|Abstract or Summary
- Numerical models are effective tools for simulating complex physical processes such as hydrodynamic and water quality processes in aquatic systems. The accuracy of the model is dependent on multiple model parameters and variables that need to be calibrated and regularly updated to reproduce changing aquatic conditions accurately. Multi-sensor water temperature observations, such as remote sensing data and in situ monitoring technologies, can improve model accuracy by providing benefits of individual monitoring technology to the model updating process. In contrast to in-situ temperature sensors, remote sensing technologies (e.g., satellites) provide the benefit of collecting measurements with better X-Y spatial coverage. However, the temporal resolution of satellite data is limited comparing to in-situ measurements. Numerical models and all source of observations have large uncertainty coming from different sources such as errors of approximation and truncation, uncertain model inputs, error in measuring devices and etc. Data assimilation (DA) is able to sequentially update the model state variables by considering the uncertainty in model and observations and estimate the model states and outputs more accurately. Data Assimilation has been proposed for multiple water resources studies that require rapid employment of incoming observations to update and improve accuracy of operational prediction models. The usefulness of DA approaches in assimilating water temperature observations from different types of monitoring technologies (e.g., remote sensing and
in-situ sensors) into numerical models of in-land water bodies (e.g., reservoirs, lakes, and rivers) has, however, received limited attention. Assimilating of water temperature measurements from satellites can introduce biases in the updated numerical model of water bodies because the physical region represented by these measurements do not directly correspond with the numerical model's representation of the water column. The main research objective of this study is to efficiently assimilate multi-sensor water temperature data into the hydrodynamic model of water bodies in order to improve the model accuracy. Four specific objectives were addressed in this work to accomplish the overall goal:
* Objective 1: Propose a novel approach to address the representation challenge of model and measurements by coupling a skin temperature adjustment technique based on available air and in-situ water temperature observations, with an ensemble Kalman filter (EnKF) based data assimilation technique for reservoirs and lakes.
* Objective 2: Investigate whether assimilation of remotely sensed temperature observations using the proposed data fusion approach can improve model accuracy with respect to in-situ temperature observations as well as remote sensing data.
* Objective 3: Investigate a global sensitivity analysis tool that combines Latin-hypercube and one-factor-at-a-time sampling to investigate the most sensitive model inputs and parameters in calculating the water age and water temperature of shallow rivers.
* Objective 4: Propose an efficient data assimilation framework to take the advantage of both monitoring technologies (e.g., remote sensing and in-situ measurements) to improve the model efficiency of shallow rivers.
Results showed that the proposed adjustment approach used in this study for four-dimensional analysis of a reservoir provides reasonably accurate surface layer and water column temperature forecasts, in spite of the use of a fairly small ensemble. Assimilation of adjusted remote sensing data using ensemble Kalman Filter technique improved the overall root mean square difference between modeled surface layer
temperatures and the adjusted remotely sensed skin temperature observations from 5.6 °C to 0.51 °C (i.e., 91% improvement). In addition, the overall error in the water column temperature predictions when compared with in-situ observations also decreased from 1.95 °C (before assimilation) to 1.42 °C (after assimilation), thereby, giving a 27% improvement in errors. In contrast, doing data assimilation without the proposed temperature adjustment would have increased this error to 1.98 °C (i.e., 1.5% deterioration). The most effective parameters to calculate water temperature were investigated and perturbed among the acceptable range to create the ensembles. Results show that water temperature is more sensitive to inflow temperature, air temperature, solar radiation, wind speed, flow rate, and wet bulb temperature respectively. Results also show in contrast to in-situ data assimilation, remote sensing data assimilation was able to effectively improve the spatial error of the model. Assimilation of in-situ observation improved the model efficiency at observation site. However, the model error increased by time and after less than two days, the model predictions of updated model were the same as base model before data assimilation. Hence, a maximum acceptable error between model and measurements was defined based on the application of model. Remote sensing data were assimilated into the model as they become available to improve the model accuracy for the entire river. In-situ data were also assimilated into the model when the error between model and observations exceeds the maximum error. Results showed that by assimilation of in-situ data one to three times per day, the average daily error reduced up to 58% comparing to situation that in-situ data were assimilated only once. In addition, the average spatial error reduced from 2.59 °C to 0.66 °C after assimilation of remote sensing data.