Meteorological Applications最新文献

The Met Office GloSea5-GC2 sub-seasonal-to-seasonal 40-member lagged ensemble consists of members who are up to 10 days different in age such that the between-ensemble-member bias is not internally consistent. Reforecasts tend to be used to convert these ensemble forecasts into anomalies from a normal state. These anomalies are however not that useful for applications where individual ensemble members are needed to drive downstream applications in the hazard and impact space. Here we explore whether there is a way of correcting for the within-ensemble bias without using reforecasts. An investigation into the individual daily precipitation distributions from the JJAS 2019 Indian monsoon season, stratified by forecast horizon, highlights how the distribution changes, and shows that the model distribution is markedly different to the observed. Initial results suggest that it could be better to use recent model forecast distribution(s) as the reference for adjusting the model rainfall accumulations as a function of lead day horizon, that is, not attempting to correct the members to a vastly different (observed) distribution shape, but a more subtle shift towards the model's best guess of reality, rather than reality itself, to remove the between-ensemble-member bias. A combination of Exponential and Generalized Pareto distributions are used for parametric quantile mapping to remove this internal ensemble bias using computationally efficient pre-computed lookup tables. Within- and out-of-sample results for the 2019 and 2020 monsoon seasons show that the method is effective in tightening precipitation gradients, with improvements in spread, accuracy and skill, especially for low accumulations.

Precipitation estimation models are typically sourced by rain gauges, weather radars and satellite observations. A relatively new technique of precipitation estimation relies on the network of Commercial Microwave Links (CMLs) employed for cellular communication networks: the rain-inducted attenuation in the links enables the precipitation estimation. In the paper, it is analysed to what extent the precipitation derived from CML attenuation data is useful in estimation of the precipitation field with the high temporal and spatial resolution required in nowcasting models. Two methods of determination of precipitation along CMLs from attenuation of signal with several frequencies were proposed. Then, in order to generate precipitation field, three approaches for assigning appropriate precipitation values to a specific point or set of pixels along the link are developed and tested. The CML-based estimates are compared with point observations from manual rain gauges and multi-source precipitation fields using daily and half-hourly accumulations. It was found that the CML-based precipitation fields are much worse than radar-derived estimates. At the same time, they had slightly poorer reliability than spatially interpolated telemetric rain gauge data and significantly higher reliability than satellite estimates. Furthermore, the impact of link characteristics, such as length and frequency, on the reliability of CML-based precipitation estimates is analysed.

This paper aims to explore machine learning techniques for post-processing high-resolution Numerical Weather Prediction (NWP) products for the early detection of convection. Data from the Arome Ensemble Prediction System and satellite observations from the Rapidly Developing Thunderstorm (RDT) product by Météo-France are used to train a recurrent neural network model to predict areas of total convection and moderate convection. The learning task is formulated as a binary classification problem using a long short-term memory (LSTM) network architecture. Results from the LSTM model are compared with an object-based probabilistic approach to forecast convection using metrics such as a receiver operating characteristics (ROC) curve, the Brier score and reliability. Results indicate that the LSTM model performs similarly to the object-based probabilistic benchmark when classifying moderate convection areas and shows improved skill when classifying areas of total convective. Finally, the LSTM model results are presented within an air traffic management context to showcase the potential use of machine learning models within an operational application.

Currently, there are three main shortcomings in meteorological drought indices: first, they rely on historical climate probability functions; second, the timescale used in calculations has a certain degree of subjectivity; third, the same index value may correspond to vastly different levels of actual drought in different climate types of regions. The purpose of this article is to establish a meteorological drought index that does not rely on historical meteorological element probability functions. Through theoretical derivation, four drought-level maintenance lines are established on the cumulative precipitation-cumulative water surface evaporation coordinate plane, and the coordinate quadrant is divided into five drought-level areas. Through forward daily rolling accumulation, the maximum distance point is selected from the dynamically changing coordinate points to determine the corresponding cumulative precipitation and cumulative evaporation. The meteorological drought index is established by the distance from the selected coordinate point to each drought-level maintenance line. Using daily precipitation and evaporation data from meteorological observation stations, the index is calculated based on the established meteorological drought index model, and compared with actual drought evolution and drought disaster records. The results show that the index can capture the development of drought well, and its changes are very consistent with drought disaster records. The index is of great significance for drought monitoring or assessment, and can provide guidance for water resource allocation, crop layout, and urban planning. Furthermore, it can also provide a way of thinking that does not rely on historical element probabilities for future drought research.

A large stretch of the east coast of Australia experienced unprecedented rainfall and flooding over a two-week period in early 2022. It is difficult to reliably estimate the likelihood of such a rare event from the relatively short observational record, so an alternative is to use data from an ensemble prediction system (e.g., a seasonal or decadal forecast system) to obtain a much larger sample of simulated weather events. This so-called ‘UNSEEN’ method has been successfully applied in several scientific studies, but those studies typically rely on a single prediction system. In this study, we use data from the Decadal Climate Prediction Project to explore the model uncertainty associated with the UNSEEN method by assessing 10 different hindcast ensembles. Using the 15-day rainfall total averaged over the river catchments impacted by the 2022 east coast event, we find that the models produce a wide range of likelihood estimates. Even after excluding a number of models that fail basic fidelity tests, estimates of the event return period ranged from 320 to 1814 years. The vast majority of models suggested the event is rarer than a standard extreme value assessment of the observational record (297 years). Such large model uncertainty suggests that multi-model analysis should become part of the standard UNSEEN procedure.

As both the population and economic output of India continue to grow, so does its demand for electricity. Coupled with an increasing determination to transition to net zero, India has responded to this rising demand by rapidly expanding its installed renewable capacity: an increase of 60% in the last 5 years has been driven largely by a quintupling of installed solar capacity. In this study, we use broad variety of data sources to quantify potential and realized capacity over India from 1979 to 2022. For potential capacity, we identify spatiotemporal patterns in solar, wind, hydro and wave power. We show that solar capacity factor is relatively homogeneous across India, except over the western Himalaya, and is highest during the pre-monsoon. Wind capacity factor is highest during the summer monsoon, and has high values off the southern coast, along the Western Ghats, and in Gujarat. We argue that wave power could be a useful source of renewable energy for the Andaman and Nicobar Islands, which are not connected to the main Indian power grid. Using gridded estimates of existing installed capacity combined with our historical capacity factor dataset, we create a simple but effective renewable production model. We use this model to identify weaknesses in the existing grid—particularly a lack of complementarity between wind and solar production in north India, and vulnerability to high-deficit generation in the winter. We discuss potential avenues for future renewable investment to counter existing seasonality problems, principally offshore wind and high-altitude solar.

Atmospheric visibility profoundly impacts daily life, and accurate prediction is crucial, particularly in conditions of low visibility characterized by high aerosol loading and humidity. This study employed the WRF-Chem model to simulate a severe wintertime haze pollution episode that transpired from January 17 to 19, 2010, in Central-East China (112–122° E, 34–42° N). The results reveal that excluding aerosol–meteorology interactions led to underestimated PM_2.5 concentrations and relative humidity in comparison with ground-based measurement data, accompanied by a significant overestimation of visibility. Aerosols can engage with meteorological elements, particularly humidity, resulting in positive feedback. Upon considering these feedback interactions, the simulation results showed an increase of 5.17% and 1.99% in PM_2.5 concentration and relative humidity, respectively, compared with the original simulation. This adjustment narrowed the bias between simulated and measured data. The overestimation of simulated visibility was reduced by 16% and 25% for the entire study period and the severe haze pollution period, respectively. These findings underscore the vital role of incorporating aerosol–meteorology interactions in visibility simulations using the WRF-Chem model. Notably, the inclusion of aerosol–meteorological feedback significantly enhances the accuracy of visibility predictions, particularly during heavily polluted periods.

The urban heat island (UHI) effect exacerbates near-surface air temperature (T) extremes in cities, with negative impacts for human health, building energy consumption and infrastructure. Using conventional weather models, it is both difficult and computationally expensive to simulate the complex processes controlling neighbourhood-scale variation of T. We use machine learning (ML) to bias correct and downscale T predictions made by the Met Office operational regional forecast model (UKV) to 100 m horizontal grid length over London, UK. A set of ML models (random forest, XGBoost, multiplayer perceptron) are trained using citizen weather station observations and UKV variables from eight heatwaves, along with high-resolution land cover data. The ML models improve the T mean absolute error (MAE) by up to 0.12°C (11%) relative to the UKV. They also improve the UHI diurnal and spatial representation, reducing the UHI profile MAE from 0.64°C (UKV) to 0.15°C. A multiple linear regression performs almost as well as the ML models in terms of T MAE, but cannot match the UHI bias correction performance of the ML models, only reducing the UHI profile MAE to 0.49°C. UKV latent heat flux is found to be the most important predictor of T bias. It is demonstrated that including more heatwaves and observation sites in training would reduce overfitting and improve ML model performance.

Models have become a key component of scientific hypothesis testing and climate and sustainability planning, as enabled by increased data availability and computing power. As a result, understanding how the perceived ‘complexity’ of a model corresponds to its accuracy and predictive power has become a prevalent research topic. However, a wide variety of definitions of model complexity have been proposed and used, leading to an imprecise understanding of what model complexity is and its consequences across research studies, study systems, and disciplines. Here, we propose a more explicit definition of model complexity, incorporating four facets—model class, model inputs, model parameters, and computational complexity—which are modulated by the complexity of the real-world process being modelled. We illustrate these facets with several examples drawn from ecological literature. Overall, we argue that precise terminology and metrics of model complexity (e.g., number of parameters, number of inputs) may be necessary to characterize the emergent outcomes of complexity, including model comparison, model performance, model transferability and decision support.