Reservoir capacity loss and sedimentation assessment of Dholbaha dam located in Punjab, India using remote sensing and bathymetric survey techniques

The hydrological and sedimentation processes of water reservoirs are significantly affected by both natural and human activities within a catchment area (Singh et al. 2023a). In catchments, smaller dams and reservoirs are primarily constructed to mitigate floods and collect rainwater during the rainy season. The collected runoff water serves various purposes, such as irrigation and fish farming (Ninija-Merina et al. 2016). However, over time, the storage capacities of these reservoirs decrease due to the inflow of silt and sediment from their respective catchment areas (Asselman & Middelkoop 1995; Fryirs 2013), with an annual loss rate of storage capacity ranging from 0.5 to 1.0% (Keller et al. 2000; Walling 2006), mainly caused by water-induced erosion (Singh 2019; Singh et al. 2023a, 2023b). Owing to current land use practices, sedimentation is occurring at a notably higher rate. Research by Annandale et al. (2003) indicates a global loss of reservoir storage capacity at approximately 0.8% due to sediment inflow, resulting in an annual financial loss of around 13 billion dollars, equivalent to 45 km³ of water, making sedimentation a crucial concern. The sediment yield of a reservoir significantly affects its physical attributes and operational planning (Kishore et al. 2021; Billi & Spalevic 2022). Therefore, for the achievement of interconnected sustainable development goals, reservoirs play a pivotal role in the water, energy, and food nexus (Quaranta et al. 2021).

Reservoir sedimentation is a complex process due to the wide array of factors influencing it, including alterations in land use within catchments, fluctuations in hydrology affecting water and sediment inflow, variations in sediment particle size, changes in reservoir operations, and the reservoir's shape and size (Strand & Penibertori 1982). According to Ninija-Merina et al. (2016), one of the primary reasons for reservoir sedimentation is soil erosion, resulting from various inadequate land management practices like deforestation, forest fires, unsustainable agricultural practices, and stream bank degradation. Sedimentation poses a significant challenge to water resource development projects, as it reduces both the dead and live storage capacities of reservoirs, undermining their anticipated benefits. Consequently, proper assessment and management of reservoirs are crucial to maintain their full storage capacity and extend their useful lifespan (Goel et al. 2002).

Reservoir capacity can be determined through direct methods (DGPS-based bathymetric or hydrographic surveys) or indirect methods (soil loss models and sediment sampling). Direct methods, such as hydrographic surveys, directly measure sediment accumulation within a reservoir (Vente et al. 2003; Mupfiga et al. 2016). However, these methods are labor-intensive, time-consuming, expensive, and demand skilled personnel and advanced tools (Thomas et al. 2009), often conducted every 5–15 years (Goel et al. 2002). In contrast, indirect methods estimate sediment content without direct measurements. Remote sensing offers a cost-effective approach for estimating reservoir capacity and sedimentation rates (Patni et al. 2017). Due to sediment entrapment concerns, remote sensing has gained attention for reservoir surveying and monitoring (Kummu & Varis 2007; Singh et al. 2021a, 2021b). GIS applications, with their capability to handle complex issues and large datasets, are efficient for three-dimensional planning (Kumar et al. 2016a, 2016b), making the combination of remote sensing and GIS a new indirect approach for sedimentation analysis in reservoirs (Bhavsa & Gohil 2015; Avinash & Chandramouli 2018). Geospatial techniques provide insights into sediment distribution patterns. The major advantage of remote sensing/satellite data is their repetitive coverage over extended periods (Dutta 2016; Pandey et al. 2016). Landsat satellite images have recently been used to estimate water spread areas of various reservoirs using indices like Normalized Difference Water Index (NDWI) or Modified Normalized Difference Water Index (MNDWI) (Ninija-Merina et al. 2016; Behera et al. 2018; Valderrama-Landeros et al. 2018).

The Kandi region, situated in the Shivalik foothills of Indian Punjab, presents one of the most fragile ecosystems of the Himalayan range due to its uneven geographical composition. Around 18% of the total area (2.14 million ha) of the Kandi region is located in Punjab (Kumar et al. 2016a, 2016b). Intensive rainfall exacerbates soil erosion in this region, characterized by rolling terrain and highly erodible soils (Sidhu et al. 2000). Soil loss exceeds 72.7 Mg/ha/year in this region (Singh 2019; Singh et al. 2021a), reaching up to 244 Mg/ha/year in certain catchments (Bhardwaj & Kaushal 2009). Several earthen dams/reservoirs exist in this area, including Chohal dam, Damsal dam, Dholbaha dam, Januari dam, Maili dam, Nara dam, Patiaria dam, Saleran dam, and Thana dam. However, due to high silt inflow relative to soil erosion, these reservoirs experience a yearly decline of 1.0–1.3% in storage capacity (Sur et al. 1999). Consequently, the elevation-area-capacity curves of these reservoirs shift, potentially leading to inaccurate water availability estimates and flawed reservoir operation plans.

As of now, various research studies have been undertaken to conduct analysis on land use and land change, morphometrics, and prioritization within numerous watersheds or reservoir catchments situated within the Kandi region of Punjab. However, a crucial aspect that remains inadequately addressed pertains to the assessment of capacity loss and sedimentation in earthen dams or reservoirs over the past 34 years from their inception. There has been an absence of studies focusing on the evaluation of reservoir capacity loss, utilizing either field survey methodologies such as bathymetric surveys or advanced geospatial techniques. The periodic assessment of reservoir capacity is paramount, as it offers invaluable insights for accurate determination of available water, effective water use scheduling, and optimization of reservoir operations. Such assessments are crucial for informed decision-making and the sustainable management of water resources within the Kandi region of Punjab. Given this context, a study was conducted to (i) estimate the live storage capacity of the Dholbaha reservoir using geospatial techniques, (ii) compare two different methods for estimating reservoir capacity, (iii) compare estimated capacity with bathymetric survey results and update the elevation-area-capacity curve, and (iv) estimate capacity loss and sediment yield using water elevation data (1987–2022) and Landsat satellite imageries (Landsat 7, 8, and 9) for multiple dates across selected years.

The soils of the Kandi region range from loamy sand to sandy loam, with shallow to moderately deep profiles and moderate moisture retention. They are generally unsuitable for cultivation, falling within Class IV to Class VIII land use classifications. The soil characteristics vary within specific ranges for electrical conductivity (EC), pH, and bulk density. The predominant vegetation of this region includes forest trees, shrubs, bushes, and grasses. The landscape is characterized by moderately dense and open forests, with a notable increase in degraded land and stream area over the past two decades, while the proportions of mixed forest and water bodies have decreased.

This region displays a range of soil textures, primarily spanning from loamy sand to sandy loam. These soils typically have shallow to moderately deep profiles, and their capacity for moisture retention falls within the low to medium range. Their suitability for cultivation is limited, falling within various land use classifications, namely Class IV to Class VIII (Sur et al. 1999; Sushanth 2018). The soil characteristics present within specific ranges. The EC of soil which is a crucial aspect affecting its fertility and nutrient availability varies from 0.10 to 0.20 dS/m. The pH, which measures the acidity or alkalinity of the soil, fluctuates between 6.5 and 7.5, representing the soil's chemical composition and its potential impact on plant growth. Additionally, the bulk density, ranging from 1.3 to 1.7 Mg/m³, provides insights into the soil's compactness and porosity.

Regarding vegetation, the prevalent flora in this region comprises forest trees, shrubs, bushes, and grasses. The landscape of the catchment area is characterized by moderately dense and open forests, interspersed with shrubs and grasses. Over a span of 2–3 decades, significant changes have occurred in the land composition of the area. The extent of degraded land has surged by approximately 55%, signifying a concerning trend of environmental degradation. Concurrently, the area occupied by streams has experienced a notable growth of 14%, possibly due to various natural and anthropogenic factors influencing water flow and landscape dynamics. In contrast, the proportions of mixed forest and water bodies have shown a decline, with a 12% decrease in mixed forest coverage and a substantial 36% reduction in water body coverage, indicating transformations in the ecosystem and potential environmental challenges.

Data regarding water surface elevation, design capacity, and bathymetric surveys of the Dholbaha dam reservoir were gathered from the Kandi Canal Irrigation Department, situated in Hoshiarpur, Punjab. Specifically, water surface elevation data spanning a 34-year period (1987–2022) was obtained. In addition, satellite images sourced from Landsat 7, 8, and 9 were obtained from the USGS EarthExplorer platform (https://earthexplorer.usgs.gov) for various dates across the chosen years.

These wavelengths were selected to maximize the ability of water to reflect light. The threshold is set at zero, resulting in the MNDWI value spanning from −1 to 1. A MNDWI value of 0 indicates the presence of water as the cover type, whereas a MNDWI value below 0 indicates the absence of water as the cover type.

The technique of density slicing was employed to delineate areas of water distribution in regions characterized by clear water and distinct land cover features. This methodology involves partitioning the complete set of grayscale pixels within an image into 256 distinct shades. The resultant output comprises a singular value encompassing all gray intensity values associated with water pixels. Through this process, water pixels are segregated from the rest of the image.

Following the identification of minimum and maximum intensity values for water pixels through histogram peak analysis, the image undergoes a thorough density-based segmentation. In cases where uncertainty arises concerning the demarcation between water and other land cover categories, a multifaceted modeling approach can be utilized. This approach aids in pinpointing water distribution areas even when uncertainties surround the delineation between water and other land cover classes.

In order to ascertain the extent of water distribution, a logical assessment is employed to determine the eligibility of pixels in the image for classification as water pixels. For instance, if a pixel satisfies predetermined threshold criteria in both the ratio image and SWIR band, it is designated as a water pixel.

The capacity of the Dholbaha dam reservoir was computed by the following methods: prismoidal and average/mid-area methods.

Reservoir capacity loss, annual depletion rate, and sediment yield can be estimated using the following relationships.

(adapted from Rahmani et al. (2018) and Patro et al. (2022))

(adapted from Rahmani et al. (2018) and Patro et al. (2022))

(adapted from Rahmani et al. (2018) and Patro et al. (2022))

The performance analysis included computation of statistical parameters, namely, mean absolute percentage error (MAE), root mean square error (RMSE), standard deviation (SD), Willmott index of agreement (d) reported by Willmott (1982), and coefficient of determination (R²).

The Dholbaha Dam reservoir, situated in the Kandi region of Indian Punjab, serves multiple purposes such as flood control, irrigation, water storage, fisheries, and more. However, due to intensified land use and associated changes, the rainy season has witnessed an escalation in soil erosion within the reservoir's catchment area. Consequently, an influx of sediment occurs, leading to a reduction in the reservoir's storage capacity (including dead, live, and gross storage) and overall lifespan. To address this issue, implementing conservation strategies holds great potential in significantly curbing sediment inflow to the dam. These measures encompass a range of practices including optimized land use approaches, establishment of vegetative cover, effective gully management, and precise water supply techniques for agriculture through minor diversions. These strategies can be adopted individually or synergistically to effectively mitigate the sediment load that enters the reservoir. In cases where the aforementioned conservation endeavors fall short of rejuvenating a waterbody – be it a lake, dam reservoir, or pond – resorting to dredging emerges as the most viable solution to restore its optimal operational condition. Employing both mechanical and hydraulic dredging techniques, sediment and debris can be removed, paving the way for the revival of the waterbody. Following the dredging process, it is advisable to implement the aforementioned conservation measures, particularly favoring mechanical dredging, to enhance the reservoir's longevity and functionality.

• i.

  The available satellite data (imageries) that can be accessed without restrictions might lack the necessary level of detail to comprehensively capture information concerning smaller features, such as diminutive water bodies. As a consequence, employing satellite data with lower resolution to assess the capacities of minor water bodies like lakes, shallow dams, and ponds could yield inaccurate outcomes.

• ii.

  Another constraint lies in obtaining satellite data with superior or finer resolution. Consequently, utilizing high-quality satellite data to estimate capacities or sedimentation rates could incur substantial expenses. This situation ultimately impacts the feasibility and potential of undertaking such endeavors.

• iii.

  Furthermore, securing satellite data (images) that correspond to the desired water surface elevations of a reservoir might prove challenging, particularly in areas characterized by frequent cloud cover or atmospheric pollution. As a result, the availability of satellite data devoid of cloud interference could also present a constraint in the analysis.

• iv.

  It is only feasible to estimate the live storage capacity of a reservoir, as it is not viable to assess sediment deposition occurring beneath the minimum drawdown level.

• i.

  Understanding the water surface elevations of government-monitored reservoirs and the accessibility of no-cost satellite data streamline the process of estimating reservoir storage capacity, saving both time and labor. In contrast, alternative approaches are expensive, time-intensive, and necessitate specialized expertise.

• ii.

  The remote sensing technique provides reservoir capacity estimates comparable to those obtained through the capital-intensive bathymetric survey method, as demonstrated in the current study.

• i.

  Over time, the quality of satellite data, including both freely accessible and paid sources, is continually improving. Therefore, greater attention should be directed toward the utilization of remote sensing techniques for estimating reservoir water storage capacity. The outcomes yielded by this approach are on par with those from hydrographic/bathymetric surveys. It's important to note that the results of remote sensing methods have the potential to further enhance as data quality continues to increase.

• ii.

  The assessment of reservoir sedimentation can benefit from a collaborative utilization of both remote sensing and bathymetric/hydrographic techniques. Employing remote sensing methods at shorter intervals and conducting bathymetric surveys at longer intervals (around 5–15 years) would be a fitting approach.

• iii.

  Utilizing microwave remote sensing data, consistently accessible throughout the year at frequent intervals, offers the advantage of obtaining cloud-free images. This data remains unaffected by factors such as illumination or weather conditions.

The bathymetric survey indicated a reservoir capacity of 914.0 ham, while remote sensing estimated it at 865.5 ham, both notably lower (by 224.4 and 272.9 ham) than the original design capacity (1,138.40 ham) of the Dholbaha reservoir. Both prismoidal and average/mid-area methods yielded identical live storage capacity results (R² = 1.0), highlighting their equal effectiveness for assessment. The Dholbaha reservoir has experienced significant reductions in storage capacities over a 34-year period, with dead, live, and gross storage diminishing by 81.5, 19.7, and 28.9%, respectively. The annual depletion rates are 2.40, 0.58, and 0.85% for the respective capacities. Notably, the loss in live storage capacity at a water surface elevation of 417.0 m amounts to 224.3 ham (19.7%) over the same timeframe, indicating a sediment yield rate of 1,203.56 m³/km²/year from the catchment. Despite differences in estimated and bathymetric survey-based live storage capacities compared to the original design capacity, there is a close statistical congruence between these approaches. This underscores the accuracy of the model projecting Dholbaha reservoir capacity within the water elevation range of 401.0 m (dead storage) to 417.0 m (live and gross storage). The Dholbaha reservoir has witnessed substantial storage capacity reductions, particularly in live storage, over the past 34 years, primarily attributed to sedimentation. The study highlights the urgent need for effective reservoir sediment management strategies to mitigate further loss of storage capacity and sustain water resources. To address the challenges associated with frequent bathymetric/hydrographic surveys, there is a necessity to explore and invest in advanced remote sensing technologies for accurate and cost-effective monitoring of reservoir capacities. Implementing a proactive sediment management plan to minimize sedimentation and regular monitoring of reservoir capacities are essential for ensuring sustainable water resource management and reservoir longevity.

The authors acknowledge the Kandi Canal Irrigation Department, Hoshiarpur, Punjab for providing water surface elevation, design data, and other related information of the Dholbaha dam.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.