Digital twin can be defined as a virtual representation of a physical asset enabled through data and simulators for real-time prediction, optimization, monitoring, controlling, and improved decision making. Recent advances in computational pipelines, multiphysics solvers, artificial intelligence, big data cybernetics, data processing and management tools bring the promise of digital twins and their impact on society closer to reality. Digital twinning is now an important and emerging trend in many applications. Also referred to as a computational megamodel, device shadow, mirrored system, avatar or a synchronized virtual prototype, there can be no doubt that a digital twin plays a transformative role not only in how we design and operate cyber-physical intelligent systems, but also in how we advance the modularity of multi-disciplinary systems to tackle fundamental barriers not addressed by the current, evolutionary modeling practices. In this work, we review the recent status of methodologies and techniques related to the construction of digital twins mostly from a modeling perspective. Our aim is to provide a detailed coverage of the current challenges and enabling technologies along with recommendations and reflections for various stakeholders.

/pdf/digital-twin-values-challenges-and-enablers-from-a-modeling-3hpoelaogn.pdf

Digital Twin: Values, Challenges and Enablers From a Modeling Perspective

https://agupubs.onlinelibrary.wiley.com/doi/epdf/10.1002/2016RG000549

Seasonal Drought Prediction: Advances, Challenges, and Future Prospects

https://eartharxiv.org/repository/object/149/download/296/

A comprehensive review of deep learning applications in hydrology and water resources

Machine learning in geo- and environmental sciences: From small to large scale

In this paper, we introduce a modular deep neural network (DNN) framework for data-driven reduced order modeling of dynamical systems relevant to fluid flows. We propose various DNN architectures which numerically predict evolution of dynamical systems by learning from either using discrete state or slope information of the system. Our approach has been demonstrated using both residual formula and backward difference scheme formulas. However, it can be easily generalized into many different numerical schemes as well. We give a demonstration of our framework for three examples: (i) Kraichnan-Orszag system, an illustrative coupled nonlinear ordinary differential equation, (ii) Lorenz system exhibiting chaotic behavior, and (iii) a nonintrusive model order reduction framework for the two-dimensional Boussinesq equations with a differentially heated cavity flow setup at various Rayleigh numbers. Using only snapshots of state variables at discrete time instances, our data-driven approach can be considered truly nonintrusive since any prior information about the underlying governing equations is not required for generating the reduced order model. Our a posteriori analysis shows that the proposed data-driven approach is remarkably accurate and can be used as a robust predictive tool for nonintrusive model order reduction of complex fluid flows.In this paper, we introduce a modular deep neural network (DNN) framework for data-driven reduced order modeling of dynamical systems relevant to fluid flows. We propose various DNN architectures which numerically predict evolution of dynamical systems by learning from either using discrete state or slope information of the system. Our approach has been demonstrated using both residual formula and backward difference scheme formulas. However, it can be easily generalized into many different numerical schemes as well. We give a demonstration of our framework for three examples: (i) Kraichnan-Orszag system, an illustrative coupled nonlinear ordinary differential equation, (ii) Lorenz system exhibiting chaotic behavior, and (iii) a nonintrusive model order reduction framework for the two-dimensional Boussinesq equations with a differentially heated cavity flow setup at various Rayleigh numbers. Using only snapshots of state variables at discrete time instances, our data-driven approach can be considered trul...

A deep learning enabler for nonintrusive reduced order modeling of fluid flows

Rapid spatio-temporal flood prediction and uncertainty quantification using a deep learning method

In a rapidly changing environment, a greater concern about the establishment and improvement of drought indices is expected. The main goal of this study is to develop and apply a time-dependent Standardized Precipitation Index (SPIt) that takes account of the possible non-stationary behaviors in precipitation records. Summer precipitation observations (1959 ~ 2011) from 21 raingauge stations in the Luanhe River basin are fitted with non-stationary Gamma distributions respectively by means of the Generalized Additive Models in Location, Scale and Shape (GAMLSS). The temporal variability of the distribution’s parameter (related to the mean) is flexibly described by an optimized polynomial function. Based on the non-stationary distribution, the SPIt is calculated and then employed to assess the spatio-temporal characteristics of summer drought in the basin. Results of the non-stationary modeling indicate an overall decreasing trend in the summer precipitation during 1959 ~ 2011, and especially a significant decrease in the period of 2000 to 2011. The SPIt is found to be more robust and reliable compared with the traditional Standardized Precipitation Index (SPI). Moreover, remarkable difference is observed between the historical drought assessments of SPIt and SPI in the Luanhe River basin, implying that the non-stationarity of hydrological time series cannot be ignored in drought analyses and forecasts. The proposed SPIt method can be a feasible alternative for drought monitoring under non-stationary conditions, intended to provide a valuable reference for further studies.

A Time-Dependent Drought Index for Non-Stationary Precipitation Series

Many drought indices were proposed to describe drought characteristics, but only few had considered environmental changes. In an attempt to incorporate climate change into meteorological drought index, a nonstationary Gamma distribution with climate indices as covariates was developed for fitting precipitation data and then used for calculating a Nonstationary Standardized Precipitation Index (NSPI) in this study. The performances of the NSPI were compared with those of the traditional Standardized Precipitation Index (SPI), showing that the NSPI capable of taking climate variations into account is more robust than the traditional SPI. Focusing on the Luanhe River basin, historical drought events were described and assessed based on the NSPI and traditional SPI. Moreover, drought characteristics, including drought frequency, peak, duration, and magnitude, were calculated by using the two indices. The results in this study indicated that NSPI using climate indices as covariates could capture drought characteristics in the Luanhe River basin, and this new drought index provides a new concept for constructing the drought index that can effectively adapt to a changing environment.

https://agupubs.onlinelibrary.wiley.com/doi/epdf/10.1002/2015JD023920

A Nonstationary Standardized Precipitation Index incorporating climate indices as covariates

https://cronfa.swan.ac.uk/Record/cronfa48945/Download/0048945-21022019132251.pdf

A domain decomposition non-intrusive reduced order model for turbulent flows

Loglinear models for three-dimensional contingency tables was used with data from 21 rainfall stations and 7 hydrometric stations in the Luanhe river basin, northeast China, for short term prediction of drought severity class. Loglinear models were fitted to drought class transitions derived from standardized precipitation index (SPI) and standardized runoff index (SRI) time series to find which series was more suitable for hydrological drought class prediction 1 and 2 months ahead, respectively. Expected frequencies for two consecutive transitions between drought classes were first calculated, and based on this the predicted drought classes 1 and 2 months ahead were obtained. The results showed that despite the contingency tables of drought class transitions presented the maintenance of the precedent drought class, results of three-dimensional loglinear modeling presented good results when comparing predicted and observed drought classes. Only for a few cases predictions did not fully match the observed drought class, mainly for 2-month lead and when the SRI values are near the limit of the severity class predicted by SRI time series. Based on the correlation analysis of SPI and SRI, we presented the well-known method of hydrological drought class prediction by SPI time series. It was found that, using loglinear regression method, the accuracy of predictions for 2-month lead predicted by SPI time series was higher than those predicted by SRI time series. When we divided the SPI and SRI time series into 2 sub-periods (pre- and post-1980 where land cover changed), we got the same drought class prediction as that predicted by the entire SPI and SRI time series, which illustrated that changes in land use did not affect predictions of hydrological drought classes in the Luanhe river basin. It could be concluded that loglinear prediction of drought class transitions is a useful tool for short term hydrological drought warning, and the results could provide significant information for water resources managers and policy makers to mitigate drought effects.

R. Hu

Papers

Rapid spatio-temporal flood prediction and uncertainty quantification using a deep learning method

A Time-Dependent Drought Index for Non-Stationary Precipitation Series

A Nonstationary Standardized Precipitation Index incorporating climate indices as covariates

A domain decomposition non-intrusive reduced order model for turbulent flows

Hydrological Drought Class Transition Using SPI and SRI Time Series by Loglinear Regression