The mission of the Climate, Weather and Water Forum (CWWF) is to facilitate an annual dialog among scientists, engineers, students, public and private enterprises and government entities on pressing issues related to climate change, weather extremes, water availability, and sustainability. We aim to:
The CWWF2024 received tremendous support from the HKUST, government institutions, international organizations, and academic journals.
Rivers are the arteries of our planet Earth, which carries water, sediment, nutrient necessary to shape the civilization today. With global warming and increasing anthropogenic pressures such as infrastructural development and consumptive water use, river flow dynamics alongside stream networks may have been significantly perturbed. But currently, a clear picture of the structural changes in river flow dynamics remains illusive, potentially having unresolved implications to the society, ecosystems, and the water cycling. Global river models are expected to serve as the most useful tools to close the knowledge gap, but their coarse spatial resolution and limited ability to represent anthropogenic pressures make them insufficient to capture the fine-scale structural changes. In this talk, I will introduce our research efforts to develop a fine-scale global river model capable of simulating river flow variability for ~3 million river reaches spanning the expansive global river networks. Expanding upon that, increasingly abundant and novel satellite observations of river hydraulic parameters and other geospatial data can be better leveraged to renew our understanding of the network-scale river flow variability in the Anthropocene. Our presentation will be focused on the methodological advancements we have made to improve the capability of global river modeling and that of retrieving river satellite observations. We will also present some initial results to quantify network-scale river flow variability, as well as discussing path forward.
Glaciers, lakes, rivers, and vegetation are particularly sensitive to the change of precipitation in the transition zone between arid and humid regions of the Tibetan Plateau. This study analyzed the spatiotemporal evolution of precipitation and its mechanisms under dynamic and thermodynamic processes based on in situ and satellite precipitation data of the Tibetan Plateau. The results indicate that the proportion of areas with annual precipitation less than 400 mm or more than 2000 mm decreases, while the proportion of areas with annual precipitation between 400 and 2000 mm increases from 1998 to 2018. This shows that precipitation in arid, semi-arid, and semi-humid regions exhibits an increasing trend, while that in southeastern humid regions exhibits a decreasing trend. Moreover, the dynamic process has a stronger effect on the change of precipitation than the thermodynamic one. In particular, the former is twice as strong as the latter, indicating that the atmospheric circulation w
Tropical cyclones (TCs) have received increasing attention because of their catastrophic damage. However, how TC has changed under climate change remains unknown. This study focuses on the change in TC and the associated impacts of extreme rainfall and floods in Mainland Southeast Asia. Here, we reveal that tropical cyclone-associated extreme rainfall and floods have substantially affected society through the high tropical cyclone-induced extreme rainfall and significant human mortality and displacement rates caused by tropical cyclone-induced floods. Moreover, the projected future intensified tropical cyclones indicate increasing tropical cyclone risks. Our findings provide an improved understanding of TCs and their impacts, which helps mitigate potential consequences of global warming in Mainland Southeast Asia and other areas facing similar challenges.
We present a novel remote sensing inversion technology for estimating total nitrogen (TN) in water bodies, addressing the challenge of accurate, large-scale monitoring. Utilizing multi-source data, our approach significantly enhances accuracy and reliability over traditional methods. Applied to the Pearl River Basin, our technique demonstrates two main advantages: a substantial improvement in accuracy, evidenced by an 86% increase in the r2-score of the validation set, and a 10% enhancement in model prediction accuracy through the identification of out-of-distribution data. This results in a model uncertainty below 0.18, showcasing robustness. The technique results in a continuous riverine TN estimation dataset from 2015 to 2023, with a spatial resolution of 5 km and temporal resolution of ~1 day. Our method offers a scalable and accurate tool for monitoring riverine nitrogen concentrations, improving the ability to track and manage water quality on a basin-wide scale. By providing detailed insights into the drivers of total nitrogen concentration in the Pearl River Basin, this study also supports further research on water pollution and ecosystem management.
Weather forecasting is a long-standing computational challenge with significant societal and economic impacts. Recent AI weather forecasting models primarily focus on forecasting tasks, 1) overlooking the inherent data complexities that contain substantial redundant information and noise, 2) neglecting the spatial relationships and solely focusing on temporal dependencies among data points. In this talk, we introduce the W-MAE weather forecasting model, applying pre-training techniques to address data denoising and spatiotemporal dependency modeling. We compare W-MAE with FourCastNet, for meteorological variables, e.g., geopotential height at 500 hPa, our W-MAE model achieves more stable 10-day prediction performance. For diagnostic variable forecasting, i.e., the precipitation forecasting, our W-MAE significantly outperforms FourCastNet (0.80 vs. 0.98) in Anomaly Correlation Coefficient (ACC). The full paper is available at https://github.com/Gufrannn/W-MAE.
Freezing rain is one of the most damaging weather phenomena in winter or early spring in many parts of the world, affecting traffic, power lines and agriculture. Thus, reliable and computationally efficient prediction of its occurrence is urgently needed in weather forecast operations. However, there are different thermodynamic processes that can lead to freezing rain, resulting in unsatisfactory forecasting performance of the state-of-the-art Numerical Weather Prediction (NWP) models. Here a data-driven forecasting method for freezing rain using machine learning technologies is proposed. Observations of weather phenomenon collected from 2515 national weather stations of China for winter of 2016 to 2019 and the corresponding atmospheric predictors derived from ERA5 reanalysis are used. The prediction function is constructed based on the classification and regression tree, and the predicting variables include temporal and vertical profiles of fundamental thermodynamic and kinematic parameters from 500hPa to 1000hPa, with a total dimension of 2304. The LightGBM (Light Gradient Boosting Machine) framework is adopted to train our prediction model and an algorithm-level approach of modifying the loss function is used to address the imbalance of classes to improve forecasting skill. Results show that the data-driven prediction model, namely DDFR (data driven forecast of freezing rain), out-performs the benchmark NWP, i.e., ECMWF IFS product. It’s improvements in terms of TS score range from 120% to 258% depending on different forecast leading times, which range from 0-12 hours. In addition, DDFR is applied in an operational NWP model of China. The problem of domain adaptation is tackled and transfer learning method is employed to adapt the original DDFR to this NWP model. The effectiveness of such adaptation has been demonstrated by its performance on both training and testing datasets.
Heavy rainfall events can lead to natural hazards (e.g. floods and debris flows) and contribute to water resources. The increasing frequency of these events globally, affecting human activities, has spotlighted the need for precipitation research. Accurate and timely quantitative precipitation estimation (QPE) is essential for improving hydrological forecasts and flash flood monitoring and prediction, in order to effectively cope with heavy rainfall events. Weather radar is a crucial tool in rainfall estimation, providing high-resolution estimates in both space and time. Moreover, knowledge of the Raindrop size distributions (DSDs) is central in calculating the bulk rainfall properties and radar variables used for developing the rainfall estimators. However, DSDs vary both spatially and temporally and even in different types of storms and weather systems. Consequently, this study seeks to optimize the rainfall estimators and improve radar QPE using disdrometer observations and dual-polarimetric S-Band radar over Beibu Gulf, South China. This study investigates the precipitation microphysics characteristics and rainfall estimators over Beibu Gulf using the second-generation Particle Size and Velocity (Parsivel2) disdrometer observations and new generation weather radar observations in the warm seasons of 2020–2023. Three distinct storm types, including convective rainfall, squall line, and typhoon, are analyzed based on their rain microphysical properties, radar polarimetric signatures, and radar QPE performance. The findings from this study are expected to offer valuable insights for improving rainfall estimation in South China and other regions where disdrometer and weather radar observations are availab
The terrestrial water storage anomaly, derived from the Gravity Recovery and Climate Experiment (GRACE), presents a remarkable opportunity for extreme weather detection. However, the practical utility of GRACE data is challenged by an 11-month data gap and several months of missing data. To address this limitation, we have developed an innovative Transformer-based deep learning model for data gap-filling. This model incorporates a self-attention mechanism using causal convolution, allowing the neural network to capture the local context of GRACE time series data. It takes into account various factors such as temperature, precipitation, terrestrial water storage, and evapotranspiration. The validation results demonstrate its robustness, with an average root mean square error of 6.05 cm and an R2 = 0.96. Moreover, our approach holds significant potential for surpassing traditional methods in predicting and filling gaps in remote sensing data and gridded observations.
Fine modeling and fast prediction of regional wind field in the middle and upper atmosphere has always been a difficult problem. We designed a neural operator method to solve this problem. We combine the idea of data assimilation with deep learning method to design a regional wind field operator suitable for near space. The annual RMSE of the zonal wind and meridional wind of the operator model at the height of 30km are 0.903 and 0.881, respectively, which is three times that of ConvLSTM. Moreover, we validate the sparse spatio-temporal modeling method of the regional wind field operator at 20/30/40/50km altitude. The result shows that the model is mesh- free, and can get high-precision modeling of different spatio-temporal resolutions, multiple regions and arbitrary positions at one time, which lays an foundation for fine regional modeling and rapid utilization of near space. 2 High-precision spatiotemporal modeling of atmospheric profiles in near space The efficacious forecasting of single-station atmospheric temperature profiles can provide essential support for the structural design and flight missions of spacecrafts in near space. However, empirical models and reference atmospheric models most are calculations of the average state of the atmosphere profiles. Numerical assimilation models require expensive computational costs to improve the accuracy for medium and long-term forecasting. It has been still a challenge to refined predict short-term temperature profiles of near space at a low-cost. We present a temperature profile operator method for refined modeling in the stratosphere by fusing the ability of Long Short-Term Memory (LSTM) networks or its variants- bidirectional LSTM (BiLSTM) to exploit time series correlated information and deep operator networks (DeepONets) to approximate the solution operator of temperature profiles. It consists of three subnetworks. The first subnetwork is used to approximate the discrete temperature profile function, the second net is applied to represent the spatial information of pressure heights, and the third branch is utilized to encode the time domain of the temperature profile operator. We first use the hourly low latitude temperature data (20- 50km) from ERA5 for training, verification and iterative testing in the next 48 hours. The results denote that the temperature profile operator network has a stable and low error of cumulative generalization, and the BiLSTM operator significantly outperform the other models. We also apply two scenarios to testing the refined applicability of the high-latitude temperature profile operator and the mid-latitude wind profile operator in the stratosphere. This work provides a novel perspective for us to study the refined single-station modeling of the upper and middle atmosp
It remains a major challenge to attribute heatwave’s lifecycle characteristics quantitatively to interwoven atmospheric and surface actions. By constructing a process-resolving, energetics-based attribution framework, here we quantitatively delineate the lifecycle of the record-breaking 2022 mega-heatwave over central-eastern China from a local energetics perspective. It is found that the cloudlessness induced radiative heating and atmospheric dynamics dominate the total energy buildup during the developing stage, while the land-atmosphere coupling and atmospheric horizontal advection act most effectively to sustain and terminate the heatwave, respectively. A reduction in anthropogenic aerosols provides a persistent positive contribution during the event, suggesting that pollution mitigation measures may actually increase the amplitudes of future heatwaves. With this framework, initial efforts are also made to unravel culprits in a model’s sub-seasonal prediction of this mega-heatwave.
Weather forecasts are intrinsically uncertain, but the impacts of that uncertainty on air quality forecasts are not explicitly quantified in current air quality forecast systems. We proposed here a surface ozone ensemble forecast system 2DCNN-SOEF, analogous to modern weather ensemble forecast systems, to represent the probability distribution of forecasted surface ozone concentrations given 30 to 50 possible future weather outcomes. The computation costs of this surface ozone ensemble forecast system were greatly reduced using deep learning techniques that emphasized the spatial patterns of weather. We applied 2DCNN-SOEF to cities in Pearl River Delta region, and showed that the surface ozone ensemble forecast system’s accuracy met the Chinese operational requirements. However, half of the ozone forecast error was due to weather forecast uncertainties, which cannot be eliminated entirely even with perfect pollutant emission estimates and chemistry models. This weather-induced innate uncertainty in air quality forecasts should be considered for effective air quality management.
The Pacific Meridional Mode (PMM) plays a critical role in affecting El Niño-Southern Oscillation (ENSO). This study examines the phase asymmetry of PMM events triggered by tropical and extratropical forcings, namely successive and stochastic events, respectively. It is shown that successive events exhibit negative asymmetry due to stronger trigger in the negative phase, while stochastic events display positive asymmetry due to stronger growth in the positive phase. The opposite phase asymmetry of two types of events respectively results in more frequent persistent La Niña events than El Niño events and more frequent episodic El Niño events than La Niña events, which increase ENSO transition complexity. This research provides a comprehensive understanding of PMM asymmetry and reconciles conflicting perspectives from previous studies. Additionally, the newly proposed contribution of positively asymmetric stochastic PMM events to more frequent episodic El Niño events in this study may enhance our comprehension of ENSO transition complexity.
Net ecosystem productivity (NEP) is a critical indicator of the CO2 sequestration capacity of terrestrial ecosystems. Resolving national carbon budgets is critical for informing land-based mitigation policy. Here we coupled Q10 equations into Penman-Monteith-Leuning model and calibrated it using collected NEP data observations from 41 eddy covariance stations located in various terrestrial ecosystems in China. Based on this carbon sink model, daily NEP data during 2003-2023 over China with 1 km resolution were quantified using MODIS satellite images. NEP shows a decreasing trend from southeast to northwest. Summing up NEP from all regions, it is found that in the past 20 years, there are evident variations of carbon sequestration capacity in different terrestrial ecosystems over China. This study optimized the NEP estimation, which can better characterize the distribution of carbon sinks in China's terrestrial ecosystems and provide a scientific foundation for developing regional carbon neutrality schemes.
Climate warming and the associated intensification of extreme climate events (such as droughts, heavy precipitation, and heatwaves) present challenges to plant growth. However, the response of plant growth to these extreme climate events in various growing periods, climate regions, and agricultural land types with different irrigation strategies remains unclear. This study utilizes ten extreme climate indices and six drought indices to predict plant growth outcomes, as indicated by the annual cumulative Gross Primary Production (GPP), across different growing seasons in Europe from 2003 to 2020. Plant growth is influenced by factors such as soil moisture, water demand, temperature sensitivity, growth stage, and irrigation practices. To examine the impact of extreme climate events on plant growth, an explainable LightGBM model is developed. This model elucidates the contribution of such events, and helps to identify their tipping points. The results demonstrate that agricultural drought and extreme absolute temperatures are key predictors in forecasting the annual accumulation GPP. Plant growth shows a high correlation with extreme climate events in arid climates, followed by cold and temperate climates. In arid climates, extreme precipitation, indicated by the maximum consecutive accumulation of precipitation amounts over a 16-day period, is a predominant predictor of annual accumulated GPP. In cold climates, agricultural drought, indicated by surface soil moisture, plays a leading role in the model prediction results. In rainfed cropland and grasslands, extreme climate events have a more pronounced effect on plant growth yield compared to irrigated croplands. The implementation of irrigation strategies involving human intervention helps mitigate the impact of extreme climate events on plant growth outcomes. Multiple extreme climate events have varying impacts on plant growth during different growing seasons. The dominant predictor influencing the prediction results of annual accumulation GPP is primarily early-season agricultural drought, indicating potential drought memory.
In July-August 2022, Yangtze River valley (YRV) experienced unprecedented hot summer, with the number of heatwave days exceeding climatology by four standard deviations. The heatwaves and associated severe droughts affected about 38 million people and caused devastating economic losses of about five billion US dollars. Here we present convergent empirical and modelling evidence to show that the record-breaking Pakistan rainfall, along with the 2022 tripe-dip La Niña, produces anomalous high pressure over YRV, causing intense heatwaves. The La Niña-induced second-highest sea surface temperature gradient in the equatorial western Pacific suppresses western Pacific convection and extends the subtropical high westward. More importantly, the tremendous diabatic heating associated with the unprecedented Pakistan rainfall reinforces the downstream Rossby wave train, extending the upper-level South Asia High eastward and controlling the entire YRV. The overlay of the two high-pressure systems.
Urban flooding poses a great threat to the megacities of the Greater Bay Area, especially under the changing environment. The urban flooding risk management practices are discussed including engineering projects and non-engineering countermeasures, taking Guangzhou as an example. For instance, the urban security assessment is carried out in Guangzhou, under the idea of containing flooding from the sources. Then, challenges requiring to be addressed are explored, based on our communications with the decision-makers. Furthermore, perspectives on future research needs are provided for better flood risk management and flood resilience improvement. This presentation aims to strengthen communications about urban flooding risk management between the academia, the industry, and the government entities, all of which share the ambition to enhance our capacity faced with more and more frequent extreme weather events.
Extratropical cyclone (EC) is a main source of precipitation at midlatitudes. Reanalysis indicates that ECs probably have an increasing contribution to the Antarctic Ice Sheet (AIS) surface mass balance in recent years. However, the change of EC’s contribution to the surface mass balance with warming still remains uncertain. Using a series of simulations with increased sea surface temperature and reduced AIS surface elevation, we found EC tended to move poleward with significantly increased track density around the AIS with warming. Accordingly, EC precipitation over the AIS, mainly in snow, also significantly increased and buffered the melting of the AIS with warming. However, such a buffering started to decrease due to increased AIS surface temperature, runoff and rainfall partly caused by heat transported by ECs, especially in austral summer once surface elevation reduced to 25%. This study highlights that ECs greatly contribute to the potential tipping point of the AIS evolution.
In August 2022, an unprecedented compound heatwave and drought event (CHDE) lasting 24 days occurred in the Yangtze River valley (YRV), leading to severe reduction of crop, fresh water and power supply. Here we construct a joint cumulative probability distribution of heatwave and drought intensity to show that the lowest probability-based index (PI) of 0.06 in 2022 was estimated as a 1-in-662-year event over 1961–2022 climate. We then detected signals of greenhouse gas forcing to the observed PI time series in a generalized extreme value framework, and determined that anthropogenic influence had increased the probability of such CHDE by about 5.28 (2.15–8.29) times compared to the counterfactual climate without anthropogenic influence. Also, the PI presented a persistently decreasing trend under medium emissions, with its value decreasing by about 0.1 to the end of the 21st century relative to the current climate, indicating that CHDE will become more extreme over YRV.
| Poster ID | Presenting Author | Affliation | Poster Title |
|---|
Five out of six La Niña events have lasted two to three years since 1998. Over the past century ten multiyear La Niña (ML) events had an accelerated trend, with half occurring in the past 25 years. These ML events induce catastrophic floods over Australia, Indonesia, tropical South America, and southern Africa and droughts over the southern U. S., equatorial Africa, India, and southeast China. Why so many double and triple La Niña events emerged recently and whether they will become common remains unknown.
We show that ML distinguishes from single-year La Niña by a prominent onset rate, which provides a precursor for predicting its accumulative intensity. The eight ML events after 1970 primarily follow either a super El Niño (SE) or a central Pacific El Niño (CPE), forming two types of ML: SE2ML and CPE2ML. The leading coupling process for ML’s onset and persistence is thermocline feedback in SE2ML and zonal advective and upwelling feedback in CPE2ML. We hypothesize that the historical increase of ML is rooted in the western Pacific (WP) warming. WP warming enhances zonal advective feedback, promoting more frequent SE and CPE events and increasing the odds for ML. It also strengthens thermocline feedback, accelerating CPE2ML’s onset, leading to a sizable heat discharge and a longer recovery. The results from the large-ensemble experiments of the CESM2 model principally support the observed ML-WP warming linkages. More extreme multiyear La Niña will exacerbate adverse socioeconomic impacts if the western Pacific continues to warm relative to the central Pacific.
(Listed in alphabetical order of the last name; To be updated)
Professor, University of Illinois at Urbana-Champaign, United States
Director of Hong Kong Observatory, Hong Kong
Professor, University of Gothenburg, Sweden
Associate Professor, The Hong Kong Polytechnic University, Hong Kong
Professor, Provost of HKUST, Hong Kong
Professor, Chonnam National University, South Korea
Professor, University of Edinburgh, United Kingdom
Professor, Tsinghua University, China
Professor, Nanjing University of Information Science and Technology, China
Professor, Tsinghua University, China
Head of the IRI Climate Group, United States
Lead Meteorologist, National Oceanic and Atmospheric Administration, United States
Professor, Hong Kong University, Hong Kong
Professor, University of Hawaii, United States
Professor, China Institute of Water Resources and Hydropower Research, China
Professor, Tsinghua University, China
Assistant professor, HKUST, Hong Kong
Professor, Southern University of Science and Technology, China
Professor, The Hong Kong Polytechnic University, Hong Kong
Developer, the Pangu team, China
Developer, the Fuxi team, China
Developer, the Fengwu team, China
Forum Chair / Associate Professor, HKUST
Event Coordinator
Technology and Registration
Planning and Promotion
Financial Coordinator
Communication and Secretariat
