Precipitation Nowcasting Research Articles

Hierarchical Partitioning Forecaster

In this work we consider a new family of algorithms for sequential prediction, Hierarchical Partitioning Forecasters (HPFs). Our goal is to provide appealing theoretical - regret guarantees on a powerful model class - and practical - empirical performance comparable to deep networks - properties at the same time. We built upon three principles: hierarchically partitioning the feature space into sub-spaces, blending forecasters specialized to each sub-space and learning HPFs via local online learning applied to these individual forecasters. Following these principles allows us to obtain regret guarantees, where Constant Partitioning Forecasters (CPFs) serve as competitor. A CPF partitions the feature space into sub-spaces and predicts with a fixed forecaster per sub-space. Fixing a hierarchical partition H and considering any CPF with a partition that can be constructed using elements of H we provide two guarantees: first, a generic one that unveils how local online learning determines regret of learning the entire HPF online; second, a concrete instance that considers HPF with linear forecasters (LHPF) and exp-concave losses where we obtain O(klogT) regret for sequences of length T where k is a measure of complexity for the competing CPF. Finally, we provide experiments that compare LHPF to various baselines, including state of the art deep learning models, in precipitation nowcasting. Our results indicate that LHPF is competitive in various settings.

GA-SmaAt-GNet: Generative adversarial small attention GNet for extreme precipitation nowcasting

In recent years, data-driven modeling approaches have gained significant attention across various meteorological applications, particularly in weather forecasting. However, these methods often face challenges in handling extreme weather conditions. In response, we present the GA-SmaAt-GNet model, a novel generative adversarial framework for extreme precipitation nowcasting. This model features a unique SmaAt-GNet generator, an extension of the successful SmaAt-UNet architecture, capable of integrating precipitation masks (binarized precipitation maps) to enhance predictive accuracy. Additionally, GA-SmaAt-GNet incorporates an attention-augmented discriminator inspired by the Pix2Pix architecture. This innovative framework paves the way for generative precipitation nowcasting using multiple data sources. We evaluate the performance of SmaAt-GNet and GA-SmaAt-GNet using real-life precipitation data from The Netherlands, revealing notable improvements in overall performance and for extreme precipitation events compared to other models. Specifically, our proposed architecture demonstrates its main performance gain in summer and autumn, when precipitation intensity is typically at its peak. Furthermore, we conduct uncertainty analysis on the GA-SmaAt-GNet model and the precipitation dataset, providing insights into its predictive capabilities. Finally, we employ Grad-CAM to offer visual explanations of our model’s predictions, generating activation heatmaps that highlight areas of input activation throughout the network.

Verification of the Deterministic and Probabilistic Radar Nowcasting of Precipitation in Warm and Cold Seasons in the European Part of Russia

Verification of the Deterministic and Probabilistic Radar Nowcasting of Precipitation in Warm and Cold Seasons in the European Part of Russia

LLMDiff: Diffusion Model Using Frozen LLM Transformers for Precipitation Nowcasting.

Precipitation nowcasting, which involves the short-term, high-resolution prediction of rainfall, plays a crucial role in various real-world applications. In recent years, researchers have increasingly utilized deep learning-based methods in precipitation nowcasting. The exponential growth of spatiotemporal observation data has heightened interest in recent advancements such as denoising diffusion models, which offer appealing prospects due to their inherent probabilistic nature that aligns well with the complexities of weather forecasting. Successful application of diffusion models in rainfall prediction tasks requires relevant conditions and effective utilization to direct the forecasting process of the diffusion model. In this paper, we propose a probabilistic spatiotemporal model for precipitation nowcasting, named LLMDiff. The architecture of LLMDiff includes two networks: a conditional encoder-decoder network and a denoising network. The conditional network provides conditional information to guide the denoising network for high-quality predictions related to real-world earth systems. Additionally, we utilize a frozen transformer block from pre-trained large language models (LLMs) in the denoising network as a universal visual encoder layer, which enables the accurate estimation of motion trend by considering long-term temporal context information and capturing temporal dependencies within the frame sequence. Our experimental results demonstrate that LLMDiff outperforms state-of-the-art models on the SEVIR dataset.

Enhancing severe weather predictions with the I-ConvGRU model: An iterative approach for radar echo time series through ConvGRU and RainNet integration

ABSTRACT Precipitation nowcasting plays a crucial role in disaster prevention and mitigation. Existing forecasting models often underutilize output data, leading to suboptimal forecasting performance. To tackle this issue, we introduce the I-ConvGRU model, a novel radar echo timing prediction model that synergizes the temporal dynamics optimization of ConvGRU with the spatial feature enhancement capabilities of RainNet. The model forecasts future scenarios by processing 10 sequential time-series images as input while employing skip connections to boost its spatial feature representation further. Evaluation of the radar echo data set from the Hong Kong Hydrological and Meteorological Bureau spanning from 2009 to 2015 demonstrates the I-ConvGRU model's superiority, with reductions of 17(3.8%) and 49(3.2%) in MSE and MAE metrics, respectively, compared with the TrajGRU model; meanwhile, the I-ConvGRU model had 52(5.8%) and 144(3.8%) lower values on the B-MSE and B-MAE metrics, respectively, than the slightly better performing TrajGRU model. Notably, it significantly improves the prediction of severe precipitation events, with the CSI and HSS metrics increasing by 0.0251(9.6%) and 0.0277(6.8%). These results affirm the model's enhanced effectiveness in radar echo forecasting, particularly in predicting heavy rainfall events.

Precipitation nowcasting method based on Spatial-Temporal Dual Discriminators

Abstract The prediction of radar echo sequence images is a spatiotemporal sequence forecasting problem, which is one of the main challenges in precipitation nowcasting. Addressing issues such as poor extraction of spatiotemporal features by previous models and blurry image sequence predictions, this study proposes a Spatial-Temporal dual Discriminator Precipitation Nowcasting Model (STD-SNGAN) based on spectral normalization generative adversarial networks (SNGAN). The model utilizes multi-scale convolution modules (Inception) to extract spatial features from radar echo images and convolutional gated recurrent units (ConvGRU) to extract temporal features between feature maps. By introducing a hidden feature sampler to enhance the extraction capability of the convolutional gated recurrent units and designing spatial-temporal dual discriminators to constrain the generator’s predicted samples, the spatiotemporal forecasting ability is enhanced. Experimental results demonstrate that the proposed STD-SNGAN model outperforms other algorithms in critical success index (CSI) and probability of detection (POD) for high echo intensity and long-term regions.

Improving Nowcasting of Intense Convective Precipitation by Incorporating Dual-Polarization Radar Variables into Generative Adversarial Networks.

The nowcasting of strong convective precipitation is highly demanded and presents significant challenges, as it offers meteorological services to diverse socio-economic sectors to prevent catastrophic weather events accompanied by strong convective precipitation from causing substantial economic losses and human casualties. With the accumulation of dual-polarization radar data, deep learning models based on data have been widely applied in the nowcasting of precipitation. Deep learning models exhibit certain limitations in the nowcasting approach: The evolutionary method is prone to accumulate errors throughout the iterative process (where multiple autoregressive models generate future motion fields and intensity residuals and then implicitly iterate to yield predictions), and the "regression to average" issue of autoregressive model leads to the "blurring" phenomenon. The evolution method's generator is a two-stage model: In the initial stage, the generator employs the evolution method to generate the provisional forecasted data; in the subsequent stage, the generator reprocesses the provisional forecasted data. Although the evolution method's generator is a generative adversarial network, the adversarial strategy adopted by this model ignores the significance of temporary prediction data. Therefore, this study proposes an Adversarial Autoregressive Network (AANet): Firstly, the forecasted data are generated via the two-stage generators (where FURENet directly produces the provisional forecasted data, and the Semantic Synthesis Model reprocesses the provisional forecasted data); Subsequently, structural similarity loss (SSIM loss) is utilized to mitigate the influence of the "regression to average" issue; Finally, the two-stage adversarial (Tadv) strategy is adopted to assist the two-stage generators to generate more realistic and highly similar generated data. It has been experimentally verified that AANet outperforms NowcastNet in the nowcasting of the next 1 h, with a reduction of 0.0763 in normalized error (NE), 0.377 in root mean square error (RMSE), and 4.2% in false alarm rate (FAR), as well as an enhancement of 1.45 in peak signal-to-noise ratio (PSNR), 0.0208 in SSIM, 5.78% in critical success index (CSI), 6.25% in probability of detection (POD), and 5.7% in F1.

Forecasting West Nile Virus With Graph Neural Networks: Harnessing Spatial Dependence in Irregularly Sampled Geospatial Data.

Machine learning methods have seen increased application to geospatial environmental problems, such as precipitation nowcasting, haze forecasting, and crop yield prediction. However, many of the machine learning methods applied to mosquito population and disease forecasting do not inherently take into account the underlying spatial structure of the given data. In our work, we apply a spatially aware graph neural network model consisting of GraphSAGE layers to forecast the presence of West Nile virus in Illinois, to aid mosquito surveillance and abatement efforts within the state. More generally, we show that graph neural networks applied to irregularly sampled geospatial data can exceed the performance of a range of baseline methods including logistic regression, XGBoost, and fully-connected neural networks.

НАУКАСТИНГ ПАРАМЕТРОВ И ЯВЛЕНИЙ ПОГОДЫ: ИНСТРУМЕНТЫ, ВОЗМОЖНОСТИ И ОГРАНИЧЕНИЯ

Specific features of nowcasting as a forecast of current weather, its needs for observational data, and main approaches to solving a forecast problem are discussed: from simple extrapolation to hydrodynamic models with assimilation of an extended set of high-resolution observations. Prospects of machine learning applications are assessed. Some peculiarities of the nowcasting of precipitation and wind, which are the parameters of particular interest for most of users, are considered. The methodological complexity of the seamless approach to the nowcasting problems, extremely limited deterministic predictability on convective spatiotemporal scales, and importance of using a probabilistic approach are noted. Some illustrations from the national nowcasting developments are presented.

Quality, Predictability, and Utility in Radar Precipitation Nowcasting Applications

Quality, Predictability, and Utility in Radar Precipitation Nowcasting Applications

Global Precipitation Nowcasting of Integrated Multi-satellitE Retrievals for GPM: A U-Net Convolutional LSTM Architecture

Abstract This paper presents a deep supervised learning architecture for 30-min global precipitation nowcasts with a 4-h lead time. The architecture follows a U-Net structure with convolutional long short-term memory (ConvLSTM) cells empowered by ConvLSTM-based skip connections to reduce information loss due to the pooling operation. The training uses data from the Integrated Multi-satellitE Retrievals for GPM (IMERG) and a few key drivers of precipitation from the Global Forecast System (GFS). The impacts of different training loss functions, including the mean-squared error (regression) and the focal loss (classification), on the quality of precipitation nowcasts are studied. The results indicate that the regression network performs well in capturing light precipitation (<1.6 mm h−1), while the classification network can outperform the regression counterpart for nowcasting of high-intensity precipitation (>8 mm h−1), in terms of the critical success index (CSI). It is uncovered that including the forecast variables can improve precipitation nowcasting, especially at longer lead times in both networks. Taking IMERG as a relative reference, a multiscale analysis, in terms of fractions skill score (FSS), shows that the nowcasting machine remains skillful for precipitation rate above 1 mm h−1 at the resolution of 10 km compared to 50 km for GFS. For precipitation rates greater than 4 mm h−1, only the classification network remains FSS skillful on scales greater than 50 km within a 2-h lead time. Significance Statement This study presents a deep neural network architecture for global precipitation nowcasting with a 4-h lead time, using sequences of past satellite precipitation data and simulations from a numerical weather prediction model. The results show that the nowcasting machine can improve short-term predictions of high-intensity global precipitation. The research outcomes will enable us to expand our understanding of how modern artificial intelligence can improve the predictability of extreme weather and benefit flood early warning systems for saving lives and properties.

DEUCE v1.0: a neural network for probabilistic precipitation nowcasting with aleatoric and epistemic uncertainties

Abstract. Precipitation nowcasting (forecasting locally for 0–6 h) serves both public security and industries, facilitating the mitigation of losses incurred due to, e.g., flash floods and is usually done by predicting weather radar echoes, which provide better performance than numerical weather prediction (NWP) at that scale. Probabilistic nowcasts are especially useful as they provide a desirable framework for operational decision-making. Many extrapolation-based statistical nowcasting methods exist, but they all suffer from a limited ability to capture the nonlinear growth and decay of precipitation, leading to a recent paradigm shift towards deep-learning methods which are more capable of representing these patterns. Despite its potential advantages, the application of deep learning in probabilistic nowcasting has only recently started to be explored. Here we develop a novel probabilistic precipitation nowcasting method, based on Bayesian neural networks with variational inference and the U-Net architecture, named DEUCE. The method estimates the total predictive uncertainty in the precipitation by combining estimates of the epistemic (knowledge-related and reducible) and heteroscedastic aleatoric (data-dependent and irreducible) uncertainties, using them to produce an ensemble of development scenarios for the following 60 min. DEUCE is trained and verified using Finnish Meteorological Institute radar composites compared to established classical models. Our model is found to produce both skillful and reliable probabilistic nowcasts based on various evaluation criteria. It improves the receiver operating characteristic (ROC) area under the curve scores 1 %–5 % over STEPS and LINDA-P baselines and comes close to the best-performer STEPS on a continuous ranked probability score (CRPS) metric. The reliability of DEUCE is demonstrated with, e.g., having the lowest expected calibration error at 20 and 25 dBZ reflectivity thresholds and coming second at 35 dBZ. On the other hand, the deterministic performance of ensemble means is found to be worse than that of extrapolation and LINDA-D baselines. Last, the composition of the predictive uncertainty is analyzed and described, with the conclusion that aleatoric uncertainty is more significant and informative than epistemic uncertainty in the DEUCE model.

A Deep Learning Model with Axial Attention for Radar Echo Extrapolation

ABSTRACT Radar echo extrapolation is an important approach in precipitation nowcasting which utilizes historical radar echo images to predict future echo images. In this paper, we introduce the self-attention mechanism into Trajectory Gated Recurrent Unit (TrajGRU) model. Under the sequence-to-sequence framework, we have developed a novel convolutional recurrent neural network called Self-attention Trajectory Gated Recurrent Unit (SA-TrajGRU), which incorporates the self-attention mechanism. The SA-TrajGRU model which combines spatiotemporal variant structure in TrajGRU and self-attention module is simple and effective. We evaluate our approach on the Moving MNIST-2 dataset and the CIKM AnalytiCup 2017 radar echo dataset. The experimental results show that the performance of the proposed SA-TrajGRU model is comparable to other convolutional recurrent neural network models. HSS and CSI scores of the SA-TrajGRU model are higher than scores of other models under the radar echo threshold of 25 dBZ, indicating that the SA-TrajGRU model has the most accurate prediction results under this threshold.

PP-Loss: An imbalanced regression loss based on plotting position for improved precipitation nowcasting

PP-Loss: An imbalanced regression loss based on plotting position for improved precipitation nowcasting

Mini-review of Pysteps: An open-source python library for precipitation nowcasting

Weather forecasting, particularly in short timeframes has been a longstanding challenge in meteorology, addressed in part by nowcasting methods. Leveraging radar data and innovative methodologies, nowcasting tools have evolved significantly, with open-source python platforms like Pysteps making is accessible to researchers to try advanced techniques. This review focus on Pysteps, a modular and user-friendly framework, offering optical flow based deterministic nowcasts and Short-Term Ensemble Prediction System (STEPS) ensemble nowcasts. Recent studies highlights its efficacy, including blending with Numerical Weather Prediction (NWP) models for improved performance beyond the nowcasting timeframe. Pysteps emerges as a versatile solution, facilitating both research innovation and operational forecasting needs, with wide range of input data and modularity.

Self-clustered GAN for precipitation nowcasting

This paper proposes a novel GAN framework with self-clustering approach for precipitation nowcasting (ClusterCast). Previous studies have primarily captured the motion vector using only a single latent space, making the models difficult to adapt to disparate space-time distribution of precipitation. Environmental factors (e.g., regional characteristics and precipitation scale) have an impact on precipitation systems and can cause non-stationary distribution. To tackle this problem, our key idea is to train a generator network to predict future radar frames by learning a sub-network that automatically labels precipitation types from a generative model. The training process consists of (i) clustering the hierarchical features derived from the generator stem using a sub-network and (ii) predicting future radar frames according to the self-supervised labels, enabling heterogeneous latent representation. Additionally, we attempt an ensemble forecast that prescribes random perturbations to improve performance. With the flexibility of representation learning, ClusterCast enables the model to learn precipitation distribution more accurately. Results indicate that our method generates non-blurry future frames by preventing mode collapse, and the proposed method demonstrates robustness across various precipitation scenarios. Extensive experiments demonstrate that our method outperforms four benchmarks on a 2-h prediction basis with a mean squared error (MSE) of 8.9% on unseen datasets.

DSADNet: A Dual-Source Attention Dynamic Neural Network for Precipitation Nowcasting

Accurate precipitation nowcasting is of great significance for flood prevention, agricultural production, and public safety. In recent years, spatiotemporal sequence models based on deep learning have been widely used for precipitation nowcasting and have achieved better prediction results than traditional methods. These models commonly use radar echo extrapolation and utilize the Z-R relationship between radar and rainfall to predict rainfall. However, radar echo data can be affected by various noises, and the Z-R correlation linking radar and rainfall encompasses several variables influenced by factors like terrain, climate, and seasonal variations. To solve this problem, we propose a dual-source attention dynamic neural network (DSADNet) for precipitation nowcasting, which is a network model that utilizes a fusion module to extract valid information from radar maps and rainfall maps, together with dynamic convolution and the attention mechanism, to directly predict future rainfall through encoding and decoding structure. We conducted experiments on a real dataset in Jiangsu, China, and the experimental results show that our model had better performance than the other examined models.

TISE-LSTM: A LSTM model for precipitation nowcasting with temporal interactions and spatial extract blocks

Precipitation nowcasting has a profound impact on humanity and society, especially in areas with heavy rainfall, playing a central role in alerting against rainstorm disasters. At present, numerous deep learning-based methods have been proposed and proven superior to traditional radar echo extrapolation techniques like the Recurrent Neural Networks (RNNs). Our study introduces a novel precipitation forecasting model named TISE-LSTM, which can use the real image of the past half hour to predict the radar echo image of the next hour. By integrating the TIB and SEB into TISL-LSTM, we can alleviate the inherent issue observed in existing models, which is the decrease in forecast accuracy for high radar echo regions with extended extrapolation time. On both the CIKM23017 and AHEM real radar echo datasets, the performance of TISL-LSTM (including POD, FAR, CSI and HSS) is improved compared to the second-ranked model by 7.53%, 2.32%, 8.93%, 2.6%, and 14.84%, 6.18%, 8.92%, 18.12%, respectively, when the precipitation threshold is set to 40. Moreover, our model obtained an optimal MAE, MSE, SSIM score. Predicted images and the graphs of each metric over extrapolation time both demonstrate that our model accurately forecasts regions with high radar echoes even with a one-hour extrapolation.

Reliable precipitation nowcasting using probabilistic diffusion models

Abstract Precipitation nowcasting is a crucial element in current weather service systems. Data-driven methods have proven highly advantageous, due to their flexibility in utilizing detailed initial hydrometeor observations, and their capability to approximate meteorological dynamics effectively given sufficient training data. However, current data-driven methods often encounter severe approximation/optimization errors, rendering their predictions and associated uncertainty estimates unreliable. Here a probabilistic diffusion model-based precipitation nowcasting methodology is introduced, overcoming the notorious blurriness and mode collapse issues in existing practices. Diffusion models learn a sequential of neural networks to reverse a pre-defined diffusion process that generates the probability distribution of future precipitation fields. The precipitation nowcasting based on diffusion model results in a 3.7% improvement in continuous ranked probability score compared to state-of-the-art generative adversarial model-based method. Critically, diffusion model significantly enhance the reliability of forecast uncertainty estimates, evidenced in a 68% gain of spread-skill ratio skill. As a result, diffusion model provides more reliable probabilistic precipitation nowcasting, showing the potential to better support weather-related decision makings.

Flood forecasting based on radar precipitation nowcasting using U-net and its improved models

Accurate and timely short-term precipitation nowcasting is important for achieving reliable flood forecasting. The data-driven approaches have performed well in radar echo extrapolation to nowcast precipitation. In this paper, U-Net and its improved models, including SmaAt-Unet, Nested-Unet, and U-Net 3Plus were applied to perform radar echo extrapolation and precipitation nowcasting for 0.5 h, 1 h, and 2 h lead times of typical rainfall processes. The nowcasted precipitation was used as input to the HEC-HMS hydrological model for flood forecasting to compare the effect of different structural improvements to U-Net on the accuracy of flood forecasting. The results demonstrated that Nested-Unet and U-Net 3Plus aided in enhancing the accuracy of the extrapolation of moderate intensity radar echoes. With fewer discrepancies and better correlation with measured rainfall, the U-Net and U-Net 3Plus precipitation nowcasting results also produced improved flood forecasting outcome. The precipitation nowcasting and flood forecasting for SmaAt-Unet were slightly worse than other models; the relative errors of both flood peak and depth for Nested-Unet at 0.5 h lead time were less than 20 %, showing a good performance. Moreover, in a separate control experiment, the accuracy of the echo extrapolation was significantly decreased when convolutional block attention module (CBAM) was added to each model. However, all models have better extrapolation accuracy than the basic ConvLSTM. In general, Nested-Unet and U-Net 3Plus were helpful to improve the accuracy of precipitation nowcasting and flood forecasting, and the forecasted flood with 0.5 h and 1 h lead times could match the actual flood processes, but the peak discharge from nowcasting with 2 h lead time were severely underestimated, while the peak occurrence time could be forecasted correctly. These conclusions and attempts can provide effective guidelines for regional precipitation nowcasting and flood forecasting.