Chapter 2: Review of Literature……………………………………………………... 9-52
2.9 Reservoir operation study
2.9.5 Models used as combination of two or more methods
In an attempt to utilize the advantages of different methods and to overcome the limitations of a particular method, researchers have also developed some hybrid optimization techniques, which are combinations of two or more algorithms.
Becker et al., (1976) used a monthly model and developed daily and hourly model for the CVP system. The monthly model output was used as an input to the daily model and output of the daily model was used as an input to the hourly model.
Chaturvedi and Srivastava (1981) analyzed six major reservoirs of Narmada basin in India using deterministic LP with simulation model.
Marino and Mohammadi (1983) developed a methodology for the monthly operation of a system of two parallel multipurpose reservoirs. The model employed LP nested in DP. At every stage of DP (i.e. months) a series of LP’s are solved. The objective of LP was to minimize the total releases from the reservoir in each month. The objective of DP was to maximize the weighted sum of monthly water and power production.
2.9.6 Artificial Neural Network (ANN) model
ANN models have also been successfully applied in reservoir operation problems.
Raman and Chandramouli (1996) derived a general operating policy for a reservoir using ANN from the deterministic DP results. They used ANN for inferring optimal release rule conditioned on initial storage, inflows and demands. Operating rule derived by ANN model was compared with a stochastic dynamic programming model, standard operating policy and the operating policy produced by multiple linear regressions from the deterministic DP results. The performance of the ANN model was reported as better than those of the other models.
Cancelliere et. al. (2002) used a neural network approach for deriving irrigation reservoir operating rules. In their approach operating rules were determined as a two step process: first, a dynamic programming technique which determines the optimal releases with minimizing sum of squared deficit as objective function, subject to various constraints was applied. Then the resulting releases from the reservoir were expressed as a function of significant variables by neural networks. They reported improved performance of ANN approach.
Thair et al., (2015), applied ANN model for Mosul-Dam reservoir located on Tigris River.
They concluded that ANN is an important Model for finding the missing data. Also, the ANN technique can accurately predict the monthly Outflow.
Ehsani et al., (2016) developed a new ANN based general reservoir operation system (GROS). Embedding GROS in a water balance model, they analyzed the hydrological impact of dam size as well as their distribution pattern within a drainage basin and concluded that for large-scale studies it is generally acceptable to aggregate the capacity of smaller dams, instead of model a hypothetical single larger dam with the same total storage capacity.
2.9.7 Fuzzy rule based modeling
The concept of fuzzy logic was first introduced by Zadeh (Zadeh, 1965). The fuzzy logic based modeling of a reservoir operation is a simple approach, which operates on an ‘if-then’
principle, where ‘if’ is a vector of fuzzy explanatory variables or premises such as the present reservoir pool elevation, the inflow, the demand, and time of the year. The ‘then’ is a fuzzy consequence such as release from the reservoir (Panigrahi and Majumdar, 2000). The fuzzy logic approach is more flexible and allows incorporation of expert opinions, which could make it more acceptable to operators (Russell and Campbell, 1996). Shrestha et al., (1996)
also confirmed that fuzzy logic is an appropriate tool to consider the impreciseness of variables like inflows, in reservoir operation modeling.
Panigrahi and Majumdar (2000) developed a fuzzy rule based model for operation of the single purpose Malaprabha irrigation reservoir in Karnataka, India. Reservoir storage, inflow, and demands were used as premises and the release as the consequence. They concluded that while a fuzzy rule based model is easy to develop and adopt for operation, it suffers from the curse of dimensionality, and therefore the applications of fuzzy logic to reservoir operation problems may remain limited to single reservoir systems.
Mousavi et al., (2005) presented a dynamic programming fuzzy rule-based (DPFRB) model for optimal operation of reservoir system. In the first step, a deterministic dynamic programming (DP) model was used to develop the optimal set of inflows, storage volumes, and reservoir releases. These optimal values were then used as inputs to a fuzzy rule-based (FRB) model to establish the general operating policies in the second step. The performance of DPFRB was compared to a model, which uses the multiple regression-based operating rules. The results indicated that DPFRB performs well in terms of satisfying the system target performance.
Kamodkar and Regulwar (2013) developed a Fully Fuzzy Linear Programming (FFLP) for the Jayakwadi reservoir stage-II, Maharashtra, India. The objective was to maximize the annual releases for irrigation and hydropower generation. The study demonstrated that use of FFLP in multipurpose reservoir system optimization presents a potential alternative to attain an optimal operating policy.
A Decision Support Model had been evolved by Umadevi et al., (2014) using fuzzy logic with different combinations of inputs for developing rules for the operation of a reservoir in South India under the prevalent varying conditions. The results of the analysis showed that fuzzy-logic can be effectively applied for evolving reservoir operation rules.
2.9.8 Genetic Algorithm (GA) model
Genetic algorithm (GA) derives its concept from Darwin’s theory of survival of the fittest and was first envisaged in 1975 by John Holland (Holland, 1992). Genetic Algorithms use a population of solutions in each iterationinstead of a single solution and hence they are called population-based approaches (Goldberg, 1989). This is the major difference between classical optimization methods and GAs. GAs use objective function information directly and do not require its derivatives or any other auxiliary information.
Oliveira and Loucks (1997) used a GA to evaluate operating rules for multi-reservoir systems demonstrating that GA can be used to identify effective operating policies. They used the real-valued chromosomes containing the coordinates of the points that define the piecewise linear operating rule functions. Their research suggests that GA may be a practical and robust way of evolving operating policies for complex reservoir systems.
Reddy and Kumar (2006) developed a Multi-objective Evolutionary Algorithm (MOEA) to derive a set of optimal operation policies for a multipurpose reservoir system, the Bhadra Reservoir system, in India. This study employs a population based search evolutionary algorithm namely Multi-objective Genetic Algorithm (MOGA). The results obtained using this algorithm were able to offer many alternative policies for the reservoir operator, giving flexibility to choose the best out of them.
Ahmed and Sarma (2005) developed a GA model for deriving the optimal operating policy and compared its performance with that of stochastic dynamic programming (SDP) for a multipurpose reservoir. The objective function of both GA and SDP was to minimize the squared deviation of irrigation release. They found that GA model releases nearer to the required demand and concluded that GA is advantageous over SDP in deriving optimal operating polices.
Jotiprakash and Shanthi (2006) developed a GA model for deriving the optimal operating policy for a multi-purpose reservoir. The objective function was to minimize the squared deviation of monthly irrigation demand deficit along with the deviation in the target storage.
They found that the GA model is advantageous in deriving optimal operating polices.
Mathur and Nikam (2009) used GA to optimize the operation of an existing multipurpose reservoir in India namely, the Upper Wardha reservoir in Maharashtra and to derive reservoir operating rules for optimal reservoir operations. Results showed that, even during low flow condition the GA model can satisfy downstream irrigation demand.
Parmar et al., (2015) applied GA to Sukhi Reservoir project in Gujarat, India to develop a policy for optimizing the release of water for the purpose of irrigation. The fitness function used is to minimize the squared difference between the monthly reservoir release and irrigation demand along with squared deviation in mass balance equation. The results derived by using GA showed that the downstream irrigation demands can be fully satisfied and also considerable amount of water saved.
2.10 Conclusions
The detailed review of the literature indicates that due to climate change, the temperature is increasing gradually and there is a possibility of temperature increase of 1.1⁰C to 6.4⁰C by 2100. This global warming is accelerating the melting of glaciers in the Himalayas, which are melting faster than the global average. However, the studies of climate variability on both short and long time scales to establish climate changes over India have been found to be limited. Brahmaputra, one of the major river systems of the Indian sub-continent which originates in the Himalayas, is expected to be more vulnerable to climate change because of the substantial contribution from snow and glaciers into its runoff at downstream. But, only a little emphasis has been placed on studying the hydrological response of this Himalayan River. The impact of climate change on hydrology of a river can be studied by using General Circulation Models. Many researchers have found the suitability and efficiency of the statistical downscaling technique more acceptable than the dynamic downscaling technique which is computationally very expensive. Comparatively, the ANN model to downscale GCMs has got wide recognition due to its ability to establish non-linear relationship between predictors and predictands.
Various hydrological models have been developed around the world to study the impact of rainfall and change in land use/land cover on the discharge of a river. Different types of models have both advantages and disadvantages. However, it is seen that even though the ANN model cannot represent the physical process of the catchment, yet it can identify the non-linear relationship between runoff and rainfall efficiently.
CHAPTER3
SNOW COVER AREA VARIATION STUDY OF BRAHMAPUTRA RIVER BASIN AND ITS IMPACT ON
DISCHARGE IN THE RIVER BRAHMAPUTRA 3.1 Introduction
River Brahmaputra, one of the major river systems of the Indian subcontinent, which originates in the Himalayas, is expected to be much vulnerable to climate change because of the substantial contribution from snow and glaciers (Singh et al., 1997a; Singh and Jain, 2002). Snow and Glaciers play essential roles on hydrological, geomorphic and ecological processes of high elevation catchments and are particularly sensitive to climate change (Beniston, 2003; Viviroli et al., 2011). Assessing the influence of ice and snow melt water on groundwater recharge, runoff and sediment transport dynamics is essential in order to provide conceptual models useful for correct management of water resources in mountain regions (Huss, 2011; Lebedeva and Semenova, 2011) and also for long-term projections of the potential hydrologic effects of climate change (Jost et al., 2012).
Snow cover area (SCA) has long been recognized as an important hydrologic variable for streamflow prediction (Martinec, 1985; Hall and Martinec, 1985). The presence of snow in a basin strongly affects moisture that is stored at the surface, and is available for future runoff.
During winter, a large extent of the mountainous area of Himalayan river basins is covered by snow, which starts ablating in the spring due to rise in temperature. IPCC (2001a) has indicated that the average global surface air temperature has increased by 0.6 ±0.2°C since the late 19th century and it is projected to increase by 1.4–5.8° C over the period 1990–2100.
Lal and Singh (2001) and Lal (2001) have reported that the average annual mean surface temperature over Indian sub-continent is likely to increase by about 2.7°C and 3.8 °C during the decades of the 2050s and 2080s, respectively. Dyurgerov and Mier (2005) and Prasad et al., (2009) mentioned that in recent years, the glaciers and snowfields of the Himalaya- Karakoram-Hindu Kush (HKH) mountain belts and Tibetan plateau are found to be amongst the fastest receding glacial and snow covers in the world. Again, Kulkarni et al., (2007) and Raina (2009) inferred that simultaneous fragmentation of the glaciers along with glacial retreats has also degraded the total areal coverage of perennial snow and ice in the HKH
region. Covering over 33,000 km2, glaciers constitute an important component of the Himalaya (Kaul, 1999). Monitoring their evolution is a key issue as the melting of all glaciers in central Asia may significantly contribute to ongoing sea level rise (Kaser et al., 2006).
Changes in glacier length, areal extent or mass balance can also be used as climate indicators in a region where climatic series (temperature, precipitation) are rare and the climate change signal is not clear (Roy & Balling, 2005; Yadav et al., 2004). In addition, runoff generated by the melting of these glaciers is an important source of water for the people living in the Himalayan valleys. Measuring ongoing glacier wastage is a first step toward the prediction of future water resources in an area and has, thus, important social and economical impacts (Barnett et al., 2005).
There has been no long-term comprehensive in-situ monitoring of snow/glacial melt contributing to the knowledge of hydrology in the river basins of the Himalayan region. As a result, studies on snow melt/glacial melt that describe a basin's hydrology lack direct evidence and sometimes appear to be inconsistent (Armstrong, 2010; Kaser et al., 2010).
Most of the upper parts of the Brahmaputra catchment is covered by Himalayan snow and have a major effect on the downstream flow characteristics of the river. Along with temperature, other climatic factors such as heavy precipitation, wind speed etc. also have impacts on melting of snow. The effect of wind is important for the melting process of snow or chipped ice (Hasebe and Kumekawa , 1994). According to the result of a laboratory experiment, performed by Hasebe and Kumekawa (1994), the relation between the ratio of volume of snowmelt water to air temperature and wind speed is nearly linear. They also found that peak discharge is sharp as the wind speed increases and Lag time from the beginning of snowmelt runoff to peak discharge is faster according to an increase in wind speed. Rainfall on existing snow cover frequently occurs in spring. At the onset of a rain storm, the snow cover often stores a portion of the rain and consequently attenuates runoff formation. However, if the rain persists and the snow cover becomes saturated, any additional melt water will increase runoff formation.
Many studies stated that the melting of glaciers is a clear indicator of climate change (Xu et al., 2009) and noted that glacier change is the most visible and obvious indicator of changing temperatures (Amstrong, 2010; Winkler et al., 2010). The rising temperature in the Himalayas would affect glacier melt (Bamett et al., 2005). In the present study, the temperature factor has been taken into consideration to study its impact on the change in snow cover area of the Brahmaputra river basin.
A widely cited estimate shows considerable variation in the contribution of melt-water across the river basins fed by the Himalayan glaciers (Eriksson et al., 2009; Xu et al., 2008) although this varies seasonally and spatially. The primary effect of climate warming on runoff is the increased streamflow caused due to larger glacier melt rates (Hock, 2005).The importance of melt-water contribution also varies from basin to basin. It is extremely important for Indus basin and it is important for Brahmaputra basin. But it plays modest roles for Ganges, Yangtze and Yellow Rivers (Immerzeel et al., 2010). By region, melt-water contributes 30 percent to the total water flow in the eastern Himalayas, 50 percent in the central and western Himalayas and 80 percent in Karakoram (Xu et al., 2009). Glacier retreat and the release of freshwater are expected to be a key element in projections of discharge from glacierised catchments over the next decades (Huss et al., 2010; Finger et al., 2012).
Using climate model data as forcing, different studies indicate an increase in discharge in spring due to earlier onset of snowmelt, but a decline later in the year due to reduced glacier extent (Stahl et al., 2008; Huss et al., 2008). Glacial melt water can also have relevant impacts on the hydrological regime of larger watersheds further downstream. Keeping all these factors in mind, the effect of variation in snow cover area on the discharge of the river Brahmaputra has also been analyzed in this study.
Mountain glaciers are used to detect and monitor local climate change in regions not typically monitored by instrumentation, as they are considered to be sensitive indicators of climate (Haeberli et al., 2007). Due to the large extent and difficult accessibility of high mountainous terrain, remote-sensing techniques provide an efficient way to collect data in such regions.
The satellites imageries are well suited for measuring the snow cover because the high albedo of snow gives better contrast with most of the other natural surfaces except clouds (Hall et al., 2002). The remote sensing-based measurements provide an opportunity to achieve large and possibly complete spatial coverage, even in very remote areas. Identification of snow cover using only the visible reflected light may be difficult because many things appear as white, such as clouds or even rocks. However, comparing the reflectance of other wavelength, such as infrared, it is possible to differentiate snow from clouds. Reflectance of snow is higher in the visible (0.5–0.7 µm) wavelengths and has a low reflectance in the shortwave infrared (1–
4 µm) wavelengths (Nolin and Liang, 2000) which enables to distinguish snow from clouds and other non- snow-covered conditions. The ratio between short-wave infrared channel and visible channel was used by Kyle et al., (1978) and Bunting and d'Entremont (1982) for snow cover mapping. This band ratio method was later utilized by Dozier (1989) to map snow cover area in the Sierra Nevada Mountains. This method is known as Normalized Difference
Snow Index (NDSI) method and is one of the very potential methods to map snow cover of an area. This NDSI method is used in the present study to prepare snow maps of the Brahmaputra River basin.
3.2 Materials and Method
3.2.1 Data used
The variation of snow cover area in the Brahmaputra river basin has been studied starting from 2002 to 2015 for four different months viz. January, April, July and October using the MODIS (Moderate Resolution Imaging Spectroradiometer) image, MOD09A1.5 (MODIS/Terra Surface Reflectance 8-Day L3 Global 500m SIN Grid)of 500m resolution consisting of seven bands (band-1 to band-7). The freely available MODIS products, with 500 m spatial resolution can provide a basis for regional snow cover mapping, monitoring and hydrological modelling (Liang et al., 2008). MODIS imageries have successfully been applied in monitoring snow cover from space using the Normalized Difference Snow Index (NDSI). Relative to similar sensors such as the Advanced Very High Resolution Radiometer (AVHRR), the MODIS sensor offers some significant advantages. The two main difficulties in estimating SCA from AVHRR images are obscuring of the ground by cloud cover, and obscuring of snow cover by vegetation canopies (especially forests). The MODIS provides observations at a nominal spatial resolution of 500m versus the 1.1km spatial resolution of the AVHRR. MODIS’s data are available continuously (spatially and temporally) and also has several spectral bands which facility is useful for distinguishing the extent of snow cover (Salomonson and Appel, 2004). Most of the MODIS accuracy assessments reported overall accuracy between 85 and 99% during clear sky conditions (Parajka and Bloschl, 2012).
The climate change scenario data used in this study are based on simulations carried out using General Circulation Models (GCMs) for the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC, 2007). The gridded temperature data of HadCM3 (Hadley center Coupled Model version 3) model of A2 scenario were used to get the average temperature of the upper portion of the Brahmaputra River basin. HadCM3 is a coupled atmosphere-ocean general circulation model (AOGCM) developed at the Hadley Centre in the United Kingdom (Gordon et al., 2000; Collins et al., 2001). In this model, spatial resolution is 2.5˚ X 3.75˚ (latitude by longitude) forming the global grid of 96×73 grid cells with 19 levels. In the oceans, this model has a resolution of 1.25˚ X 1.25˚ (latitude by