Search

Machine learning-based data-driven models have achieved great success since their invention. Nowadays, the artificial neural network (ANN)-based machine learning methods have made great progress than ever before, such as the deep learning and reinforcement learning, etc. In this study, we coupled the ANN with the K-nearest neighbor method to propose a novel hybrid machine learning (HML) hydrological model for flood forecast purpose. The advantage of the proposed model over traditional neural network models is that it can predict discharge continuously without accuracy loss owed to its specially designed model structure. In order to overcome the local minimum issue of the traditional neural network training, a genetic algorithm and Levenberg–Marquardt-based multi-objective training method was also proposed. Real-world applications of the HML hydrological model indicated its satisfactory performance and reliable stability, which enlightened the possibility of further applications of the HML hydrological model in flood forecast problems.

Kan, G., He, X., Ding, L., Li, J., Hong, Y., Ren, M., Lei, T., Liang, K., Zuo, D., & Huang, P. (March-April, 2017). Daily streamflow simulation based on the improved machine learning method. Water Technology and Sciences (in Spanish), 8(2), 51-60. Daily streamflow simulation has usually been implemented by conceptual or distributed hydrological models. Nowadays, hydrological data, which can be easily obtained from automatic measuring systems, are more than enough. Therefore, machine learning turns into an effective and popular tool which is highly suited for the streamflow simulation task. In this paper, we propose an improved machine learning method referred to as PKEK model based on the previously proposed NU-PEK model for the purpose of generating daily streamflow simulation results with better accuracy and stability. Comparison results between the PKEK model and the NU-PEK model indicated that the improved model has better accuracy and stability and has a bright application prospect for daily streamflow simulation tasks.

The Generalized Likelihood Uncertainty Estimation (GLUE) method has been thrived for decades, huge number of applications in the field of hydrological model have proved its effectiveness in uncertainty and parameter estimation. However, for many years, the poor computational efficiency of GLUE hampers its further applications. A feasible way to solve this problem is the integration of modern CPU-GPU hybrid high performance computer cluster technology to accelerate the traditional GLUE method. In this study, we developed a CPU-GPU hybrid computer cluster-based highly parallel large-scale GLUE method to improve its computational efficiency. The Intel Xeon multi-core CPU and NVIDIA Tesla many-core GPU were adopted in this study. The source code was developed by using the MPICH2, C++ with OpenMP 2.0, and CUDA 6.5. The parallel GLUE method was tested by a widely-used hydrological model (the Xinanjiang model) to conduct performance and scalability investigation. Comparison results indicated that the parallel GLUE method outperformed the traditional serial method and have good application prospect on super computer clusters such as the ORNL Summit and Sierra of the TOP500 super computers around the world. [ABSTRACT FROM AUTHOR]

Small watersheds are ideal objects for studying the evolution of hydrological and water resources at small scales. Whether precipitation products can meet the runoff simulation of small watersheds is the main purpose of this study. With NOAA-CPC-US precipitation as a reference, in nine small watersheds of the United States, accuracy of the precipitation products such as PERSIANN, PERSIANN-CDR, TRMM-3B42V7, GPM-IMERG, StageIV, and ERA5 is analyzed. By driving the CREST hydrological model using these datasets, the runoff simulation effects were evaluated. Result shows the precipitation products match the NOAA-CPC-US from high to low in the order of StageIV, PERSIANN-CDR, GPM-IMERG, PERSIANN, ERA5, and finally TRMM-3B42 V7. These datasets have relatively low accuracy in the northern high latitude area and the western mountains, while accuracy is better in the central, southern, and eastern parts of the United States. In the runoff simulation effectiveness evaluation, the daily runoff of watersheds is simulated during the same verification period. The results show that: NOAA-CPC-US and StageIV have good simulation effect in small watersheds. In the northern and western United States, the PERSIANN, PERSIANN-CDR, GPM-IMERG, and ERA5 for runoff simulation should be used with caution. TRMM-3B42V7 is not suitable for runoff simulation in small watershed. [ABSTRACT FROM AUTHOR]

Spatial and temporal characteristics of temperature extremes have been investigated in Huang-Huai-Hai (HHH) region based on the daily series of temperature observations from 162 meteorological stations. A total of 11 indices were used to assess the changes of temperature pattern. Linear trend analyses revealed that the daily maximum temperature (TXx) increased at α = 0.05 level with a magnitude of 0.15 °C per decade on the regional scale during the period of 1961-2014. More pronounced warming trend of the daily minimum temperature (TNn) was detected at a rate of 0.49 °C per decade (α = 0.01 level). Consequently, a decreasing trend of the temperature range of TXx and TNn (extreme temperature range) was observed. The frequency of hot days (TXf90) and annual average of warm events (warm spell duration indicator, WSDI) showed significant increasing trends, while that of cold nights (TNf10) and cold events (cold spell duration indicator, CSDI) showed opposite behaviors. Both warm winter (W-W) and hot summer (H-S) series displayed significant increasing trends at α = 0.01 confidence level. The cold winter (C-W) series showed a decreasing trend at α = 0.01 confidence level, while the cool summer (C-S) series showed a nonsignificant decreasing trend that is not passing the 90% confidence level (α = 0.1). Abrupt increments of warm­related extremes (TXx, TXf90, WSDI) have been detected since 1990s, and a steadily decreasing trend of cold related extremes (TNf10, CSDI) was found since 1970s. Ten hot summers out of 11 and nine warm winters out of 10 occurred after 1990s. Altitude has a large impact on spatial pattern of extreme temperature indices, and the urban heat island effect also has an impact on amplitude of variation in extreme temperature. Trend magnitudes are significantly larger at sites with high altitudes for warm­related indices (TXx, TXf90, WSDI), while those involving cold-related indices (TNn, TNf10) are remarkably larger for stations with low altitudes. [ABSTRACT FROM AUTHOR]

The temporal variations and spatial patterns of drought in Shandong Province of Eastern China were investigated by calculating the standardized precipitation evapotranspiration index (SPEI) at 1-, 3-, 6-, 12-, and 24-month time scales. Monthly precipitation and air temperature time series during the period 1960-2012 were collected at 23 meteorological stations uniformly distributed over the region. The non-parametric Mann-Kendall test was used to explore the temporal trends of precipitation, air temperature, and the SPEI drought index. S-mode principal component analysis (PCA) was applied to identify the spatial patterns of drought. The results showed that an insignificant decreasing trend in annual total precipitation was detected at most stations, a significant increase of annual average air temperature occurred at all the 23 stations, and a significant decreasing trend in the SPEI was mainly detected at the coastal stations for all the time scales. The frequency of occurrence of extreme and severe drought at different time scales generally increased with decades; higher frequency and larger affected area of extreme and severe droughts occurred as the time scale increased, especially for the northwest of Shandong Province and Jiaodong peninsular. The spatial pattern of drought for SPEI-1 contains three regions: eastern Jiaodong Peninsular and northwestern and southern Shandong. As the time scale increased to 3, 6, and 12 months, the order of the three regions was transformed into another as northwestern Shandong, eastern Jiaodong Peninsular, and southern Shandong. For SPEI-24, the location identified by REOF1 was slightly shifted from northwestern Shandong to western Shandong, and REOF2 and REOF3 identified another two weak patterns in the south edge and north edge of Jiaodong Peninsular, respectively. The potential causes of drought and the impact of drought on agriculture in the study area have also been discussed. The temporal variations and spatial patterns of drought obtained in this study provide valuable information for water resources planning and drought disaster prevention and mitigation in Eastern China. [ABSTRACT FROM AUTHOR]

The influence of initial state variables on flood forecasting accuracy by using conceptual hydrological models is analyzed in this paper and a novel flood forecasting method based on correction of initial state variables is proposed. The new method is abbreviated as ISVC (Initial State Variable Correction). The ISVC takes the residual between the measured and forecasted flows during the initial period of the flood event as the objective function, and it uses a particle swarm optimization algorithm to correct the initial state variables, which are then used to drive the flood forecasting model. The historical flood events of 11 watersheds in south China are forecasted and verified, and important issues concerning the ISVC application are then discussed. The study results show that the ISVC is effective and applicable in flood forecasting tasks. It can significantly improve the flood forecasting accuracy in most cases. [ABSTRACT FROM AUTHOR]

In the field of hydrological modelling, the global and automatic parameter calibration has been a hot issue for many years. Among automatic parameter optimization algorithms, the shuffled complex evolution developed at the University of Arizona (SCE-UA) is the most successful method for stably and robustly locating the global “best” parameter values. Ever since the invention of the SCE-UA, the profession suddenly has a consistent way to calibrate watershed models. However, the computational efficiency of the SCE-UA significantly deteriorates when coping with big data and complex models. For the purpose of solving the efficiency problem, the recently emerging heterogeneous parallel computing (parallel computing by using the multi-core CPU and many-core GPU) was applied in the parallelization and acceleration of the SCE-UA. The original serial and proposed parallel SCE-UA were compared to test the performance based on the Griewank benchmark function. The comparison results indicated that the parallel SCE-UA converged much faster than the serial version and its optimization accuracy was the same as the serial version. It has a promising application prospect in the field of fast hydrological model parameter optimization. [ABSTRACT FROM PUBLISHER]