Project topics and materials
Get Complete Project Material File(s) Now! »
Hydrodynamic-ecological modeling of lakes and reservoirs
Numerical models have been applied as an effective tool to understand physical, biological and chemical processes in surface freshwaters (Fenocchi et al., 2019; Leon et al., 2011; Toffolon et al., 2014). Models are valuable for lake and reservoir water quality studies. On one hand they can be used to understand ecosystem physical and ecological processes in response to environmental factors. On the other hand, they can serve as a management tool and to forecast water quality in view of possible future climate change scenarios.
Several 1D and 3D hydrodynamic models, coupled to ecological models, have been developed for the study of lake ecosystems. The choice of the spatial dimension of the model should usually depend on the objectives of the study or the questions being addressed and the characteristics of the ecosystem. 1D vertical models with a horizontally averaged output have been frequently used in lake ecosystem studies (Hodges 2009). They require a small computational effort and less input data and are thus considered for long term applications or future projections. However, the need for 3D hydrodynamic models has been recognized in order to accurately capture the physical processes (e.g. vertical and horizontal mixing) (Soulignac et al., 2018).
The application of hydrodynamic models to inland water bodies has been widely used to understand their ecological status (Fadel et al., 2017; Pinardi et al., 2015; Soulignac et al., 2018). Reservoirs exhibit different physical and water quality characteristics from natural lakes (Doubek & Carey, 2017). The river can influence the reservoir water level but also the circulation pattern, and hence strongly impacts the ecosystem responses, while for natural lakes the impact of a stream flow is generally limited (Tufford & McKellar, 1999).
Compared to large lakes and reservoirs, the application of 3D hydrodynamic models to medium and small-size reservoirs with considerable water level fluctuations is seldom seen, especially in semi-arid climates. Reservoir physical dynamics drives the distribution and transport of water quality variables in particular physical mixing or wind induced circulation (Falconer et al., 1991).
The latter can influence chemical and biological properties in particular the distribution of harmful algal blooms and their shoreline accumulation. Water quality is also influenced by thermal stratification acting as a barrier in the warm season and controlling the distribution of nutrients, heat and dissolved substances in the water column.
Prior knowledge about water temperature and currents is crucial for assessing the water quality of reservoir. Several 3D hydrodynamic modeling studies were undertaken in order to understand the hydrodynamics and the thermal regime of reservoirs (e.g. Chanudet et al., 2012). In the tropical Itumbiara reservoir in Brazil (814 km2), an investigation of surface circulation and thermal structure was conducted using the 3D Estuary and Lake Computer Model ELCOM (Curtarelli et al., 2014). The results showed that the surface circulation was mainly wind-driven and that the thermal structure was influenced by several meteorological variables and the river inflow.
In other studies, 3D hydrodynamic and water quality or ecological models were jointly applied for identifying key factors affecting the water quality of reservoirs. They highlighted the importance of 3D hydrodynamic models for capturing temporal and spatial distributions of biogeochemical variables (Romero et al., 2004). For example, describing the transport and dynamics of algal blooms (Mao et al., 2015) and contaminants (Xu et al., 2017) under the influence of hydrodynamics.
Ecological modeling
In the last several decades, the rise in the number of eutrophic lakes and reservoirs affected by harmful algal blooms has led to the use of ecological models for research purposes such as understanding ecological processes in response to hydrodynamics and climate change as well as a predictive tool to support lake management. Most recently, modeling cyanobacteria in aquatic ecosystems has become the subject of several studies due to their negative impacts on water quality and their ability to induce human health risks.
Several ecological models, most of which are mechanistic and dynamic, have been developed to study lake and reservoir eutrophication in combination with 3D hydrodynamic models (Anagnostou et al., 2017). These models usually integrate chemical, physical and biological processes. In a recent review by Vinçon-Leite & Casenave (2019), the application of ecological models has been shown to often prevail across lakes in the United States and Asia with a large surface area. Reservoirs on the other hand received less attention.
In the largest reservoir in Western Europe, the Alqueva (250 km2), located in a semi-arid region of Portugal, a coupled hydrodynamic-eutrophication modeling approach was used to simulate flow, temperature and water quality in order to explain the existence of distinct ecological regions in the reservoir. This spatial and temporal variability was attributed to particular hydrological and geomorphological characteristics in addition to local climate features (Lindim et al., 2011).
Coupling in situ measurements, satellite remote sensing and modeling techniques
The coupling of hydrodynamic-ecological models and satellite remote sensing data is advancing in the field of limnology (Nouchi et al., 2019). Synoptic in situ data acquisition and the examination of its horizontal variability within a short period of time through traditional field monitoring is difficult and expensive. With current research in the remote sensing field, it has become possible to enhance data acquisition. The recent launch of Sentinel-2 satellites provides a greater coverage of water constituents with an increased spatial (10 m) and temporal resolution (5 days) suitable for the calibration/validation of models and the monitoring of small lakes and reservoirs (section 1.2). However, the development and validation of remote sensing algorithms of Sentinel-2 is still required for accurate data assimilation to model applications. Owing to the lack of thermal channels in Sentinel-2 satellites, the TIRS onboard Landsat 8, with an appropriate atmospheric correction algorithm, is a valuable tool for small lakes and reservoirs, providing regular thermal data each 16 days.
Because of the difficulties in conventional monitoring methods, most model validation efforts are restricted to a few observation points. Remote sensing can overcome limitations imposed by this type of monitoring and thus provide synoptic data suitable for a thorough calibration or validation of models. For instance, Allan et al. (2016) used surface temperature data acquired from Landsat thermal imagery to validate a 3D hydrodynamic model of Lake Rotoehu. Yet, using remote sensing data for model calibration or validation is still uncommon for lakes and reservoirs.
Furthermore, remote sensing data can be used as initial data for 3D models. For example, Curtarelli et al. (2015) used remote sensing and 3D hydrodynamic modeling to represent phytoplankton dynamics at the Itumbiara Hydroelectric Reservoir in Brazil. They used remote sensing data to force the hydrodynamic model (generate time series of river inflow temperature) and to investigate the horizontal distribution of phytoplankton.
However, remote sensing does not provide information over the entire vertical domain of the ecosystem. For instance, remote sensing estimations such as water surface temperature and Chl-a concentration are limited to the top 100 μm and to the Secchi depth respectively. Therefore, an assimilation of both in situ and remote sensing data is required into modeling applications.
Further, remote sensing tools do not offer an explanation for the spatial distribution of water constituents shown in remotely sensed images. In this case, numerical models which cover the entire spatial (vertical and horizontal) and temporal domains can provide means to interpret these images. For example, in a study by Hedger et al. (2002), a coupled 3D computational fluid dynamics-ecological model was used to explain the spatial distribution of Chl-a concentration in the remote sensing image of Loch Leven Lake, Scotland. Further, in order to identify the circulation pattern responsible for algal patchiness at Mantua Superior Lake, Italy, Pinardi et al. (2015) used a 3D hydrodynamic model to analyze the distribution of phytoplankton from remote sensing images of Chl-a. Soulignac et al. (2018) showed the capability of 3D modeling and remote sensing for understanding the spatio-temporal heterogeneities of phytoplankton in Lake Geneva.
Table of contents :
Résumé
Abstract
Acknowledgements
Table of contents
List of Figures
List of Tables
General Introduction
Major objectives
General approach
Thesis overview
1. Chapter 1: State of the art
1.1. Cyanobacterial blooms in inland waters: traits and environmental drivers
1.2. Remote sensing of lakes and reservoirs
1.2.1. Overview
1.2.2. Spaceborne sensors
1.2.3. Water surface temperature
1.2.4. Chl-a a concentration
1.3. Hydrodynamic-ecological modeling of lakes and reservoirs
1.3.1. Hydrodynamic modeling
1.3.2. Ecological modeling
1.4. Coupling in situ measurements, satellite remote sensing and modeling techniques
1.5. Previous modeling and remote sensing studies in Karaoun reservoir
2. Chapter 2: Materials and Methods
2.1. Study site
2.2. Data collection
2.2.1. Meteorological and hydrological data
2.2.2. In situ measurements
2.2.3. Laboratory analysis
2.3. Remote sensing of water surface temperature
2.3.1. The radiative transfer equation
2.3.2. The general single channel algorithm
2.3.3. The mono-window algorithm
2.3.4. Data collection and processing
2.4. Remote sensing of Chl-a
2.4.1. Satellite data and pre-processing
2.4.2. Chl-a algorithm
2.5. Remote sensing of transparency
2.5.1. Data collection and image processing
2.5.2. Algorithm development
2.6. 3D hydrodynamic modeling
2.6.1. Model description and configuration
2.6.2. Model simulations
2.6.3. Correction of satellite thermal data
2.7. 3D ecological modeling
2.7.1. Model description and configuration
2.7.2. Oxygen
2.7.3. Nutrients
2.7.4. Description of the BLOOM model
2.7.5. Model parameters
2.7.6. Input data and initial conditions
2.7.7. Model calibration and validation
2.8. Statistical indicators
3. Chapter 3: Field data analysis
3.1. Hydrological and meteorological conditions
3.2. Water temperature and thermal stratification
3.3. Phytoplankton composition
3.4. PC and Chl-a
3.5. Nutrients
3.6. Conclusions
4. Chapter 4: Lake surface temperature retrieval from Landsat-8 and retrospective analysis
4.1. Algorithms performance
4.2. Historical analysis of water surface temperature
5. Chapter 5: Thermal regime of reservoirs: A satellite and 3D modeling approach
5.1. Validation of the hydrodynamic model with field measurements
5.1.1. Water level
5.1.2. Water temperature and thermal stratification
5.2. Validation of the hydrodynamic model with satellite images
5.2.1. Efficiency of the correction of satellite skin temperature with measured sub-surface temperature
5.2.2. Analysis of temperature at the reservoir surface in satellite and model maps
5.3. Discussion
5.3.1. Model simulations and in situ measurements
5.3.2. Landsat 8 water surface temperature
5.3.3. Spatial distribution of surface temperature
6. Chapter 6: 3D ecological modeling at Karaoun Reservoir
6.1. Remote sensing of Chl-a and transparency
6.1.1. Transparency
6.1.2. Chl-a
6.2. Ecological modeling
6.3. Model performance
6.4. Satellite and model simulations of Chl-a concentrations
6.3. Spatial distribution of Chl-a concentrations
7. Conclusions and perspectives
References .
