Sea surface temperature (SST) is one of the most important parameters in the global ocean-atmospheric system, changes of which can have profound effects on the global climate and may lead to extreme ...weather events such as droughts and floods. Therefore, predicting the dynamics of future SSTs is of vital importance which can help identify these extreme events and alleviate the losses they cause. In this paper, a machine learning method combining the long short-term memory (LSTM) deep recurrent neural network model and the AdaBoost ensemble learning model (LSTM-AdaBoost) is proposed to predict the short and mid-term daily SST considering that LSTM is good at modelling long-term dependencies but suffers from overfitting, while AdaBoost has strong prediction capability and is not easily overfitted. By combining these two strong and heterogeneous models, the prediction errors related to variance may cancel out each other and the final results can be improved. In this method, the historical time-series satellite data of SST anomaly (SSTA) is used instead of SST itself considering that the fluctuations of SSTs are very small compared to their absolute magnitudes. The seasonality of the SSTA time series is first modelled using polynomial regression and then removed. Then, the deseasonalized time series are used to train the developed LSTM model and AdaBoost model independently. Daily SSTA predictions are made using these two models, and eventually, their predictions are combined as final predictions using the averaging strategy. A case study in the East China Sea that predicts the daily SSTA 10 days ahead shows that the proposed LSTM-AdaBoost combination model outperforms the LSTM and AdaBoost separately, as well as the optimized support vector regression (SVR) model, the optimized feedforward backpropagation neural network model (BPNN), and the stacking LSTM-AdaBoost model (S_LSTM-AdaBoost), when judged using multiple error statistics and from different perspectives. The results suggest that the LSTM-AdaBoost combination model using the averaging strategy is highly promising for short and mid-term daily SST predictions.
•A LSTM-AdaBoost combination method is proposed to predict short and mid-term SST.•36-year satellite-derived time series daily SST data are used.•A case study has been demonstrated in the East China Sea.•The proposed method outperforms LSTM, AdaBoost, SVR, BPNN and S_LSTM-AdaBoost.
Environmental efficiency (EE) assessment is an efficient way to evaluate the degree of coordination between economy and environment. Most of the studies measured country- or region-level EEs, while ...the EE disparities among cities were not well investigated. By incorporating the socioeconomic and remote sensing data, this study measured the static and dynamic EEs of 11 provinces and 131 cities in the Yangtze River Economic Zone (YREZ) in China based on a super efficiency data envelopment analysis (SEDEA) and Malmquist index (MI) methods during 2003–2014. The influential factors of EE imbalance in the YREZ area were explored by the panel tobit model. Results show that large gaps exist in city's environmental efficiency. Cities in the Yangtze River Delta (YRD) show higher EEs than that in the Chengyu Urban Agglomeration (CUA) and Urban Agglomeration in the Middle Reaches of the Yangtze River (UAMR) areas. 15 cities have an EE below 0.2 and only 2 cities above 1 in 2014. The overall average EE exhibited a declining trend during 2003–2014. The number of cities below the average environmental efficiency increased from 70 (53.4%) to 83 (63.4%) over the time period studied. The MI results indicate that management and scale optimization level is the main factor hindering total factor productivity (TFP) growth. The tobit experiment reveals that GDP per capita played a negative impact on EE for most of the YREZ area during 2003–2014. The degree of opening up and industrial structure acted positively on city's environmental efficiency. These conclusions may be a helpful reference for decision makers to coordinate the economy and environment in the YREZ area.
•City-level environmental efficiency was measured in the Yangtze River Economic Zone during 2003–2014.•GDP per capita was found acting negatively on environmental efficiency during this period.•Remote sensing PM2.5 data was incorporated into the undesirable outputs.•The number of cities below the average environmental efficiency increased from 70 (53.4%) to 83 (63.4%).
Spatiotemporal forecasting (STF) extends traditional time series forecasting or spatial interpolation problem to space and time dimensions. Here, we review the statistical, physical and artificial ...intelligence (AI) methods, data and model uncertainties, predictability and future directions for STF problems. Statistical STF methods have limitations in high-level feature extractions and long-term memory modeling. Physical models are computationally intensive and are imperfect in model structure and parameterization. AI models lack the interpretability and require elaborate training but can model complex nonlinear and non-Gaussian problems. Integrating data-driven and physical model-driven methods could facilitate the improvement of interpretability and forecasting accuracy. The predictive uncertainty comes from data and models, which could be measured by probability distribution and Bayesian inference, respectively. The predictive uncertainty is generally missing in AI models and could be resolved by incorporating Bayesian frameworks. The predictability of dynamic earth systems is spatiotemporally heterogeneous and is generally examined by diagnostic and prognostic approaches. Diagnostic methods analyze the predictability empirically from a theoretical perspective, while prognostic methods investigate the predictability through real experiments. Unraveling the predictability in space and time and the predictability sources will greatly improve earth system understanding and operational forecasting development. Current STF systems are largely not user-friendly to provide probabilistic and understandable forecasting services in near real-time. Intelligent STF systems should automatically prepare various data sources, train the models in a self-adaptative way and provide timely predictive information services for users to make decisions. This review provides state-of-the-art advances in forecasting sciences and highlights new directions for new-generation STF systems.
•Review of methods, uncertainties and predictability of spatiotemporal forecasting.•Integration of artificial intelligence and physical model to improve forecasting.•Probabilistic modeling of forecasting uncertainty by Bayesian methods.•Development of intelligent spatiotemporal forecasting systems for decision-making.
Seasonal precipitation forecasting is valuable for regional water management and agricultural food security. Current numerical models have large uncertainty in model structure, parameterization and ...initial conditions. Here, a data-driven multi-model ensemble is constructed using a series of statistical and machine learning methods with varying inputs. Deterministic precipitation forecasts are produced by the weighting of ensemble members using Bayesian model averaging (BMA) and probabilistic forecasts are generated by sampling from BMA predictive probability density function (PDF). Three mathematical metrics are used to evaluate the performance of precipitation forecasts, including Pearson’s correlation coefficient (PCC), root mean square error skill score (RMSESS) and continuous ranked probability skill score (CRPSS). The results demonstrate that the accuracy in the statistical ensemble is significantly higher than the North American multi-model ensemble (NMME) for both deterministic and probabilistic precipitation forecasts, especially at 1-month lead. Statistical models are considerably enhanced by incorporating wavelets, which decomposes the raw precipitation series into several different levels, potentially representing underlying precipitation patterns at different time-frequency scales. Selecting some good ensemble members can improve the ensemble performance, instead of including all the ensemble members with some inefficient models. Overall, the statistical ensemble can be considered as an effective complement of numerical models in both deterministic and probabilistic precipitation forecasts.
In the last half-century, geoscience research has advanced due to multidisciplinary technologies, among which Information and Communication Technology (ICT) has played a vital role. However, ...scientifically organizing these ICTs toward improving geoscience measurements, data processing, and information services has encountered tremendous challenges. This paper reviews a profound revolution in geoscience that has resulted from the Geospatial Sensor Web (GSW), serving as a new cyber-physical spatio-temporal information infrastructure for geoscience on the World Wide Web (WWW). In contrast to previous experiment-based and sensor-based paradigms, the GSW-based paradigm is able to accomplish the following: (1) achieve integrated and sharable management of diverse sensing resources, (2) obtain real-time or near real-time and spatiotemporal continuous data, (3) conduct interoperable and online geoscience data processing and analysis, and (4) provide focusing services with web-based geoscience information and knowledge. As a benefit of the GSW, increasingly more geoscience disciplines are enjoying the value of real-time data, multi-source monitoring, online processing, and intelligence services. This paper reviews the evolution of geoscience research paradigm to demonstrate the scientific background of GSW. Then, we elaborates on four key methods provided by GSW, namely, integrated management, collaborative observation, scalable processing and fusion, and focusing service web capacity. Furthermore, current GSW prototypes and applications for environmental, hydrological, and natural disaster analysis are also reviewed. Moreover, four challenges to the future GSW in geoscience research are identified and analyzed, including integration with the Model Web initiative for sophisticated geo-processing, integration with humans for pervasive sensing, integration with Internet of Things (IoT) to achieve high-quality performance and data mining, and integration with Artificial Intelligence (AI) to provide smart geoservices. We have concluded that GSW has become an indispensable cyber-physical infrastructure, and will play a greater role in geoscience research and application.
Tea is a cash crop that improves the quality of life for people in the Tanuyen District of Laichau Province, Vietnam. Tea yield, however, has stagnated in recent years, due to changes in temperature, ...precipitation, the age of the tea bushes, and diseases. Developing an approach for monitoring tea bushes by remote sensing and Geographic Information Systems (GIS) might be a way to alleviate this problem. Using multi-temporal remote sensing data, the paper details an investigation of the changes in tea health and yield forecasting through the normalized difference vegetation index (NDVI). In this study, we used NDVI as a support tool to demonstrate the temporal and spatial changes in NDVI through the extract tea NDVI value and calculate the mean NDVI value. The results of the study showed that the minimum NDVI value was 0.42 during January 2013 and February 2015 and 2016. The maximum NDVI value was in August 2015 and June 2017. We indicate that the linear relationship between NDVI value and mean temperature was strong with R 2 = 0.79 Our results confirm that the combination of meteorological data and NDVI data can achieve a high performance of yield prediction. Three models to predict tea yield were conducted: support vector machine (SVM), random forest (RF), and the traditional linear regression model (TLRM). For period 2009 to 2018, the prediction tea yield by the RF model was the best with a R 2 = 0.73 , by SVM it was 0.66, and 0.57 with the TLRM. Three evaluation indicators were used to consider accuracy: the coefficient of determination ( R 2 ), root-mean-square error (RMSE), and percentage error of tea yield (PETY). The highest accuracy for the three models was in 2015 with a R 2 ≥ 0.87, RMSE < 50 kg/ha, and PETY less 3% error. In the other years, the prediction accuracy was higher in the SVM and RF models. Meanwhile, the RF algorithm was better than PETY (≤10%) and the root mean square error for this algorithm was significantly less (≤80 kg/ha). RMSE and PETY showed relatively good values in the TLRM model with a RMSE from 80 to 100 kg/ha and a PETY from 8 to 15%.
•The proposed absorber achieves triple-band absorption.•The proposed absorber has a good working polarization angle tolerance.•The proposed absorber are beneficial to the fabrication of the device.
...In this paper, a triple-band perfect metamaterial absorber based on Cu-dielectric-Cu triple-layer nanostructure is reported. The top metal film structure consists of a ring and four pairs of capacitor plates, which has a frequency selection effect, allowing the absorber to resonate in the near infrared range. Theoretical study shows that the absorption of the three absorption peaks (872.54 nm, 1008.69 nm and 1138.62 nm) are 87.1%, 99.9% and 99.6%, respectively. The average absorption is 95.53%, including two perfect absorption peaks. Changing the structural parameters can affect its absorption peaks and resonant wavelengths. At the same time, due to the high symmetry of the absorber, it is not sensitive to the polarization angle and incident angle. Whether in the TE mode or the TM mode, the absorber at a wide incident angle (0-60°) also exhibits good operating angle polarization tolerance. Therefore, the perfect metamaterial absorber we designed can be widely used in sensing.
By means of critical coupling and impedance matching theory, we have numerically simulated the perfect absorption of monolayer graphene. Through the critical coupling effect and impedance matching, ...we studied a perfect single-band absorption of the monolayer graphene and obtained high quality factor (Q-factor = 664.2) absorption spectrum which has an absorbance close to 100% in the near infrared region. The position of the absorption spectrum can be adjusted by changing the ratio between the radii of the elliptic cylinder air hole and the structural period. The sensitivity of the absorber can be achieved S = 342.7 nm/RIU (RIU is the per refractive index unit) and FOM = 199.2 (FOM is the figure of merit), which has great potential for development on biosensors. We believe that our research will have good application prospects in graphene photonic devices and optoelectronic devices.
The combination of critical coupling and coupled mode theory in this study elevated the absorption performance of a graphene-based absorber in the near-infrared band, achieving perfect absorption in ...the double bands (98.96% and 98.22%), owing to the guided mode resonance (the coupling of the leak mode and guided mode under the condition of phase matching, which revealed 100% transmission or reflection efficiency in the wavelet band), and a third high-efficiency absorption (91.34%) emerged. During the evaluation of the single-structure, cross-circle-shaped absorber via simulation and theoretical analysis, the cross-circle shaped absorber assumed a conspicuous preponderance through exploring the correlation between absorption and tunable parameters (period, geometric measure, and incident angle of the cross-circle absorber), and by briefly analyzing the quality factors and universal applicability. Hence, the cross-circle resonance structure provides novel potential for the design of a dual-band unpatterned graphene perfect absorber in the near-infrared band, and possesses practical application significance in photoelectric detectors, modulators, optical switching, and numerous other photoelectric devices.
Downscaling considerably alleviates the drawbacks of regional climate simulation by general circulation models (GCMs). However, little information is available regarding the downscaling using machine ...learning methods, specifically at hydrological basin scale. This study developed multiple machine learning (ML) downscaling models, based on a Bayesian model average (BMA), to downscale the precipitation simulation of 8 Coupled Model Intercomparison Project Phase 5 (CMIP5) models using model output statistics (MOS) for the years 1961–2005 in the upper Han River basin. A series of statistical metrics, including Pearson’s correlation coefficient (PCC), root mean squared error (RMSE), and relative bias (Rbias), were used for evaluation and comparative analyses. Moreover, the BMA and the best ML downscaling model were used to downscale precipitation in the 21st century under Representative Concentration Pathway 4.5 (RCP4.5) and RCP8.5 scenarios. The results show the following: (1) The performance of the BMA ensemble simulation is clearly better than that of the individual models and the simple mean model ensemble (MME). The PCC reaches 0.74, and the RMSE is reduced by 28%–60% for all the GCMs and 33% compared to the MME. (2) The downscaled models greatly improved station simulation performance. Support vector machine for regression (SVR) was superior to multilayer perceptron (MLP) and random forest (RF). The downscaling results based on the BMA ensemble simulation and SVR models were regarded as the best performing overall (PCC, RMSE, and Rbias were 0.82, 35.07, mm and −5.45%, respectively). (3) Based on BMA and SVR models, the projected precipitations show a weak increasing trend on the whole under RCP4.5 and RCP8.5. Specifically, the average rainfall during the mid- (2040–2069) and late (2070–2099) 21st century increased by 3.23% and 1.02%, respectively, compared to the base year (1971–2000) under RCP4.5, while they increased by 4.25% and 8.30% under RCP8.5. Additionally, the magnitude of changes during winter and spring was higher than that during summer and autumn. Furthermore, future work is recommended to study the improvement of downscaling models and the effect of local climate.
PDF
