Real-time flood forecasting updating is essential in improving the forecasting performance and preventing flood damages. The advanced dynamic system response curve (DSRC) method has been validated to be effective by adjusting precipitation based on simulated streamflow errors. However, the time lag between input-output signals is not explicitly considered in the original DSRC, resulting in the problem that the most recent precipitation information is not utilized in updating the forecasting. Moreover, regularization techniques are normally introduced in DSRC to ensure the numerical stability of error estimation, however, the commonly used Ridge estimator can result in excessive adjustment of precipitation. To address these critical issues, we proposed an improved precipitation adjustment framework (DSRC-ARMA) that integrated the DSRC method and the autoregressive-moving average (ARMA) model, such that the most recent precipitation information can be used for a complete precipitation adjustment. Moreover, alternative regularized estimators (i.e., the Lasso and Elastic Net estimators) were introduced and cross-compared to prevent the excessive adjustment issue. The performance of the proposed framework was evaluated in two basins in China. The results showed that the DSRC-ARMA method outperformed the original DSRC method in terms of overall goodness-of-fit (e.g., Nash-Sutcliffe efficiency improved from 0.94 f 0.03 to 0.95 f 0.04 and 0.89 f 0.05 to 0.91 f 0.05, respectively in Dapoling (DPL) and Jianyang (JY) basin) and particularly capturing the peak flows (relative error of peak flow decreased from 13.6 f 7.3 % to 5.2 f 3.7 % and from 10.1 f 7.8 % to 5.9 f 3.5 % in DPL and JY, respectively). For different regularized estimators, the Ridge estimator was most suitable for the rainfall events without intermittent non-rainfall time segments (due to its veracity feature); while the Lasso estimator performed better for intermittent rainfall events, due to its feature of sparsity that can confine non-rainfall period errors to be zeros and thus avoid excessive adjustment. Overall, the proposed precipitation adjustment framework holds the potential to enhance the real-time flood forecasting accuracy, thereby offering a valuable approach for flood mitigation.

Forecasting landslide deformation is challenging due to influence of various internal and external factors on the occurrence of systemic and localized heterogeneities. Despite the potential to improve landslide predictability, deep learning has yet to be sufficiently explored for complex deformation patterns associated with landslides and is inherently opaque. Herein, we developed a holistic landslide deformation forecasting method that considers spatiotemporal correlations of landslide deformation by integrating domain knowledge into interpretable deep learning. By spatially capturing the interconnections between multiple deformations from different observation points, our method contributes to the understanding and forecasting of landslide systematic behavior. By integrating specific domain knowledge relevant to each observation point and merging internal properties with external variables, the local heterogeneity is considered in our method, identifying deformation temporal patterns in different landslide zones. Case studies involving reservoir-induced landslides and creeping landslides demonstrated that our approach (1) enhances the accuracy of landslide deformation forecasting, (2) identifies significant contributing factors and their influence on spatiotemporal deformation characteristics, and (3) demonstrates how identifying these factors and patterns facilitates landslide forecasting. Our research offers a promising and pragmatic pathway toward a deeper understanding and forecasting of complex landslide behaviors. (c) 2025 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/ 4.0/).

Uncertainty plays a key role in hydrological modeling and forecasting, which can have tremendous environmental, economic, and social impacts. Therefore, it is crucial to comprehend the nature of this uncertainty and identify its scope and effects in a way that enhances hydrological modeling and forecasting. During recent decades, hydrological researchers investigated several approaches for reducing inherent uncertainty considering the limitations of sensor measurement, calibration, parameter setting, model conceptualization, and validation. Nevertheless, the scope and diversity of applications and methodologies, sometimes brought from other disciplines, call for an extensive review of the state-of-the-art in this field in a way that promotes a holistic view of the proposed concepts and provides textbook-like guidelines to hydrology researchers and the community. This paper contributes to this goal where a systematic review of the last decade's research (2010 onward) is carried out. It aims to synthesize the theories and tools for uncertainty reduction in surface hydrological forecasting, providing insights into the limitations of the current state-of-the-art and laying down foundations for future research. A special focus on remote sensing and multi-criteria-based approaches has been considered. In addition, the paper reviews the current state of uncertainty ontology in hydrological studies and provides new categorizations of the reviewed techniques. Finally, a set of freely accessible remotely sensed data and tools useful for uncertainty handling and hydrological forecasting are reviewed and pointed out.

In contrast to boreal winter when extratropical seasonal predictions benefit greatly from ENSO-related teleconnections, our understanding of forecast skill and sources of predictability in summer is limited. Based on 40 years of hindcasts of the Canadian Seasonal to Interannual Prediction System, version 3 (CanSIPSv3), this study shows that predictions for the Northern Hemisphere summer surface air temperature are skillful more than 6 months in advance in several midlatitude regions, including eastern Europe-Middle East, central Siberia-Mongolia-North China, and the western United States. These midlatitude regions of statistically significant predictive skill appear to be connected to each other through an upper-tropospheric circumglobal wave train. Although a large part of the forecast skill for the surface air temperature and 500-hPa geopotential height is attributable to the linear trend associated with global warming, there is signifi- cant long-lead seasonal forecast skill related to interannual variability. Two additional idealized hindcast experiments are performed to help shed light on sources of the long-lead forecast skill using one of the CanSIPSv3 models and its uncoupled version. It is found that tropical ENSO-related sea surface temperature (SST) anomalies contribute to the forecast skill in the western United States, while land surface conditions in winter, including snow cover and soil moisture, in the Siberian and western U.S. regions have a delayed or long-lasting impact on the atmosphere, which leads to summer forecast skill in these regions. This implies that improving land surface initial conditions and model representation of land surface processes is crucial for the further development of a seasonal forecasting system.

A landslide involves the downward movement of a mass of rock, debris, earth, or soil. Landslides happen when gravitational forces and other types of shear stresses on a slope surpass the shear strength of the materials. Additionally, landslides can be triggered by processes that weaken the shear strength of the slope's material. Shear strength primarily depends on two factors such as frictional strength, which is the resistance to movement between the interacting particles of the slope material, and cohesive strength, which is the bonding between those particles. A landslide is a terrible natural disaster that causes much damage to both human life and the economy. It often occurs in steep mountainous areas or hilly regions, ranging in scale from medium to large. It progresses slowly (20-50 mm/year), but when it occurs, it can move at a speed of 3 m/s. Therefore, early detection or prevention of this disaster is an essential and significant task. This paper developed a method to collect and analyze data, with the purpose of determining the possibility of landslide occurrences to reduce its potential losses. center dot The proposed method is convenient for users to grasp information of landslide phenomenon. center dot A machine learning model is applied to forecast landslide phenomenon. center dot Internet of things (IoT) system is utilized to manage and send a warning text to individual email address and mobile devices.

Climate change has accelerated the frequency of catastrophic wildfires; however, the drivers that control the time-to-recover of forests are poorly understood. We integrated remotely sensed data, climate records, and landscape features to identify the causes of variability in the time-to-recover of canopy leaf area in southeast Australian eucalypt forests. Approximately 97% of all observed burns between 2001 and 2014 recovered to a pre-fire leaf area index (+/- 0.25 sd) within six years. Time-to-recover was highly variable within individual wildfires (ranging between = 5 years), across burn seasons (90% longer January to September), and year of fire (median time-to-recover varying four-fold across fire years). We used the logistic growth function to estimate the leaf area recovery rate, burn severity, and the long-term carrying capacity of leaf area. Time-to-recover was most correlated with the leaf area recovery rate. The leaf area recovery rate was largest in areas that experienced high burn severity, and smallest in areas of intermediate to low burn severity. The leaf area recovery rate was also strongly accelerated by anomalously high post-fire precipitation, and delayed by post-fire drought. Finally we developed a predictive machine-learning model of time-to-recover (R2: 0.68). Despite the exceptionally high burn severity of the 2019-2020 Australian megafires, we forecast the time-to-recover to be only 15% longer than the average of previous fire years. Australian eucalypt forests have evolved different strategies to recover from fire. While the meteorological drivers of bushfire are reasonably well understood, the various processes explaining how long a forest takes to recover from fire are not. We investigated a range of static (landscape) and dynamic (vegetation condition or meteorological) factors that could influence how long a forest's canopy leaf area would take to recover from fire. Time-to-recover after fire is highly variable, ranging from less than 1 year to more than 5 years even within an individual burn location. More intense fires cause greater forest canopy damage and generally (but not always) lead to longer recovery times, whereas wetter conditions after the fire can accelerate recovery. Using these factors and others, we developed a model capable of predicting the time-to-recover and applied it to the 2019-2020 Australian megafires. Our analysis suggests the canopy damage caused by these fires was far more severe than fires in years prior. This would normally lead to a prolonged time-to-recover, however we predict that anomalously high rainfall in the year following the fires will shorten recovery time, compensating for the high burn severity. Ultimately we predict the time-to-recover will be only slightly longer than average. Pre-fire leaf area, burn severity, and post-fire meteorological conditions combine to determine time-to-recover after fire Large geographic variation in time-to-recover can be explained by mean climate and landscape differences Time-to-recover can be predicted with high accuracy using information limited to the first year following fire

An unprecedented heat wave occurred over the Pacific Northwest and southwest Canada on 25-30 June 2021, resulting in all-time temperature records that greatly exceeded previous record maximum temperatures. The impacts were substantial, including several hundred deaths, thousands of hospitalizations, a major wildfire in Lytton, British Columbia, Canada, and severe damage to regional vegetation. Several factors came together to produce this extreme event: a record-breaking midtropospheric ridge over British Columbia in the optimal location, record-breaking midtropospheric temperatures, strong subsidence in the lower atmosphere, low-level easterly flow that produced downslope warming on regional terrain and the removal of cooler marine air, an approaching low-level trough that enhanced downslope flow, the occurrence at a time of maximum insolation, and drier-than-normal soil moisture. It is shown that all-time-record temperatures have not become more frequent and that annual high temperatures only increased at the rate of baseline global warming. Although anthropogenic warming may have contributed as much as 1 degrees C to the event, there is little evidence of further amplification from increasing greenhouse gases. Weather forecasts were excellent for this event, with highly accurate predictions of the extreme temperatures. SIGNIFICANCE STATEMENT: This paper describes the atmospheric evolution that produced an extreme heat wave over the Pacific Northwest during June 2021 and puts this event into historical perspective.

The rapid decline in groundwater around the world poses a significant challenge to sustainable agriculture. To address this issue, agricultural managed aquifer recharge (Ag-MAR) is proposed to recharge the aquifer by artificially flooding agricultural lands using surface water. Ag-MAR requires a carefully selected flooding schedule to avoid affecting the oxygen absorption of crop roots. However, current Ag-MAR scheduling does not take into account complex environmental factors such as weather and soil oxygen, resulting in crop damage and insufficient recharging amounts. This paper proposes MARLP, the first end-to-end data-driven control system for Ag-MAR. We first formulate Ag-MAR as an optimization problem. To that end, we analyze four-year in-field datasets, which reveal the multi-periodicity feature of the soil oxygen level trends and the opportunity to use external weather forecasts and flooding proposals as exogenous clues for soil oxygen prediction. Then, we design a two-stage forecasting framework. In the first stage, it extracts both the cross-variate dependency and the periodic patterns from historical data to conduct preliminary forecasting. In the second stage, it uses weather-soil and flooding-soil causality to facilitate an accurate prediction of soil oxygen levels. Finally, we conduct model predictive control (MPC) for Ag-MAR flooding. To address the challenge of large action spaces, we devise a heuristic planning module to reduce the number of flooding proposals to enable the search for optimal solutions. Real-world experiments show that MARLP reduces the oxygen deficit ratio by 86.8% while improving the recharging amount in unit time by 35.8%, compared with the previous four years.

Reliable subseasonal forecasts of high summer temperatures would be very valuable for society. Although state-of-the-art numerical weather prediction (NWP) models have become much better in representing the relevant sources of predictability like land and sea surface states, the subseasonal potential is not fully realized. Complexities arise because drivers depend on the state of other drivers and on interactions over multiple time scales. This study applies statistical modeling to ERA5 data, and explores how nine potential drivers, interacting on eight time scales, contribute to the subseasonal predictability of high summer temperatures in western and central Europe. Features and target temperatures are extracted with two variations of hierarchical clustering, and are fitted with a machine learning (ML) model based on random forests. Explainable AI methods show that the ML model agrees with physical understanding. Verification of the forecasts reveals that a large part of predictability comes from climate change, but that reliable and valuable subseasonal forecasts are possible in certain windows, like forecasting monthly warm anomalies with a lead time of 15 days. Contributions of each driver confirm that there is a transfer of predictability from the land and sea surface state to the atmosphere. The involved time scales depend on lead time and the forecast target. The explainable AI methods also reveal surprising driving features in sea surface temperature and 850 hPa temperature, and rank the contribution of snow cover above that of sea ice. Overall, this study demonstrates that complex statistical models, when made explainable, can complement research with NWP models, by diagnosing drivers that need further understanding and a correct numerical representation, for better future forecasts.

Human activities have substantially altered present-day flow regimes. The Headwater Area of the Yellow River (HAYR, above Huanghe'yan Hydrological Station, with a catchment area of 21,000 km(2) and an areal extent of alpine permafrost at similar to 86%) on the northeastern Qinghai-Tibet Plateau, Southwest China has been undergoing extensive changes in streamflow regimes and groundwater dynamics, permafrost degradation, and ecological deterioration under a warming climate. In general, hydrological gauges provide reliable flow records over many decades and these data are extremely valuable for assessment of changing rates and trends of streamflow. In 1998-2003, the damming of the Yellow River by the First Hydropower Station of the HAYR complicated the examination of the relations between hydroclimatic variables and streamflow dynamics. In this study, the monthly streamflow rate of the Yellow River at Huanghe'yan is reconstructed for the period of 1955-2019 using the double mass curve method, and then the streamflow at Huagnhe'yan is forecasted for the next 20 years (2020-2040) using the Elman neural network time-series method. The dam construction (1998-2000) has caused a reduction of annual streamflow by 53.5-68.4%, and a more substantial reduction of 71.8-94.4% in the drier years (2003-2005), in the HAYR. The recent removal of the First Hydropower Station of the HAYR dam (September 2018) has boosted annual streamflow by 123-210% (2018-2019). Post-correction trends of annual maximum (Q(Max)) and minimum (Q(Min)) streamflow rates and the ratio of the Q(Max)/Q(Min) of the Yellow River in the HAYR (0.18 and 0.03 m(3).(-)s(-1).yr(-1) and -0.04 yr(-1), respectively), in comparison with those of precorrection values (-0.11 and -0.004 m(3).s(-1).yr(-1) and 0.001 yr(-1), respectively), have more truthfully revealed a relatively large hydrological impact of degrading permafrost. Based on the Elman neural network model predictions, over the next 20 years, the increasing trend of flow in the HAYR would generally accelerate at a rate of 0.42 m(3).s(-1).yr(-1). Rising rates of spring (0.57 m(3).s(-1).yr(-1)) and autumn (0.18 m(3).s(-1).yr(-1)) discharge would see the benefits from an earlier snow-melt season and delayed arrival of winter conditions. This suggests a longer growing season, which indicates ameliorating phonology, soil nutrient availability, and hydrothermal environments for vegetation in the HAYR. These trends for hydrological and ecological changes in the HAYR may potentially improve ecological safety and water supplies security in the HAYR and downstream Yellow River basins.