Springer Science and Business Media LLC

Jurassic facility-sliding strata have been identified as a fundamental factor affecting the occurrence of rainfall-reservoir induced landslides in western Hubei Province, China Three Gorges Reservoir area. Regional landslide susceptibility mapping is the most effective method for landslide prediction and mitigation. To solve the current problem of identifying the true landslides and non-landslides, a novel hybrid model based on the two steps self-organizing mapping-random forest (two steps SOM-RF) algorithm is proposed. The identified high and very high susceptibility zones are located within the hydro-fluctuation belt and regions with low altitude based on the datasets before 2014. Deviation and variance of other ten datasets are generated to evaluate the reliability of the maps for the problem of unbalanced sample sizes. Two typical landslides occurred in 2017 have been found in the very high susceptibility zone, which emphasized the validity of susceptibility mapping. To verify the effectiveness of selecting true landslides and non-landslides based on two steps SOM model, recorded landslides and non-landslides randomly chosen from landslide-free areas are put into the single RF model for comparison. The receiver operating characteristic curve and Accuracy index are applied to compare the performance of the landslide susceptibility maps created based on two steps SOM-RF and the single RF model. The results demonstrate that the consideration of true landslides and non-landslides is effective in producing a more accurate landslide susceptibility map with superior prediction skill and higher reliability.

Droughts cause severe damage to the economy, society, and environment. Drought forecasting plays an important role in establishing mitigation drought damage plans. In this study, a hybrid model involving long short-term memory and a climate model (LSTM-CM) is constructed for drought prediction. LSTM-CM was compared to the long short-term model stand-alone (LSTM-SA) and climate prediction model GloSea5 (GS5). The performance of models was evaluated based on the Pearson correlation coefficient (CC), mean absolute error (MAE), root mean squared error (RMSE), and skill score (SS). GS5 displayed physical robustness in predictions and did not reduce the amplitude or shift results. However, GS5 prediction tends to have a large bias caused by the inputs, model structure, and parameters. The MAEs of GS5 at 1, 2 and 3 months (0.41, 0.68, and 0.89) were higher than those of LSTM-SA (0.38, 0.61, and 0.89). The LSTM-SA reduced bias, but predictions were characterized by shifts, small variance, and failure to capture drought occurrences in long-lead-time cases. LSTM-CM yielded enhanced drought predictions by encompassing the low bias of LSTM-SA and the physical process simulation ability of GS5; thus, it inherited the good features of these models and limited the poor features. The SS values based on the CC, MAE, and RMSE of LSTM-CM compared to those of GS5 for 1-, 2-, and 3-month lead time predictions were improved from 29.17 to 54.29, 22.47 to 34.15, and 1.75 to 35.09%, respectively. LSTM-CM can accurately detect drought events and displayed less uncertainty in prediction than LSTM-SA and GS5.

Watershed management, disaster warning, and hydrological modeling require accurate spatiotemporal precipitation data sets. This paper presents a comprehensive assessment of a gauge-satellite-based precipitation product that merges the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR) satellite precipitation product (SPP) and ground precipitation data at 134 rain gauges in the Xijiang River basin, South China. Two regression-based schemes, principal component regression (PCR) and multiple linear regression (MLR), were used to combine the gauge-based precipitation data and PERSIANN-CDR SPP and were compared at daily and annual scales. Furthermore, a hydrological model Variable Infiltration Capacity was used to calculate streamflow and to evaluate the impact of four different precipitation interpolation methods on the results of the hydrological model at the daily scale. The result shows that the PCR method performs better than MLR and can effectively eliminate the interpolation anomalies caused by terrain differences between observation points and surrounding areas. On the whole, the combined scheme consistently exhibits good performance and thus serves as a suitable tool for producing high-resolution gauge-and satellite-based precipitation datasets.

In regions with limited flow and catchment data needed for the configuration and calibration of hydraulic and hydrological models, employing spatial flood modeling and mapping enables authorities to predict the spatial extent and severity of floods. This study leveraged flood inventory data, coupled with various conditional variables, to formulate a novel Ensemble model. This ensemble model combined four hybridized models based on Support Vector Machine (SVM), Naïve Bayes (NB), Decision Classification Tree (DCT), and Artificial Neural Network (ANN), all of which were optimized using the metaheuristic Symbiotic Organisms Search algorithm (SOS). The precision of the flood inundation map generated by the four hybrid models and the ensemble model was assessed using standard metrics. The results demonstrated that the ensemble model outperformed other models, with an accuracy metric of 0.99 Area Under the Curve (AUC) during the training stage and 0.96 during the testing stage. This underscores the effectiveness of the ensemble approach in flood preparedness and response applications. Furthermore, a comparison was conducted, comparing the performance of the developed ensemble model against other studies within the state of West Bengal. The findings highlighted a significant improvement in the ensemble model's performance with an AUC score of 0.96 in validation compared to studies in similar areas within West Bengal with AUC score ranged from 0.73 to 0.92. In conclusion, the methodology employed in this study holds promise for application in other regions worldwide that face challenges related to limited data availability for accurate flood inundation mapping.

The LAgrangian Transport Eulerian Reaction Spatial (LATERS) Markov model was developed to predict upscaled bimolecular reactive transport in a flow around an array of solid cylinders. This method combines the stochastic Lagrangian Spatial Markov model (SMM) to predict transport and a volume averaged reaction rate equation to predict reactions of the form $$A+B\rightarrow \emptyset$$ . Here, we extend the LATERS Markov model to upscale bimolecular reactive transport in a Darcy flow through an idealized heterogeneous porous medium. In agreement with previous literature, the accuracy of the prediction is a function of the Damköhler (Da) numbers, i.e., high Da are more challenging because of incomplete mixing. It was found that a key component which must be incorporated into the upscaled model in these high Da systems is the idea that nearby A and B particles should be more likely to react than those that are farther apart. This is here achieved by appropriately reducing the resolution of the spatial grid employed to resolve the reactive process.

A problem often encountered in agricultural and ecological modeling is to disaggregate daily precipitations into vectors of hourly precipitations used as input values by crop and plant models. A stochastic model for rainfall data, based on transformed censored latent Gaussian process is described. Compared to earlier similar work, our transform function provides an accurate fit for both the body and the heavy tail of the precipitation distribution. Simple empirical relationships between the parameters estimated at different time scales are established. These relationships are used for the disaggregation of daily values at stations where hourly values are not available. The method is illustrated on two stations located in the Paris basin.

Precipitation (P) and temperature (T) are key components of the hydrometeorological system, and their dependence structures exhibit significant dynamic changes, including non-stationary behavior, in response to environmental variations. These changes affect local hydrological processes and impact the predictability of the hydrometeorological system. However, the dynamics of dependence structures among meteorological factors during corresponding and adjacent seasons, as well as their underlying causes, have not been fully revealed. Therefore, this study comprehensively explored the dynamics of the precipitation-temperature dependence structure (PTDS) and temperature-temperature dependence structure (TTDS), and their possible causes. Firstly, non-stationary of PTDS was identified using a copula model. Then the main drivers of PTDS were determined by the random forest (RF) model and variable projection importance (VIP) criteria. These drivers include both conventional factors such as local meteorological factors (e.g., P, T, wind speed (WS), vapor pressure, relative humidity and sunshine duration (SD)) and teleconnection factors (e.g., Sunspots, the Arctic Oscillation, Pacific Decadal Oscillation (PDO), El Niño-Southern Oscillation (ENSO)). Additionally, the normalized difference vegetation index (NDVI) was used to investigate the response of dependence structure to vegetation dynamics. Finally, the ridge regression model was applied to construct driver models for the dynamics of dependence structures. The Loess Plateau was selected as the study area because of its high ecological sensitivity and typical human afforestation activities. The results show that: (1) non-stationarity in the PTDS occurred in different seasons and at various stations; (2) the primary drivers of PTDS and TTDS dynamics are predominantly local meteorological factors; (3) there is a strong correlation between SD and ENSO, and the impacts of PDO on local meteorological factors (WS and T) play a crucial role in driving the PTDS dynamics; and (4) NDVI is the main driver, primarily influencing T and ultimately affecting the dynamics of PTDS and TTDS. These findings suggest that there are significant ecological impacts through radiative or non-radiative feedback mechanisms under warming scenarios. Overall, this study provides new insights into the drivers and mechanisms behind the dynamics of dependence structures among meteorological elements. It contributes to a deeper understanding of the changing local hydrometeorological processes.

Previous studies attempting to quantify the contributions of climate change and human activities to runoff variations in a changing environment have widely focused on an annual scale, while seasonal scales are rarely involved. China is the largest socialist country in the world, and its economic development is affected by its Five-year Plan policy. To this end, the upper Han River Basin, the largest tributary of the Yangtze River Basin, was selected as a case study. A seasonal-scale Budyko framework was extended based on the monthly abcd model in this study for quantifying the effects of climate and human activities changes on seasonal runoff variations, especially exploring the effect of China’s Five-year Plan policy. Results disclose that: (1) the change point in the annual runoff series is 1990, and the monthly abcd model achieves good performance in the prechange and postchange period; (2) the relative contribution of climate change to runoff variations is 62.99%, 70.20%, 87.54%, and 90.30% in spring, summer, autumn, and winter, respectively, indicating that climate change is the dominant factor controlling runoff variations in every season; (3) the contributions of climate change and human activities show obvious dynamic and seasonal characteristics, which are strongly impacted by the implementation of the Five-year Plan policy. Generally, the findings of this study provide valuable references for local reasonable water resource planning and management to make timely and appropriate use of the water supply.

Separability in the context of multidimensional point processes assumes a multiplicative form for the conditional intensity function. This hypothesis is especially convenient since each component of a separable process may be modeled and estimated individually, and this greatly facilitates model building, fitting, and assessment. This is also related to the problem of reduction in the number of dimensions. Following previous approximations to this problem, we focus on the conditional intensity function, by considering nonparametric kernel-based estimators. Our approach calculates thinning probabilities under the conditions of separability and nonseparability and compares them through divergence measures. Based on Monte Carlo experiments, we approximate the statistical properties of our tests under a variety of practical scenarios. An application on modeling the spatio-temporal first-order intensity of forest fires is also developed.