE computes the areal precipitation distribution by inverse distance weighting. Some spatially distributed inputs required for LANDPINE are prepared in Arc Hydro and then converted in IDRISI format to make it readable for LANDPINE.

Data from 1991 to 1995 and from 1997 to 2000 have been used for calibration and validation respectively for both models. HBV model has shown good performance for both goodness of fitting and coefficient of determination, R2 0.86 with -0.24% deviation in water volume during calibration and R2, 0.82 with -8.89% deviation in water volume during validation. The LANDPINE model has also performed well with some deviation in its free parameters from the suggested values for the Scandinavian catchments. The simulation resulted in R2, 0.80 and 0.82 for the calibration and verification period respectively. The respective accumulated water volume differences are -1.04 mm and -1105 mm. The accumulated difference during verification of the model seems high compare to calibration of the model. Both models also have produced good snow simulation during calibration and validation periods.

Both models have shown satisfactory performance in normal flow situation but they are unable to catch the peaks during its monsoon periods. Output results such as simulated runoff, areal precipitation distribution, snowpack distribution, snow out from the catchment and temperature distribution from both models are compared and found consistent and promising.

In conclusion, the results obtained from both HBV and LANDPINE model calibration for Modi catchment indicates the well performance of the models and they are equally applicable for tropical catchments in Nepal, though they are developed for the temperate region like Scandinavia. These models may be a useful tool for further hydrological analysis like climate change, flood forecasting, long term computation of availability of water, planning and operation of hydro power projects, and upgrading of existing power projects.

CHAPTER 1: INTRODUCTION

1.1 BACKGROUND

Nepal, a country with an uncommonly diverse geography, is roughly trapezoidal in shape with approximately 800 km in length and 200 km in width. Its total area is 147181 sq km , and is located between 800 03ʹ E to 880 12ʹ E longitude and 260 22ʹ N to 300 25ʹN latitude. It is a landlocked country situated between India to the east, west and south and China to the north.

Nepal is commonly divided into three physiographic areas, the Himalayan region, Mountainous region and the plain Terai region. These ecological belts that run east to west and are vertically intersected by Nepal’s major, north to south flowing river systems. There are more than 6000 rivers originating from high Himalayas in the northern part of the country and flowing down south. The southern lowland plains are the part of the northen rim of the indo gangetic plains. They were formed and fed by three major rivers; Gandaki, Koshi and Karnali.

Nepal, with its rich water resources, presently has untapped hydroelectric potential amounting around 43,000 MW. The abundance of water resources and favorable topography has provided ample opportunities for its development. The power harnessed so far is however minimal (less than 2% of its total capacity). Because of its renewable nature , environmental friendliness compared to other power projects and easy availability ,water resource is supposed to be the most prominent one in terms of economic development of the country.

Availability of water is the basic input parameter to evaluate the capacity of power production from any river. Hydrological study is the primary process to evaluate those input parameters and hydrological models are now a days the essential tools for reliable hydrological study to determine the capacity of power projects form planning to operation stage. Therefore hydrological models are very important for the development of hydropower projects.

Several hydrological models are avaiable but HBV is a simple lumped model often used in hydrological study in hydropower development in Scandanavian countries and LANDPINE recently developed distributed model in NTNU, which has great potential to assess the hydrological parameters in micro levels.

In this study, a comparision of hydrological models for inflow computation in the Modi river, a tributary of Kali Gandaki river, has been carried out using HBV, a lumped, and Landpine a distributed model. Results obtained from this study could be a valuable tool for planning and operation of existing and planned power projects. There are more than three planned and one existing projects in the Modi basin.

1.2 MOTIVATION

Nepal is one of the country having substantial water resources in the world. However it has not able to utilize its full resources because of many constraints associated with it. Limitation of financial resources, computation of availability of resources, planning and design, management of skilled manpower etc are some examples of it. The basic input parameter for the development of hydropower is the availability of water during the year. Because of the influence of monsoon in the climate, the precipitation greatly varies throughout the year. Therefore, the discharge in the Nepalese rivers greatly varies throughout the year. About 80% of the discharge is passing through the river during its monsoon period. So, the computation of real inflow from any river is the difficult task and is very important from the hydropower production point of view.

The water distribution is uneven throughout the year due to the extreme spatial and temperal variability in the climate and the rainfall. The variation of mean monthly stream flows between dry and wet season is very high in most of the rivers of the country.

For the efficient planning, management and monitoring of water resources, hydrological and meteorological stations are the key resources to get the basic input information. But the existing density of hydrological and meteorological stations is very low in the country. Therefore understanding the problems and the nature of the hydrological system will help to improve the utilization of water resource projects efficiently in the country.

Hydrological modeling is one of the tools which provide the concept of complex hydrological process and help to determine the inflow coming from stream. The HBV is a lumped hydrological model and LANDPINE is the distributed and considers the influence of climate change, land use changes, human activities on the catchment.

The work of this thesis is to deal with the comparison of inflow computation from Modi catchment by using both HBV and LANDPINE model. The work includes data analysis, model calibration and validation, comparison of two modeling approaches and finally the comparison of model parameters of Modi catchment with Upper Trishuli 3A catchment of Nepal. Both are the sub catchments of Gandaki basin which is one of the major basin among three basins of Nepal Koshi, Gandaki and Karnali. The intention of the thesis is purely academic purpose of understanding the hydrological process and their representation in modeling system. The outcome can be utilized, modified and used as an information source for further study of hydrological analysis of the Modi catchment and its surrounding.

1.3 OBJECTIVE OF THE STUDY

The main objective of this study is to calibrate the hydrological models in Modi catchment by using two modeling approaches, lumped and distributed and then to compare the results from these two approaches. The HBV and LANDPINE models are used as lumped and distributed models respectively. The following tasks have been done to meet the objectives of the study.

- Collection, preparation and analysis of input data and filling the missing gaps in data series.
- Extraction of the catchment from USGS Hydro 1K data set and preparation of hypsographic curve by using Arch Hydro.
- Evaluation of precipitation distribution and computation of areal precipitation
- HBV model calibration and validation
- Preparation of distributed data and calibration and validation of LAND PINE model
- Comparison of output results from both modeling approaches.

The main objectives of the study are as follows:

1. An overview of the modeling practices in hydrology, principles of modeling approaches.
2. Preparation of input data, data analysis, evaluation and correction if necessary of the climatic and run off data for the analysis.
3. Calibration of the HBV and LANDPINE model and testing the model against the validation period.
4. Evaluation of the modeling parameters and results obtained from both modeling approaches.
5. Comparison and evaluation of model parameters of Modi catchment with Upper Trishuli 3A catchment.
6. Preparation of report describing the work done during the study period.

1.4 RESEARCH METHODOLOGY

The research methodology is based on qualitative and quantative analysis of the results obtained from HBV, a lumped and LANDPINE, a distributed model and the observed data from the catchment and its surroundings. This study work utilizes all sources of information and data, climatological, hydrological, topography map of the project area, research and scientific papers and literature found from the internet relevent to the subject. The stream flow data and Meteorological data for selected stations are collected from Department of Hydrology and Meteorology,Kathmandu, Nepal.

Topographic maps with scales 1:50000 and 1:25000 produced by Government of Nepal, Survey Department are used for the study area of Modi catchment.The Ditigital Elevation Model (DEM), the land use and the vegetation types are obtained from USGS HYDRO 1K database. The digital maps in raster form are processed using Arc GIS with Arc Hydro tools to find the catchment area and hypsographic curve. For the HBV model calibration, M S Excel is used.

1.5 THESIS ORGANIZATION

This thesis is composed of main ten chapters. The first chapter describes the intodructory part dealing with background, motivation and objective of the study.The second chapter contains about the description of the project area.The third chapter contains literature review describing about the model types used in hydrological analysis.

Data collection and processing contains in the fourth chapter including stream flow data, precipitation and temperature, filling of missing data and checking its quality. Meteorological parameters, precipitation and temperature gradient, areal precipitation distribution which have great influence during hyrological modelling are described in the fifth chapter of the report.

The sixth and seventh chapter contain about the calibration and validation of HBV and LANDPINE model including input data preparation respectively. Results of both HBV and LANDPINE and discussions are also included in the respective chapters.

In chapter eight, comparison of computed results between HBV and LANDPINE is presented while in the ninth chapter, the input parameters and results obtained from this study are compared with the different catchment, Upper Trishuli 3A catchment which has already been studied .

Finally, the chapter ten contains conclusion from this study work and some recommendations have been proposed for further hydrological modelling study to check/verify the results obtained from this study.

CHAPTER 2: DESCRIPTION OF PROJECT AREA

2 DESCRIPTION OF PROJECT AREA

2.1 LOCATION AND COVERAGE

The study area, the Modi catchment is situated in Western development region in the mountain range of Nepal. It is a sub catchment of Kali Gandaki river which is one of the major tributary of Gandaki river basin, the central basin of the country. The catchment has a surface area of 578 km2 and is located between latitudes 280 15ʹ N to 280 37ʹN and longitudes 83 038ʹ E to 840 00ʹ E.

illustration not visible in this excerpt

Figure 2.1:Location of Modi Catchment

The climate of the basin is influenced by the physiography of the region and becomes colder and colder with the increase in altitude. Therefore the altitude is the main factor controlling the temperature in the catchment. The precipitation also increases with the increase in altitude up to certain level and thereafter its value reduces. The south east monsoon is responsible for most of the rainfall in the basin. The mean annual precipitation is estimated around 3970 mm . The monsoon starts at late June and continues until the late September. The winter starts in November and lasts until February and it gets little precipitation by short winter monsoon. The climate becomes progressively warmer from February until the beginning of the next monsoon. The mean annual temperature of the basin is 8.410C which increases from north to south.

2.2 TOPOGRAPHY

Nepal is situated in a very high elevation lap varying from 70 m in the south plain area to the highest mountain in the world, the Mount Everest 8848 m in the north. There are several mountains from central part to the northen part of the country and the peaks and the numbers are increasing towards north. The diverse topography of the country generally features all kind of terrains for example rugged mountains, rolling plains and low lands. The major river basins include the Koshi, Gandaki and Karnali flowing towards south and become tributaries of the Ganges system. The longest river is the Karnali and the largest river in terms of discharge is the Koshi.

illustration not visible in this excerpt

Figure 2.2 :Physiography of the Nepal Himalaya (after Dahal and Hasegawa)

The low land in the southern part of the country extending from east to west known as the Terai, this region comprises both cultivable land and dense jungle. The second and by far the largest part of Nepal are formed by the Mahabharat, Churia, and Himalayan mountain ranges, extending from east to west and their altitude increases towards the north. Eight of the world's highest mountains are situated in the Himalaya range.

According to the topography map produced by government of Nepal scale (1:50000), the elevation of Modi catchment varies from 700 m to 8091 m above sea level at the highest level of the catchment boundary. However, the catchment obtained from Arc Hydro only covers the elevation from 709 m to 7116 m above sea level. Area between 7116 m and 8091 m is very small and not considered for this study. The modi river starts from the Annuparna glacier and flows around a distance of 45 Km along the valley of the mountain up to the outlet of the study area and it gets Kali Gandaki river at the catchment out let some kilometers downstream from the existing gauging station. The catchment is elongated from north towards south with a mean width of 12 Km. The average length is around 45 Km with an average gradient 0.14 .

2.3 GEOLOGY, SOIL, VEGETATION AND LAND USE

According to physiographic division of Nepal, the project area lies in midlands, fore Himalaya and higher Himalaya. The main rock types present in mid lands are schist, phyllite, gneiss, quartzite, granite, limestone and geologically belongs to the Lesser Himalayan Zone. The fore Himalaya consists of gneisses, schists, phyllites and marbles and mostly belongs to the northern edge of the Lesser Himalayan Zone.The Higher Himalaya range starts from elevation 5000 m and gneisses, schists, migmatites and marbles belong to the this zone.

The vegetation in Modi catchment varies from open shrub land to forest area.The upper part of the catchment is occupied by grass and barren area. The lower part of the catchment consists of forest and cultivated land.

illustration not visible in this excerpt

Table 2.1:Physiographical division of the Nepal Himalaya (modified after Upreti, 1999)

The latest physiographic data shows that Nepal comprises around 4.27 million hectares (29% of total land area) of forest, 1.56 million hectares (10.6%) of scrubland and degraded forest, 1.7 million hectares(12%) of grassland, 3.0 million hectares (21%) of farmland, and about 1.0 million hectares (7%) of uncultivated lands. Forest cover in the Terai and Hill areas is reported to have decreased at an annual rate of 1.3% and 2.3%, respectively between 1978/79 and 1990/91.(Nepal Biodiversity Resource Book Protected Areas, Ramsar Sites, and World Heritage Sites)

2.4 HYDROMETEOROLOGY

2.4.1 General Climatic Condition of Nepal

Nepal has great variation in climate due to uneven distribution of elevation. Although Nepal lies near the northern limb of the tropics, because of rugged topography, there is a very wide range of climates, from the summer tropical heat and humidity of the lowlands to the colder dry continental and alpine winter climate through the middle and northern mountainous sections.From south to north five defined climatic zones exist in the country. These are tropical, sub tropical, temperate, alpine and sub arctic. However the dominant climatic influence is the south east monsoon.Basically, there are four climatic seasons: March to May (Spring), June to August (Summer), September to November (Autumn) and December to February (Winter). The monsoon is approximately from the end of June to the middle of September.

During Northern Hemispheric summer, the southeast trade wind blows from the Southern Hemisphere, due to an intense pressure gradient towards India.In Nepal the monsoon air masses advance from east to west, and the level of maximum moisture advection will increase due to orographic ascending of monsoon air masses. In the eastern lowland this level is below 1500 meter and when it arrives in Kathmandu valley it is raised to 3000 meter above mean sea level (Nayaju, 2000). Hence the precipitation distribution will be different for the same height barrier for different location.

illustration not visible in this excerpt

Figure 2.3 : Generalized geological cross section of the Nepal Himalaya (modified after Dahal 2006)

Based on altitude,Nepal is divided into five climatic zones .

- Tropical and sub tropical zone of below 1200 m in altitude
- Cool, temperate zone of 1200 m to 2400 m in altitude
- Cold zone of 2400 m to 3600 m in altitude, and
- Artic zone above 4400 m in altitude.

2.4.2 Meteorological Networks in Nepal

Department of Hydrology and Meteorolgy is responsible for the establishment of metrological stations throughout the country. It has established and maintained nation-wide networks of 337 precipitation stations, 154 hydrometric stations, 20 sediment stations, 68 climatic stations, 22 agrometeorological stations, 9 synoptic stations and 6 aero-synoptic stations. Mostly the stations are scattered on the mid mountain and southern part of Terai region. There are very few stations in high altitude area which measures the solid precipitation and other meteorological data. A few climate gauging stations have equipped with measuremet of all meteorological records like precipitation, temperature, evaporation, wind velocity etc. Most of the stations are working as rain gauge stations. There are seven metrological stations in the vicinity of the Modi catchmnet.

2.4.3 Precipitation

The amount of annual rainfall and its its distribution is also different in different parts of the country. Up to certain elevation, the total annual precipitation increases with increase in elevation. The rainfall pattern over Nepal is very much dictated by its topography.There are particularly three regions with high rainfall (>3000mm), namely in the middle hilly region of central part of Nepal, region around (86 E, 28 N) and other in the north eastern part. South of these regions receives relatively less rain which mostly lie in the lee side of the lower mountain range in the south. In general, the eastern and central regions of Nepal receive more rainfall in comparison to the western region.

The amount of precipitation also varies according to the season. About 80% of total precipitation falls during monsoon (June to September); about 15% falls during pre-monsoon (March - May) and post monsoon (October); and less than 5% falls during winter (November - February). It is observed that July is the wettest month and November is the driest month of the year. As the summer precipitation (monsoon) advances from east to west so the winter precipitation advances from the west to east but its extension and intensity is so weak compared to summer monsoon that its appreciable effect is limited to Mid Western and Far Western regions. The average precipitation in Nepal is 1800 mm and its distribution ranges from less than 250 mm in the north-central portion near the Tibetan plateau to more than 5000 mm in along the southern slopes of Annapurna Range in central Nepal (Nayaju, 2000). The Modi catchment lies in the southern part of the Annuparna Range.In summer, precipitation comes from eastern side of the country and gradually decrease towards the westren side, the monsoon coming from the bay of Bangal is mainly responsible for this precipitation . Therefore, the trend of precipitation over the country is complex and difficult to define in a single function of elevation.

illustration not visible in this excerpt

Figure 2 . 4 :Seasonal Precipitation Distribution in Nepal

illustration not visible in this excerpt

Figure 2 . 5 :Mean annual precipitation in Nepal ( source: Department of Hydrology and Meteorology)

Two operating meteorological stations within the catchment and one outside from it are used for the hydrological analysis of the Modi catchment. The two stations within the catchment are Lumle (Index No. 0814) and Ghandruk (Index No.0821) and the station outside the catchment used for the analysis is Baglung( Index No.0605) which lies in eastern side around 12 km far from the boarder of the catchment . The annual precipitation data from each station is used and there is no temperature data available in Ghandruk station at all. So the average temperature of Lumle and Baglung stations is used as the input temperature data for the analysis. There are seven metrological stations in the vicinity of the catchment. The location and annual rainfall of each station is presented in Table 2.2.

illustration not visible in this excerpt

Table 2.2 : Location and annual rainfall of different stations

2.4.4 Temperature

Air temperature is a temporal and spatial dependence metrological parameter. Mean annual temperature is greatly influenced by altitude. A study conducted in Annuparna Region shows that the temperature gradient during the monsoon between the lowland stations and a highland station located in Machhapuchhre basecamp is found between -0.07 0C to -0.078 0C per 100 m rise of elevation reflecting well mixed atmospheric condition where as during the dry winter season the gradient typically varies between -0.08 0C to 0.09 0C per 100 m and the study area also lies in the same region. However an average of 0.04 0C to 0.06 0C fall of temeperature with increase of 100 m elevation known as lapse rate is used for this study.

Analysis of maximum temperature data from 49 stations in Nepal for the period 1971–94 reveal warming trends after 1977 ranging from 0.06°C to 0.12°C yr−1 in most of the Middle Mountain and Himalayan regions, while the Siwalik and Terai (southern plains) regions show warming trends less than 0.03°C yr−1. The subset of records (14 stations) extending back to the early 1960s suggests that the recent warming trends were preceded by similar widespread cooling trends. Distributions of seasonal and annual temperature trends show high rates of warming in the high-elevation regions of the country (Middle Mountains and Himalaya), while low warming or even cooling trends were found in the southern regions. This is attributed to the sensitivity of mountainous regions to climate changes. The seasonal temperature trends and spatial distribution of temperature trends also highlight the influence of monsoon circulation.

Spatial distributions of maximum temperature trends in Nepal show high warming trends in most of the Himalayan region and the Middle Mountains, while low warming or even cooling trends are observed in most of the Terai and the Siwalik regions. Though the actual mechanisms are not well understood, monsoon circulation may play an important role in the distribution of seasonal temperatures as well as temperature trends.

illustration not visible in this excerpt

Figure 2 . 6 :Temperature variation trend in different region of Nepal *

2.4.5 Evaporation

Evaporation is the loss of water and it is the primary process of water transfer in the hydrological cycle. The evaporation is a variable aaccounting on temperature, humidity, solar radiation and wind speed.There are very few stations for measuring evaporation and located below 2000 masl.The potential evaporation is calculated in the mean elevation of the catchment by using Thornthwaite’s formula and found 319 mm per year.

CHAPTER 3: LITERATURE REVIEW ON HYDROLOGICAL MODELLING

3 LITERATURE REVIEW ON HYDROLOGICAL MODELLING

Hydrological systems are very complex and have great contribution in the control of water balance in the world. As a result, its effects can be seen in natural as well as in manmade systems around the world. Affects in physical, chemical and biological conditions in the environment, administrative conflicts in water use are some of the examples of affects caused by hydrological systems. The hydrological cycle is a major part of hydrological system and its components are precipitation, evaporation, snow melt, infiltration, run off, ground water movement etc and the hydrological modeling is a system concept comprising the effects of all the components involving in the hydrological cycle.

illustration not visible in this excerpt

Figure 3.1 :Hydrological cycle concept (Chow V T, 1989)

From the hydropower generation point of view, the runoff /available water in nature is of great importance. It is a basic recourse for hydropower and is very important to know its availability during its planning and operation of the system. Planning, projecting and controlling water economy measures and constructions need optimum information on flow conditions at any given location. It is very difficult to get required data in each and every required point. So, hydrological models could be the suitable means to predict the discharge at any required location. They could also be important tools for understanding the process and estimating the hydrological variables of interest required for water resources management.

3.1 HYDROLOGICAL MODELS

A hydrological model is the representation of an actual system. Inputs and outputs, the measurable hydrological variables are linked with a set of equations and the system transformation acts as connector between the input and the output parameters.

Because of the complex nature of hydrological process, it’s a very difficult task to develop mathematical expression for quantitative prediction of parameters involving in the system. The development of hydrological models in the real world is important because of the following reasons.

- Limitation of hydrological measurement techniques
- Limitation on measurements in space and time
- Verification and analysis of the quality of measured data.

illustration not visible in this excerpt

Figure 3 . 2 :Precipitation runoff modeling concept

(Source: Lecture Note, HPD, NTNU by Prof Ånund)

3.2 THE MODELLING PROCESS

There are various modeling procedures for the use of conceptual rainfall run off modeling. The procedures may vary according to the case. A procedure comprising a sequence of steps in a hydrological model application is refereed as modeling protocol.

The first step in the modeling process is to define the purpose of the model application and describe the conceptual model of the user’s perception of the key hydrological process by mathematical expressions and logical statements. The perceptual model is the summery of perceptions of catchment response due to rain fall under different conditions.

Almost all hydrological models contain equations involving variety of input and state variables. The inputs related to geometry of the catchment remains constant during the particular period of the simulation. The other types of input parameters that define the time variable boundary conditions during simulations may change in the modeling process.

Model calibration in general is the process of manipulation of a specific model to reproduce the response of the catchment in different conditions within the range of accuracy specified in the performance criteria. This is usually done with trial and error adjustment of parameters. The validation of the model is the checking of observed value with estimated one by using the same parameters of its calibration. The model is said to be validated if it fulfills the criteria within acceptable limits as specified in the performance criteria. A typical modeling procedure suggested by Anderson and Woessnor is shown in Figure 3.3.

illustration not visible in this excerpt

Figure 3.3 :The different steps in hydrological model application- a modeling protocol, (modified after Anderson and Woessnor 1992)

3.3 CLASSIFICATION OF HYDROLOGICAL MODELS

There are several systems for classification of hydrological models. The models are classified according to three main criteria.

- Randomness (Deterministic, Stochastic)
- Spatial variation (Lumped or Distributed)
- Time variability (Time dependent and Time independent)

illustration not visible in this excerpt

Figure 3.4 :Classification of hydrological models according to process description

Deterministic model is a mathematical model in which outcomes are precisely determined through known relationships among inputs and outputs, without any random variation. A given input always produces the same output. The stochastic model on the other hand uses ranges of values for variables in the form of probability distributions. Deterministic models can be classified to whether the model gives a lumped or distributed description of the considered area and whether the description of the hydrological processes is empirical, conceptual or more physical based. The three main groups of deterministic models are:

- Empirical models ( Black box models)
- Lumped conceptual models ( Grey box models)
- Distributed process description based models (White box models)

A black box or an empirical model uses the mathematical equations without any physical process in the catchment. It only uses the analysis of concurrent input and output time series.

A conceptual model is one that is constructed on the basis of the physical process in the catchment that we can observe on the catchment. Physically sound structures and equations together with empirical one are used in conceptual model. However, the physical significance is not usually so clear that the parameters can be accessed from direct measurements. The only way to estimate the parameters from calibration is by applying concurrent input and output time series. A conceptual model, usually a lumped type is often called a grey box model.

illustration not visible in this excerpt

Figure 3.5 :Lumped and distributed Models

(Source: Lecture Note, HPD, NTNU by Prof Ånund)

In lumped model, catchment is handled as a homogeneous unit and model parameters apply to the whole area where as in the distributed model, the parameters like land use, soil class, elevation may differ from grid to grid.

A physically based model describes the natural system using the basic mathematical representations of flow of mass, momentum and various forms of energy. For catchment models, a physically based model in practice also has to be fully distributed. This type of model is also called a white box model. This model consists at its most basic ‘human friendly’ and considers all activities in a scientific way.

3.4 THE HBV MODEL

3.4.1 Background and Introduction

The HBV model was originally developed by SMHI in the early 70´s to assist hydropower operations. The aim was to create a conceptual hydrological model with reasonable demands on computer facilities and calibration data. The HBV approach has proven flexible and robust in solving water resource problems and applications now span a broad range. The HBV model is named after the abbreviation of Hydrologiska Byråns Vattenbalansavdelning (Hydrological Bureau Water balance-section). This was the former section at SMHI, the Swedish Meteorological and Hydrological Institute, where the model was originally developed.

The HBV model is a conceptual precipitation run off model which is used to simulate the run off process in a catchment based on observed data for precipitation, air temperature and potential evapotranspiration. The HBV model is based on water balance equation which can be described as:

illustration not visible in this excerpt

Where:

illustration not visible in this excerpt

The model computes snow accumulation, snow melt, and actual evapotranspiration, storage in soil moisture and ground water and run off from the catchment. HBV can be used as a semi-distributed model by dividing the catchment into sub basins. Each sub basin is then divided into zones according to altitude, lake area and vegetation. The model is normally run on daily values of rainfall, air temperature, and daily or monthly estimates of potential evaporation. Now, the original HBV model has been revised several times both in Sweden and Norway.

3.4.2 Properties and Applications of HBV model

HBV is a lumped rainfall-runoff model with a distributed snow routine. It uses mostly 10 elevation zones and a 5-point statistical distribution of snow in each zone. Precipitation and temperature are the driving variables; elevation lapse rates are fixed or calibrated. A rather unique feature is the lack of an infiltration routine; all precipitation is assumed to enter the unsaturated zone. This assumption is generally valid for Scandinavian till catchments, where Hortonian overland flow hardly occurs. The soil water storage includes a non-linear term. The model has the following properties:

Some extent a linear model: Most of the mathematical expressions in the model are linear. Some parts of the model like soil moisture routine are nonlinear and the run off generating response function is based on linear reservoirs with some modifications.

Basically a lumped model: Catchment is treated as a single unit without any considerations to the spatial distributions within the catchment but the snow routine is distributed among different elevation zones.

Conceptual model: The HBV model is based on some considerations of the physical structure and process in the catchment. This structure is based on hydrological knowledge and it is verified during the development and testing of the model. Only the most important processes and storages in the catchment have been included in the model.

Deterministic model: The HBV model has no components controlled by probability. It means two equal sets of input always yield the same output, if run through the model from identical start conditions and with identical model parameters.

The HBV model contains a number of parameters that need to be given values before its application. The model parameters can be grouped as free and confined parameters. Confine parameters may be determined from maps, field surveys, or other sources of information about the catchment. The confined parameters are never changed, once they have been determined. Catchment area, lake percentage, area elevation curve are the examples of confined parameters. Free parameters are determined by a process of calibration of the model. They are normally determined before the model is taken in operational use and kept constant for later use. Degree day factors for snow melt, field capacity in soil etc are the examples of free parameters. A properly calibrated HBV model may have different applications some are as follows:

- Run off forecasting
- Flood forecasting
- To generate run off time series from meteorological data (precipitation, air temperature)
- To fill in missing run off observations
- As a tool in quality control of runoff data
- To determine the effects of changes in the catchment
- To study the effects of climate change

3.4.3 HBV Model Structure

Like other rainfall runoff models the HBV model is also based on a conceptual representation of a few main components in the land phase of the hydrological cycles. The runoff from a catchment is computed from observed climatic data like precipitation, air temperature and estimated potential evapotranspiration. The model computes storage water balances in different layer of the catchment to accomplish the relation between the climatic data and runoff and it also shows the dynamic variation in storage water in response to the varying meteorological inputs. The standard version of the HBV model uses the four main storage components snow routine, soil moisture, upper zone and lower zone. In addition a separate river and lake storage may be used when needed.

illustration not visible in this excerpt

Figure 3.6 :Main structure of the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

illustration not visible in this excerpt

Figure 3.7 :Catchment area division and hypsographic curve

(Source: Hydropower Development Series, Vol 7, Hydrology)

3.4.4 Catchment Description

The most important characteristics of the catchment are described by the following parameters in the HBV model:

- Catchment area (km2)
- Area of natural lakes (km2)
- Area of regulated lakes (km2)
- Area elevation curve (Hypsographic curve)

The catchment is divided in ten elevation levels as shown in Figure 3.7. The area of each zone with its average elevation is used to construct the hypsographic curve.

3.4.5 The Snow Routine

The standard snowmelt routine of the HBV model is a degree-day approach, based on air temperature, with a water holding capacity of snow which delays runoff. Melt is further distributed according to the temperature lapse rate and is modeled differently in forests and open areas. A threshold temperature is used to distinguish rainfall from snowfall. The snowpack is assumed to retain melt water as long as the amount does not exceed a certain fraction of the snow. When temperature decreases below the threshold temperature, this water refreezes gradually. A snow distribution can be made in each zone by subdividing it into a number of subareas with different snow accumulation. At each zone the model computes the following variables:

- Air temperature based on observed temperature at climatic stations and the air temperature lapse rate

- Amount of precipitation based on observed precipitation and the precipitation lapse rate
- Precipitation type based on air temperature
- Snow melt or refreezing based on air temperature

illustration not visible in this excerpt

Figure 3.8 :The snow routine in the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

The main results of the computations in the snow routine are the following three variables which are computed for each elevation zone and time step.

- Snow storage in mm of water equivalent
- Free (liquid) water contents in snow in mm
- Snow melt in mm/time step

3.4.6 The Soil Moisture Routine

The soil moisture accounting of the HBV model is based on statistical distribution of storage capacities in a basin. This is the main part controlling runoff formation. This routine is based on the three parameters, BETA, LP and FC, as shown in the Figure 3.9. BETA controls the contribution to the run off response routine (dUZ) and the increase in soil moisture storage (dSM) for a precipitation or snow melt input of 1 mm into the soil moisture storage. The BETA controls the nature of the equation. If BETA is not equal to 1, the equation will be non linear. Usually, BETA has a value in the range 2-3, making the equation strongly non linear. The structure results a small percentage contribution of run off when the soil moisture is low, and a high contribution when the soil moisture is high.

illustration not visible in this excerpt

Figure 3.9 :Soil moisture routine in the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

Threshold (LP) is a soil moisture value above which evapotranspiration reaches its potential value, and field capacity (FC) is the maximum soil moisture storage in the model. The parameter LP is given as a fraction of FC. The soil moisture is depleted by evapotranspiration. The actual evapotranspiration (EA) is a function of potential evapotranspiration (EP) and relative soil moisture storage (SM/FC). The evapotranspiration in the model is only computed from the snow free part of the catchment. BETA, LP and FC, all are free parameters and only computed from model calibration.

3.4.7 The Runoff Response Routine

The runoff generation routine is the response function which transforms excess water from the soil moisture zone to runoff. It also includes the effect of direct precipitation and evaporation on a part which represents lakes, rivers and other wet areas. The system contains two linear reservoirs called upper zone and lower zone which delay the run off in time and are the origin of the quick (superficial channels) and slow (base-flow) runoff components of the hydrograph. By choosing suitable values for the parameters the model can obtain both a quick response for high flows and slow response for the low flows, as normally seen in observed hydrographs. The upper zone conceptually represents the quick runoff components, both from overland flow and from groundwater drained through more superficial channels, interflow. The lower zone conceptually represents the groundwater and lake storage that contributes to base flow in the catchment.

illustration not visible in this excerpt

Figure 3 . 10 :The runoff response routine in the HBV model (Source: Lecture Note, HPD, NTNU by Prof Ånund)

3.5 LANDPINE

3.5.1 Introduction

LANDPINE is a distributed hydrological rainfall runoff model developed by Trond Rinde to study the effects of land use changes in runoff from catchment. The model was developed during a research program, called HYDRA following the flood occurred in the largest basin in the Norway, Glomma river basin in the spring 1995. The model was implemented within a simulation tool for modeling of hydrological related phenomena, called PINE (Process Integrating Network) which provides a high level of flexibility with respect to representation of the real world systems in simulation setup. It operates in integration with a geological information system, usually used for preparation of input data and analysis and presentation of simulation results. The LANDPINE model is based on the following two requirements:

- An explicit representation of the catchment characteristics that will change if land use is changed

- Not requiring more meteorological data i.e. only precipitation and temperature data

3.5.2 Model Description

The model accounts on a distributed basis for interception in high and low vegetation, storage of water on the ground surface, evapotranspiration, accumulation and melting of snow, infiltration, retention of water in the soil, and generation of surface run off and outflow from the soil. Water movement in rivers and outflows from ground water reservoirs are linked by the help of an aggregated response function. The Figure 3.11 shows the general structure of the LANDPINE model.

illustration not visible in this excerpt

Figure 3.11 :Model structure of LANDPINE (After Rinde T, 1998)

3.5.3 Meteorological Input

Temperature and precipitation are the major input data for the model. These data can be imported to the model either as records of point measurements from a number of gauging stations or as distributed values, generated by a meteorological model. The threshold, tx, temperature is used to classify the type of precipitation as rain or snow and a rain/snow correction factor is used for each cases. Then a spatial interpolation is performed to produce the temperature and precipitation distributions over the catchment by using inverse distance weighting method. Both the temperature and precipitation values are adjusted for difference between local surface elevation (cell elevation) and the corresponding elevation (climatic station) of the input value. A single linear precipitation gradient, pgrad, is used for adjusting the precipitation while for temperatures; different lapse rates with and without precipitation are the wet adiabatic lapse rate represented by the parameter , tcgrad, and the dry adiabatic lapse rate by the parameter, tpgrad, are used for each time steps.

Apart from the meteorological inputs, the distributed inputs required for the LANDPINE are shown in Table 3.1, and these data are generally imported into the model as digital maps.

illustration not visible in this excerpt

Table 3.1 :Distributed input parameters for LANDPINE

illustration not visible in this excerpt

Figure 3.12 :Model Structure for grid cell process in LANDPINE (After Rinde T., 1998)

The following equations are used for temperature correction with elevation for each cell.

The following equations are used for precipitation correction with catch error and elevation correction for each grid.

The model structure for the grid cell processes is shown in Figure 3.12.

3.5.4 High Vegetation

High vegetation is represented in the model in terms of five parameters. These are as follows:

illustration not visible in this excerpt

For the high vegetation, interception capacity and potential evaporation are calculated along the actual values for these responses. Precipitation that is not intercepted is considered as through fall to the ground.

3.5.5 Interception Capacity

For a fully grown forest stand with a complete canopy, the interception capacity (HINTCAP) is calculated as:

illustration not visible in this excerpt

VEGCOR accounts for the fact that a stand may not be fully grown and SESCOR accounts for the seasonal variation in the canopy cover. Depending on the vegetation height, VEGCOR varies according to the relationship.

illustration not visible in this excerpt

VEGCOR = 1.0 represents a fully grown stand. SESCOR may vary from values LAIMIN/LAIMAX to 1.0 and is calculated on the basis of accumulated degree days above a given threshold temperature, tvlow. The equation for calculation of SESCOR is:

illustration not visible in this excerpt

The degree days start to accumulate when temperature rises above tvlow. Then SESCOR attains its maximum value when the accumulated degree days reaches or rises beyond a value given by the parameter tvsum.

3.5.6 Potential Evaporation

Potential evaporation (POTET) is calculated from monthly average values for potential evaporation per day which are provided as input to the model. These values refer to average climatic conditions and a standardized surface type. To account for possible deviations from their average values during simulation period, the values are corrected for deviation between the actual temperature and long term monthly mean temperature by a linear function of TMPCOR, given by the following equation.

etmp is the dependency of potential evaporation on temperature deviation from the long-term monthly mean temperature, temp mnd

In periods with precipitation, an increase in air humidity hence reduction in POTET is assumed and a reduction factor PRCCOR is introduced.

Potential evaporation rates can also be adjusted for deviations in wind speeds from their monthly mean values by applying wind correction factor, WNDCOR and the value is given by:

Kwnd is a factor for dependency of potential evaporation on wind speed deviation from long term monthly mean wind.

illustration not visible in this excerpt

Potential evaporation is also dependent on the vegetation height since the exchange of air masses generally occurs more efficiently high above the ground than the close to it. High vegetation therefore leads to higher potential evaporation that low vegetation. This effect is dealt with an adjustment factor, HGTCOR, which accounts for changes in potential evaporation for deviation of assumed vegetation height, VEGHGT from standard/ Default height for fully grown trees, dveght.

ehg t is the dependency of potential evaporation on VEGHGT deviation from dveght. Hence the final equation for potential evaporation computation for each grid cell is:

Where, epmnd is the input of monthly average potential evaporation for each month.

3.5.7 Actual Interception and Evaporation

Actual evaporation from interception in high vegetation (AEH) is taken as the potential evaporation rate multiplied by the VEGCOV and a correction factor which accounts for reduction in evaporation if the intercepted precipitation is in the form of snow. This factor, SNWCOR, is set to either unity for time steps with temperature higher than tx or to eredsnw less than 1 for time steps with temperature lower than tx. In each time step, incoming precipitation (PREC) is used to fill up the interception storage (HINT) which is depleted by the actual evaporation (AEH). When the interception capacity is reached, excess precipitation falls as through fall (TRUFAL) to ground or the snow surface. The equations for the above describing process are as follows:

illustration not visible in this excerpt

3.5.8 Snow

If local air temperature is lower than the rain /snow threshold, the trough fall from high vegetation goes to increase the snow layer. If the temperature is higher, the through fall instead goes to increase the liquid water content in the snow, or if no snow is present, it passes on to the lower interception storage. In each grid cell, the snow distribution is assumed to be linear. A distribution factor, snwdst, is then used to specify the relative magnitudes of the maximum and minimum storage values in the cell. If the distribution factor is set to unity, the snow pack (SNWPCK) becomes homogeneous. If it is set equal to 2.0, the maximum value (MAXPCK) becomes twice the average value, and the minimum value (MINPCK) becomes zero. If it is set higher than 2.0, only partial snow cover will be simulated. These three conditions are shown in Figure 3.13.

illustration not visible in this excerpt

Figure 3 . 13 : Snow Distribution within area-element (After Rinde T., 1998)

Snow melt (SNWMLT) is calculated on the basis of actual air temperature, TEMP, a threshold for melting, tx, and a melt factor cx, according to degree day principle.

illustration not visible in this excerpt

In forested and partly forested areas the melt factor is reduced. These causes reduced melt intensity and hereby delayed snow melting in forested areas compared to open land. If air temperature is lower than the melt threshold, refreezing of liquid water (SNWFRZ) in the snow is calculated through the use of the parameter, cfr, which accounts for the facts that refreezing occurs at a much lower rate than melting.

illustration not visible in this excerpt

Snow melt as well as refreezing, is assumed to be homogeneous across the snow surface. Melted snow is added to the liquid water content in the snow (SNWWAT). A separate parameter, lwmax, specifies the maximum relative amount of such water that can be withheld in the snow. If the relative amount becomes larger than this fraction, the excess forms outflow from the snow (SNWOUT). Through fall on the snow-free parts contributes directly to SNWOUT. The equations for the above process are given below.

illustration not visible in this excerpt

3.5.9 Low Vegetation and Land Surface

The leaf area index for low vegetation (LAILOW) is the only parameter separately defined for low vegetation. Similar to high vegetation, interception storage is also computed for the low vegetation. The interception capacity (LINTCAP) computation is based on the same principles used in HINTCAP computation but it is lumped together with a wetting storage for the land surface (SRFSTR).

illustration not visible in this excerpt

The lumped storage is filled by outflow from the snow routine. When the storage capacity is exceeded, excess water will go to the soil (TOSOIL). Potential evaporation for this storage is taken as the potential evaporation that was calculated for the high interception storage reduced according to the actual evaporation that has already occurred in the high interception storage (AEH). Actual evaporation from the lower storage (AEL) is assumed to occur at a potential rate as long as water is left in the storage. The equations with this routine are given below.

illustration not visible in this excerpt

3.5.10 Lake

There is different procedure for the computations when there is a lake instead of the land surface. A lake is taken as an infinite reservoir, therefore the lake storage (LAKESTR) represents an excess storage recharged by SNWOUT and depleted by the AEL. However SNWOUT in this case is the direct precipitation falling over the lake surface since there will be no loss of water due to interception or evaporation from the high vegetation. And AEL is computed in the same way as for the low vegetation with the exceptions that SNWCOV and AEH will always be zero.

illustration not visible in this excerpt

The contribution of flow from the lake to the ground water reservoir (LAKERUN) is therefore computed as:

illustration not visible in this excerpt

3.5.11 Root Zone/ Soil Moisture Zone

Infiltration capacity of soil controls the inflow to the root zone .Root zone or soil moisture zone is the layer from which plant roots can extract water during transpiration, its upper boundary is the soil surface, while its lower boundary is indefinite and irregular and can extend to a water table in some cases. Water enters by infiltration and leaves via evapotranspiration and gravity drainage.

In LANDPINE, any outflow from low vegetation and land surface (TOSOIL) which is beyond the infiltration capacity (INFCAP) of the soil will directly contribute surface runoff (SRFRUN) while inflow rates below the INFCAP (INFLTR) will enter the root zone.

The storage of water in the zone (SOLMST) is limited by the field capacity (FLDCAP). If the root-zone is fully saturated, any incoming water will simply pass directly in the form river runoff. If the zone is partially filled, the entering water is split into two fractions. The first will increase moisture content in the root zone itself, and the remaining (SOILOUT) contributes to runoff. The SOILOUT is defined by a relationship as shown below.

illustration not visible in this excerpt

Figure 3 . 14 : SOLMST / FLDCAP vrs SOILOUT / INFILTR curve with different β

Loss of water in the root-zone can only take place through transpiration (AET). Under wet conditions AET is assumed to occur at a potential rate which now is reduced by the actual evaporation that has already taken place in the high and the low interception storage. Then the transpiration rate is proportionally reduced with the soil saturation deficit. However, since there is a certain saturation deficit below which plant uptake of water is significantly reduced, the transpiration rate is first considered when the relative soil moisture content falls below a specified threshold value represented by a factor, lp. An additional reduction factor (HGTCOR), accounts for the lower transpiration intensity for smaller trees that for fully grown stand. These equations representing these processes are given below.

illustration not visible in this excerpt

3.5.12 Surface Flow and Ground Water

Unlike the previous routine, outflow from ground water reservoirs and flow through rivers and lakes is not described in a distributed manner. The sum of soil and surface runoff from all the grid cells are averaged and provided as input to a lumped response routine consisting of two linear tanks. This routine is similar to the one as used in the HBV model. The upper tank accounts for the retention of water in rivers, whereas the lower tank describes outflow from ground water reservoirs.

illustration not visible in this excerpt

Figure 3 . 15 :Response routine for ground water and surface water runoff (After Rinde T., 1998)

Figure 3.15 shows the model structure of response routine for ground water and surface water runoff. The average soil outflow, avgsoilflw, and average surface flow, avgsurflw, from grid cell is computed as:

illustration not visible in this excerpt

When the input of avgsoil flow exceeds the percolation (PERC), the storage in the upper tank (uz) will start to fill and at the same time be drained as low flow. This flow is controlled by low outflow coefficient k1. If uz exceeds a threshold (trsh), a quicker outflow which is controlled by coefficient k2 will start.

illustration not visible in this excerpt

And the storage in the lower tank (lz) is recharged by direct precipitation on lake area and perc. Drainage from this tank will take place through the outgoing base flow whose magnitude is controlled by coefficient, k0.

illustration not visible in this excerpt

Finally, all the three outflows from the two tanks are lumped together with the average surface flow (avgsurflw) to form the total runoff generated from the catchment, runout.

illustration not visible in this excerpt

CHAPTER 4: DATA COLLECTION AND PROCESSING

4.1 GENERAL

Precipitation is the main source of all waters entering in the land and it’s a very complex process to estimate actual runoff from precipitation. So, extensive analysis of metrological data is needed to calculate the actual runoff from any catchment. Hydrologists need to understand how the amount, rate, duration and quality of precipitation are distributed in space and time in order to access, predict, and forecast hydrological responses of a catchment. Because of the almost impossibility to measure precipitation in all the surface of the catchment , only point measurements of precipitations are taken and converted in areal form. So, there is always possibility to occur error due to the complex nature of regional areal distribution of precipitation. The uncertainties associated with a value of regional precipitation consist due to the following errors:

- Errors due to point measurement

- Errors due to uncertainty in converting point measurement to areal precipitation

To overcome the above errors, it is necessary to check the data for its quality, continuity and consistency before its application to the hydrological analysis.

4.2 CATCHMENT CHARACTERISTICS

Modi River is one of the major tributaries of Kali Gandaki River which is one of the major tributary of Gandaki river basin which lies in the central part of Nepal. The river originates from the southern part of Annuparna Himalaya and flows towards south. Due to the advancement of technology, it is becoming very easy to calculate and analyze the data in hydrological analysis. Arc Hydro is one of the tool developed by ESRI, has been used to produce the catchment and hypsographic curve of the study area which is briefly described below.

Digital elevation model (DEM) 1K X 1K datasets and Stream Network are downloaded from the website for Asia Region. Arc GIS with Arc Hydro tools has been used for further processing the DEM and Stream Network. Data sets have been projected with spatial reference WGS_1984_UTM_ZOME_45N, as per catchments location.

Terrain processing tool in Arc Hydro has been used to execute for further processing of catchment delineation and the sequences involved in the process are as follows:

- DEM Reconditioning
- Fill Sinks
- Flow Direction
- Flow Accumulation
- Stream Definition
- Stream Segmentation
- Catchment Grid Delineation
- Catchment Polygon Processing
- Drainage Line Processing
- Adjoint Catchment Processing
- Drainage Point Processing

DEM Reconditioning: This function modifies the DEM by imposing linear features onto it. It is an implementation of the AGREE method developed Centre for Research in Water Resources at the University of Texas at Austin in 1997. The function needs as input a raw DEM and linear feature, river network that both have to be present in the map document.

Fill Sinks: This function fills the sinks in a grid. If a cell is surrounded by cells with higher elevation cells, the water is trapped in that cell and cannot flow. The fill sinks function modifies the elevation value to eliminate these problems.

Flow Direction: This function computes the flow direction for a given grid. The values in the cells of the flow direction grid indicate the direction of the steepest descent from that cell.

Flow Accumulation: This function computes the flow accumulation grid that contains the accumulated number of cells upstream of a cell, for each cell in the input grid.

Stream Definition: This function computes a stream grid contains a value of ‘1’ for all the cells in the input flow accumulation grid that have a value greater than the given threshold. All other cells in the stream grid contain no data.

Stream Segmentation: This function creates a grid of stream segments that have a unique identification. Either a segment may be a head segment, or it may be defined as a segment between two segment junctions. All the cells in a particular segment have the same grid code that is specific to that segment.

Catchment Grid Delineation: This function creates a grid in which each cell carries a value (Grid code) indicating to which catchment the cell belongs. The value corresponds to the value carried by the stream segment that drains that area, defined in the steam segment link grid.

Catchment Polygon Processing: The three functions catchment polygon processing, drainage line processing and adjoint catchment processing convert the raster data developed so far to vector format.

Drainage Line Processing: This function converts the input stream link grid into drainage line features class. Each line in the feature class carries the identifier of the catchment in which it resides.

Adjoint Catchment Processing: This function generates the aggregated upstream catchments from the ‘Catchment’ feature class. For each catchment that is not a head catchment, a polygon representing the whole upstream area draining to its inlet point is constructed and stored in a feature class that has an ‘Adjoint Catchment’ tag. This feature class is used to speed up the point delineation process.

Drainage Point Processing: This function allows generating the drainage points associated to the catchments.

After successful execution of the above function, the Point Delineation as well as Batch point Generation followed by Watershed Processing (Batch Watershed Delineation) tools was used to delineate the catchment. Figure 3.1 is the result of the DEM, and catchment of the project from GIS and Arc Hydro tools.

illustration not visible in this excerpt

Figure 4.1 : Area Elevation curve with DEM and Catchment Delineation of Modi Catchment.

4.2.1 CLIMATIC AND HYDROLOGICAL DATA

Hydro meteorological data, temperature, precipitation and runoff are the main parameters for precipitation run off model. The required data for the study of the catchment is collected from Department of Hydrology and Meteorology, Kathmandu, Nepal. The data from different three meteorological stations Lumle index No.0814, Ghandruk index No. 0821 and Baglung index No.0605 have been used for the study of the catchment. The two stations, Lumle and Ghandruk lie in southern part of the catchment and the Baglung lies about 12 km far from the boarder of the catchment in the south west of the catchment. The location and annual precipitation based on the recorded period is shown in Table 4.1.

illustration not visible in this excerpt

Table 4.1 : Details of meteorological stations (source: Hydrological Estimations in Nepal)

Stream flow measurements records at gauging station No. 406.5 at Jhapre Bagar, Nayapool, in Modi river are available from 1976. But the available data set used for this study is from 1987 to 2000 and they are used for the calibration and validation for both model approaches, HBV and LANDPINE. Runoff data with missing values are discarded for the analysis of the study. The daily discharge data from 1991 to 1995 and from 1997 to 2000 are used for calibration and validation periods respectively.

The meteorological and hydrological data available are shown in time scale in Figure 4.2. The climatic data from different stations with reasonable missing are filled by interpolation, regression and different well known standard approaches. For precipitation data, normal ratio method is used for filling the gaps because of more than 10% variation of annual precipitation among the stations used for this study. The hydrological data used for this study is complete and there is no missing at all.

illustration not visible in this excerpt

Figure 4.2 :Availability of hydro meteorological data in time scale

4.3 PRINCIPLES OF DATA ANALYSIS

4.3.1 Corrections to Point Measurements

Precipitation is the input to the land phase of the hydrologic cycle and its accurate measurement is the essential part of quantitative and qualitative hydrologic analysis. There are many reasons for concerning about the accuracy of precipitation data, and these reasons must be understood and accounted for both scientific and applied hydrologic analysis. Size of orifice, orientation of the plane and height of orifice, application of wind shield, splashing, evaporation from gauge etc are the common factors that usually affect the measurement accuracy of precipitation. So, some corrections are necessary to overcome the effects caused by the above factors.

4.3.2 Estimation of Missing Data

Without complete data set it’s very difficult to get the required output from hydrological analysis. Therefore, complete data sets are required on many variables such as rainfall, stream flow, evapotranspiration and temperature. Unfortunately, records of hydrological processes are usually short and often have missing observations. The existence of data gaps might be attributed to a number of factors such as interruption of measurements because of equipment failure, effects of extreme natural phenomena such as heavy rainfall or landslides or of human-induced factors such as wars and civil unrest, mishandling of observed records by field personnel, or accidental loss of data files in the computer system.

Before using the precipitation data of the stations, it is necessary first to check the data for continuity and consistency. Several approaches are used to estimate the missing values, Station average; Normal ratio, Inverse distance weighting, and Regression methods are commonly used methods to fill the missing data.

In station average method, the missing record is computed as the simple average of the values at the nearby stations. This method is only used when the annual precipitation value at each of the neighboring stations differs by less than 10% from the annual precipitation of the station with missing data. Simple arithmetic average procedure is followed to estimate the missing values.

illustration not visible in this excerpt

The annual average precipitation values are 5284 mm, 1842 mm and 3281 mm in Lumle, Baglung and Ghandruk respectively, i.e. the difference among these values is greater than 10%. Therefore, Normal Ratio method has been used to fill the missing gaps. The missing value can be estimated by weighting the observations at the G gauges by their respective annual average value.

illustration not visible in this excerpt

Simply, MS Excel is used to fill the missing data. There is no precipitation data at 1996 and no temperature data at all for Ghandruk station. So, temperature data are used from only two stations Lumle and Baglung and the Linear Regression method is used to fill two or more missing temperature data in respective stations while for single missing temperature data, simply, interpolation is done. The daily discharge data series has no missing. The discharge data with missing is discarded.

Based upon the availability of discharge data, less missing precipitation data and analyzing precipitation run off response from the available data, the model simulation period is decided. Different alternatives were tested for calibration and the period from 1991 to 1995 found the best among different alternatives.

4.3.3 Checking the Consistency of Point Measurements

If the conditions relevant to the recording station have undergone a significant change during the period of record, inconsistency would arise on the rainfall data of that station. Some of the common causes of inconsistency are:

- Shifting of recording station
- Changes in the ecosystem due to calamities such as deforestation, landslide etc.
- Observational error from both personal and instrumental

Double mass curve technique is the common method for checking the inconsistency of the record. This technique is based on the principle that when each recorded data comes from the same parent population, they are consistent. The curve is a plot on arithmetic graph paper, of cumulative annual precipitation collected at a gauge where measurement conditions have changed significantly against the average of the cumulative annual precipitation for the same period of years collected at several gauges in the same region. The latest record is used as the first entry. If the double mass curve reveals a significant change in slope and it is due to change in measurement condition at a particular station, then the values of the earlier period should be adjusted to be consistent with later period records before computing of areal averages. The adjustment is done by applying correction factor, K, on the records before the slope change given by the following relationship.

4.4 QUALITY CHECK OF AVAILABLE DATA FOR THE STUDY

Two common methods, comparison and double mass curve technique are applied for the quality checking of the data.

[...]