Based on the insensitivity of the specific differential phase (KDP) to path-integrated attenuation (PIA), partial beam blockage (PBB), and wet radome (WR) issues, the bias of reflectivity (ZH) is corrected for X-band gap-filling radars. TheKDP-ZHrelationship established using the raindrop size distribution (DSD) data collected by local disdrometers is considered in the correction procedure. The practical performance of this correction procedure is extensively evaluated during a severe rainfall event. The results show that (i) this correction methodology can simultaneously address the PIA, PBB, and WR effects onZHmeasurements at X-band frequency, and the correctedZHat X-band agrees very well with DSD-derivedZHand its C-and S-band counterparts; (ii) radar quantitative precipitation estimates (QPE) based onZHare significantly improved after the correction in both the PBB-affeceted and PBB-unaffected areas. The applications of correctedZHand the improved radar QPE can improve the gap-filling capability of X-band radars for surrounding C- and S-band radar observations.

Precipitation products at high spatial and temporal resolutions are critical for applications such as flood monitoring and water resources management. However, the spatial resolution of commonly used satellite precipitation products, such as the NOAA/Climate Prediction Center morphing product (CMORPH version 1; 8 km by 8 km resolution), is often not sufficient for such applications. This article develops a generative adversarial network (GAN) model termed PrecipSRGAN for satellite precipitation product downscaling. PrecipSRGAN uses the National Oceanic and Atmospheric Administration (NOAA) Stage IV quantitative precipitation estimates (QPE) as references and incorporates the digital elevation model (DEM) information to improve satellite-based precipitation feature extraction in order to create a super-resolution version (e.g., 4 km by 4 km) of CMORPH. This deep learning-based downscaling method is evaluated against a traditional statistical interpolation approach in terms of correlation coefficient (CC), normalized mean error (NME), and root mean square error (RMSE), using Stage IV QPE and rain gauge observations as ground references. The results show that PrecipSRGAN outperforms the statistical interpolation method, achieving a composite CC of 0.74, compared to 0.41 of the interpolation approach.

For ground-based weather radar systems, reflectivity is particularly crucial for monitoring severe convective events. However, its limited coverage poses challenges in acquiring reliable radar data, especially for oceanic and mountainous regions. In contrast, geostationary meteorological satellites offer near-global coverage and near-real-time cloud-top observations. This paper introduces a novel deep learning-based radar reflectivity emulation method to reconstruct surface radar observations from cloud-top satellite data, termed Dynamic Residual Convolution-based Network (DRC-Net), aiming to provide more accurate and reliable reflectivity data in regions lacking radar coverage. It uniquely combines dynamic convolution, which focuses attention on convolutional kernels for dynamically adjusting weights based on input, with residual convolution, effectively enhancing network’s ability to capture intricate radar echo details. Experimental results demonstrate that DRC-Net outperforms existing methods in various assessment indices, including Probability Of Detection (POD), False Alarm Ratio (FAR), Critical Success Index (CSI) and Heidke Skill Score (HSS). Generalization tests and case studies further illustrate its effectiveness in reconstructing radar reflectivity across various regions, particularly in mountainous and oceanic areas.

It is well-established that the intensity of tropical cyclones (TCs) is highly associated with their structure. This understanding allows us to utilize geostationary satellite cloud products for the recognition of storm morphology, thereby enabling the accurate estimation of TC intensity. This study adapts the Swin-Unet architecture as a backbone model for an objective deep learning (DL)-based TC intensity estimation method over the Western North Pacific. The model incorporates several key components, including the self-attention mechanism, shift-window mechanism, and Unet structure. Additionally, the model innovatively introduces the rotation index and dispersion index as part of the loss function to characterize storm morphology and perform more comprehensive feature extraction from spatiotemporal data. The model’s input in the study includes five cloud products from the Fengyun series geostationary satellites: sectional image (SEC), cloud top temperature (CTT), temperature of the brightness black-body (TBB), precipitation estimation (PRE), and humidity profile derived from cloud analysis (HPF). Results show that the model obtains an exceptionally low mean absolute error (MAE) of 3.71 m/s and root mean square error (RMSE) of 5.05 m/s. Furthermore, the ablation study (component-impact analysis) quantifies the contribution of the rotation index and dispersion index which enhance the model’s estimation performance to some extent. Finally, through an analysis of feature importance across the five cloud products, HPF, CTT, and TBB received higher importance scores, indicating the model concentrates on the thermodynamic and dynamic features that are strongly associated with TC convective activities. This study is expected to provide technical support for real-time TC intensity estimation in coastal regions and contribute to disaster warning systems.

The microphysical processes' spatial variety and temporal evolution tendency during super typhoon Leima is investigated through measurements from an S-band polarimetric radar and six disdrometers. The results show that collision-coalescence/the collision-breakup occurred near/around the center of the typhoon, deduced from radar-measured and DSD-derived ZDR. In the vertical time-series view of ZDR, the collision-breakup may transition to the collision-breakup near the surface around the altitude of 0.9 km, and the collision-coalescence, collision-breakup, and coalescence-breakup balance may dominate during different periods of typhoon precipitation.

A ground-based weather radar is commonly used for observing severe convective weather. However, the limited coverage of the radar poses difficulties in obtaining reliable radar observations for oceanic and mountainous regions. An effective solution is to derive radar data from meteorological satellite observations using deep-learning methods. This study proposes a novel feature redistribution module-based convolutional neural network (FR-CNN) to retrieve radar composite reflectivity (CREF) data from geostationary satellite observations. Differing from existing skip connection (SC)-based CNNs, FR-CNN adopts a feature redistribution module (FRM) to alleviate the problem of information scarcity during network propagation. In the FRM, a parallel attention block (PAB) is introduced to preserve key feature information and improve the retrieval ability of the FR-CNN. The evaluation results show that the FR-CNN can effectively reconstruct radar reflectivity data and has a better performance than other methods like U-Net in terms of assessment indices including the probability of detection (POD), false alarm ratio (FAR), and critical success index (CSI).

Accurate and real-time estimation of surface precipitation is crucial for decision-making during severe weather events and for water resource management. Polarimetric weather radar serves as the primary operational tool employed for quantitative precipitation estimation (QPE). However, the conventional parametric radar QPE algorithms overlook the dynamic spatiotemporal characteristics of precipitation. In addition, challenges such as radar beam attenuation and imbalanced distribution of precipitation data further compromise the estimation accuracy. This article develops a 3-D star neural network (StarNet) for polarimetric radar QPEs that integrate physical height prior knowledge and employ a reweighted loss function. To better cope with the dynamic characteristics of precipitation patterns, 3-D convolution is introduced within StarNet to effectively capture the spatiotemporal features between successive radar volume scanning data. In particular, multidimensional polarimetric radar observations are utilized as inputs, and surface gauge measurements are employed as training labels. The feasibility and performance of the StarNet model are demonstrated and quantified using U.S. Weather Surveillance Radar-1988 Doppler (WSR-88D) observations collected near Melbourne, Florida. The experimental results show that the StarNet model enhances the prediction accuracy of moderate to heavy precipitation events and improves the estimation performance over long distances, with a mean absolute error (MAE) of 1.55 mm, a root mean square error (RMSE) of 2.63 mm, a normalized standard error (NSE) of 25%, a correlation coefficient (CC) of 0.92, and a BIAS of 0.94 for hourly rainfall estimates. The results suggest that StarNet is able to effectively map the connection between polarimetric radar observations and surface rainfall.

Multilabel remote sensing (RS) image classification aims to predict multiple semantic labels from an RS image. Previous methods [e.g., graph convolution networks (GCNs)] focus on mining the relationships of multiple labels, neglecting that the scene information is closely related to labels. To remedy this deficiency, in this article we propose a novel end-to-end deep neural network for multilabel RS image classification. In the proposed network, we use the GCN as the base model and introduce several new components to improve the classification performance. First, we explore hybrid contrastive learning (CL), including supervised transformation-based CL and unsupervised mix-based CL, to explicitly learn discriminative scene representations. Then, we apply the GCN-based classifier to the learned scene representations to obtain initial label prediction scores. Meanwhile, we pass the scene representations to a softmax layer to predict the probability that each image belongs to each specific scene class and use the scene-to-label information with the law of total probability to calibrate the initial label prediction scores. Finally, we incorporate CL, scene classification, and multilabel classification into a unified learning framework using uncertainty to weigh different losses. Experimental results on two benchmark RS datasets demonstrate the superiority of our proposed network for multilabel image classification.

Hyperspectral image (HSI) and LiDAR data are complementary to each other, which can be combined to improve the classification performance. However, existing deep network models do not sufficiently consider their complementarity to design the network structure and loss functions. Moreover, there lacks a hierarchical mutual-assistance learning mechanism that leverages the modality-shared features to enhance the modality-specific ones and vice versa. In view of these, we propose a novel height information guided hierarchical fusion-and-separation network (HFSNet) for joint classification of HSI and LiDAR data. HFSNet consists of three major components, i.e., dual-structure feature encoders (DSFEs), feature fusion-and-separation blocks (F2SBs), and an edge decoder (ED). Specifically, the transformer and convolutional neural network are introduced in DSFEs to encode the spectral and spatial information of HSI and LiDAR data, respectively. In F2SBs, the deformable convolution-based height information guided fusion module and the modality separation refinement module are proposed to sequentially extract modality-shared and modality-specific features. Additionally, the ED is incorporated into our model to predict the LiDAR edge map from the HSI feature to improve the model’s generalization ability. As such, the learned features from HSI and LiDAR data are deeply fused and mutually enhanced. Experiments on three benchmark datasets show the superiority of HFSNet to the state-of-the-art methods for jointly classifying HSI and LiDAR data with limited training samples.

Due to the limited spatial or temporal resolution, current satellite-based products from VIIRS (Visible Infrared Imaging Radiometer Suite), MODIS (Moderate Resolution Imaging Spectroradiometer), and GOES ABI (Geostationary Operational Environmental Satellite Advanced Baseline Imager) are often inadequate for real-time applications such as wildfire edge detection and wildfire suppression. The overarching goal of this research is to employ deep learning to create high spatial and temporal resolution fire products needed by fire managers for real-time wildfire monitoring and perimeter mapping. To create a super-resolution GOES fire product, this paper will adapt an Enhanced Super-Resolution Generative Adversarial Network (ESRGAN) to train the 2-km GOES ABI data (Channels 2, 7, 14, and 15 in particular). The concurrent overlapping higher spatial resolution VIIRS data (Channels I1-I5, 375 m resolution) from Suomi NPP, NOAA-20, and NOAA-21 (nominal temporal resolution is 12 h or less depending on the latitude) will be used as target labels in training the deep learning super-resolution model for downscaling GOES ABI data. The adapted ESRGAN model can then be applied to the real-time high temporal resolution GOES ABI data to produce high spatial resolution, i.e., 375-m, fire products.

The unique network consisting of three different wavelength (C-, S- and X-band) radar in Eastern China is utilized to demonstrate their coordination capability of better coverage of low atmospheric layers in convective situations. After correcting attenuation effects on horizontal reflectivity (ZH) of all three wavelength radars based on the self-consistency approach, X-band radar exhibits even larger ZHthan S- and C-band, which exactly captures the severe developing convective rainbands on the lower atmospheric layer. In this sense, the X-band radar effectively plays a gap-filling coordination role in this S- and C-band radar network in convective situations.

Real-time and accurate precipitation estimation is critical for environmental protection and water resources management. Compared to traditional methods, i.e., radar reflectivity (Z) and rainfall rate (R) relations, relying on local raindrop size distributions, the deep learning model can fit the functional relationship between radar observations and rainfall rate measurements. However, the black-box nature of deep learning models makes it difficult to explain the physical mechanisms behind their results. To address this problem, this study proposes DQPENet, a deep learning model for polarimetric radar QPE utilizing dense blocks. We employ a permutation test to understand the relative importance of different radar data input variables. Additionally, we propose a regression importance value (RIV) method for the precipitation estimation task to visualize feature importance regions. Our experimental results show that radar reflectivity and specific differential phase at the lowest elevation angle are the two most important observables for the model’s precipitation estimation. Furthermore, we find that radar data closer to the rain gauge are more influential on the model’s results, indicating that the deep learning model is able to capture the underlying physical mechanism of atmospheric data.

Precipitation retrieval biases exist in composite satellite precipitation products (SPPs) due to the limitations of passive microwave (PMW) sensors in resolving shallow and/or heavy rain. The resulting underestimation and overestimation of precipitation intensity and potential precipitation position errors lead to inconsistent and unstable performance of SPPs through different rainfall types at different geophysical locations. This study aims to correct biases and alleviate the effects of uncertainties in precipitation estimates from NOAA Climate Prediction Center (CPC) MORPHing technique (CMORPH) and NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG). Specifically, a deep convolutional neural network (CNN) model is employed to capture and correct the precipitation error patterns in CMORPH and IMERG using the ground-based operational Quantitative Precipitation Estimation and Segregation Using Multiple Sensors (QPESUMS) system product over Taiwan as the training references. The improved performance of satellite precipitation products is quantitatively validated, which demonstrates the capability of the devised deep learning based bias correction approach.

Accurate precipitation retrieval using satellite sensors is still challenging due to the limitations on spatio-temporal sampling of the applied parametric retrieval algorithms. In this research, we propose a deep learning framework for precipitation retrieval using the observations from Advanced Baseline Imager (ABI), and Geostationary Lightning Mapper (GLM) on GOES-R satellite series. In particular, two deep Convolutional Neural Network (CNN) models are designed to detect and estimate the precipitation using the cloud-top brightness temperature from ABI and lightning flash rate from GLM. The precipitation estimates from the ground-based Multi-Radar/Multi-Sensor (MRMS) system are used as the target labels in the training phase. The experimental results show that in the testing phase, the proposed framework offers more accurate precipitation estimates than the current operational Rainfall Rate Quantitative Precipitation Estimate (RRQPE) product from GOES-R.

Research on nowcasting through dual-polarization weather radar data using deep learning approach is rare but worth exploring. This paper lightens a previous work, the MCT (Multivariate Channel Transformer) model, which leads to the design of the MSF (Multivariate Swin Fusion) model. The commonalities between the two are as follows: on one hand, both fuses several dual-polarization observables including reflectivity (Z), specific differential phase (Kdp), and differential reflectivity (Zdr) to more comprehensively consider meteorological particle features; on the other hand, they introduces the attention mechanism to more fully fuse multi-frame, multi-variate, and multi-scale features. In the experimental evaluation, this study first selects observation data from KMLB radar in FL, USA, and uses traditional optical flow method, deep learning TrajGRU method, etc. as controls. The results show that both MCT and MSF perform better than the control, and the 60min forecast scores of both are 8.78/9.31 for RMSE and 0.46/0.18/0.07 for CSI (20/35/45dBZ), and this conclusion is verified by case study. Further, the role of the attention mechanism is verified by ablation experiments.

The task of nowcasting by deep learning using multivariate, rather than just reflectivity, is limited by poor interpretability. The previous experiment designed MCT (Multivariate Channel Transformer), a deep learning model capable of nowcasting with dual-polarization radar data. Four analytical methods are designed to further explore the contribution of polarization parameters: (i) Case studies of different meteorological processes. (ii) A permutation test ranking the significance of each variable. (iii) Visualization of the feature maps obtained by forward propagation of the input data. (iv) Data downscaling of polarimetric radar data. The results show that the polarization parameters serve as a guide to predict the location and shape of strong reflectivity, as well as the energy retention of strong echoes at 40-50 dBZ. The contributions of Zdrand Kdpare more evident in the prediction results after 30 min, and the importance of Kdpexceeds that of Zdrin case of strong convective weather.

The National Weather Service has initiated an effort to upgrade the scan strategy of the operational Weather Surveillance Radar – 1988 Doppler (WSR-88D) to improve its hydrometeorological applications. The lowest scan elevation angle has been changed from 0.5° to 0° (or even lower in future) for some WSR-88D stations. Using the KMUX WSR-88D radar deployed in mountainous terrain in Northern California as an example, this article quantifies the impacts of lowering the minimum scan elevation angle of WSR-88D radar on surface rainfall mapping, with an emphasis on shallow orographic precipitation. In order to estimate surface rainfall using radar observations, polarimetric radar rainfall relations are established using local disdrometer data, which are then implemented with the KMUX radar observations at 0° and 0.5° scan elevation angles to derive rainfall estimates. Comparative evaluation of the radar-based rainfall estimates using rainfall measurements from surface rain gauges has demonstrated the superior performance of the lower scan elevation angle. When the distance from the radar is long (i.e., when the radar beam is likely within or above the melting layer), the improvement gained by the 0° scan relative to the 0.5° scan is 16.1% and 19.5% in terms of the normalized standard error (NSE) and the Pearson correlation coefficient (CORR), respectively.

Precipitation is the primary driver of the regional/global hydrological cycle. Accurate precipitation estimation is essential for disaster management and climate change studies. Geostationary meteorological satellites provide a wealth of temporal and spatial information for quantitative precipitation estimation (QPE), but how to extract useful precipitation-related information from raw satellite data has always been a challenge. As a result, there is still a gap between the current satellite precipitation products and ground-based precipitation observations. Recent advancements in deep learning (DL) techniques now allow the development of more accurate QPE algorithms. This letter proposes a PREcipitation estimation algorithm based on Multi-task learning (PREM) using satellite observations extracted from the Geostationary Operational Environmental Satellite-R (GOES-R) series. The key design points of PREM are as follows: 1) the detection (rain/no-rain separation) and rainfall rate estimation are integrated in PREM; 2) infrared (IR) radiance data are the main input to PREM, and ground-based radar precipitation products are used as the target in the training phase; and 3) in addition, PREM integrates lightning data (called PREM-L) to investigate its impact on heavy precipitation estimation. Independent verification shows that PREM performs better than the operational GOES-R products, and PREM-L outperforms PREM when precipitation is heavy.

Accurate estimation of surface precipitation with high spatial and temporal resolution is critical for decision making regarding severe weather and water resources management. Polarimetric weather radar is the main operational instrument used for quantitative precipitation estimation (QPE). However, conventional parametric radar QPE algorithms such as the radar reflectivity (Z) and rain rate (R) relations can not fully represent clouds and precipitation dynamics due to their dependency on local raindrop size distributions and the inherent parameterization errors. This article develops four deep learning (DL) models for polarimetric radar QPE (i.e., RQPENetD1, RQPENetD2, RQPENetV, RQPENetR) using different core building blocks. In particular, multi-dimensional polarimetric radar observations are utilized as input and surface gauge measurements are used as training labels. The feasibility and performance of these DL models are demonstrated and quantified using U.S. Weather Surveillance Radar - 1988 Doppler (WSR-88D) observations near Melbourne, Florida. The experimental results show that the dense blocks-based models (i.e., RQPENetD1and RQPENetD2) have better performance than residual blocks, RepVGG blocks-based models (i.e., RQPENetRand RQPENetV) and five traditionalZ-Rrelations. RQPENetD1has the best quantitative performance scores, with a mean absolute error (MAE) of 1.58 mm, root mean squared error (RMSE) of 2.68 mm, normalized standard error (NSE) of 26%, and correlation of 0.92 for hourly rainfall estimates using independent rain gauge data as references. These results suggest that deep learning performs well in mapping the connection between polarimetric radar observations aloft and surface rainfall.

Satellite precipitation products such as NASA’s Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (GPM) (IMERG) have served as valuable sources for oceanic precipitation information. However, quantifying their uncertainty over oceans remains challenging due to limited reference observations. This study uses a variety of advanced in situ rainfall measurements collected during NASA’s Salinity Processes in the Upper Ocean Regional Study-2 (SPURS-2; August 2016–November 2017) to characterize the uncertainty of IMERG over the western edge of the tropical Eastern Pacific Fresh Pool. A censored, shifted gamma distribution (CSGD) model is implemented for uncertainty quantification of IMERG at its gridded resolution (i.e., 30 min, 0.1°), and rainfall observations from multiple passive aquatic listeners (PALs), mooring gauges, and gauges on the research vessel (R/V) are used to train and validate the model’s performance. The results indicate that the drifting PAL-trained CSGD models perform similarly as the mooring gauge-trained models, showing good fits to the data and successful representations of bias and random error. The CSGD-estimated quantiles align with the empirical quantiles derived from references, demonstrating the model’s ability to reproduce the conditional density distributions of possible “true” precipitation given the IMERG estimates. Verification with independent data from the R/V and GPM Level-2A Dual-frequency Precipitation Radar precipitation product (2ADPR) confirms the model’s performance in generating reliable point-scale rainfall ensembles and in improving heavy rainfall estimation by accounting for IMERG uncertainty. With the newly available oceanic rainfall dataset from 58 PALs globally, this error modeling framework can be readily applied to estimate the uncertainty of IMERG for applications in other ocean regions.

Missing or low-quality data regions usually happen to weather radars. One of the most common situations is beam blockage or partial beam blockage. Therefore, correction of weather radar observations that are partially or fully blocked is an indispensable step in radar data quality control and subsequent quantitative applications, especially in complex terrain environments such as the western United States. In this article, we propose a deep learning framework based on generative adversarial networks (GANs) for restoring partial beam blockage regions in polarimetric radar observations using local and global contextual information. Due to the diverse precipitation types, blockage conditions, and ground information in different areas, two radars deployed in two different regions characterized by different precipitation types are used to demonstrate the proposed methodology. Both are S-band operational Weather Surveillance Radar – 1988 Doppler (WSR-88D): KFWS located in Fort Worth, northern Texas, and KDAX located in Davis, northern California. For training the GAN model, this article simulates the partial beam blockage situations by manually cropping observation sectors of both KDAX and KFWS radar data. The trained models were tested using independent precipitation events in Texas and California to demonstrate the model effectiveness in inpainting “missing" data. In addition, this paper cross-tested the data with different precipitation features to examine the generalization capacity of the beam blockage correction models. The beam blockage correction performance is also compared with a traditional linear interpolation approach. The results show that for both domains the continuity of precipitation observations is greatly improved after applying the deep learning-based inpainting approach. For the KFWS test data, some visible discrepancies exist between the results from models trained based on convective and stratiform precipitation events in Texas and California, respectively, yet both models outperform the traditional interpolation method. For the KDAX test data, both the model trained using the KFWS data from convective precipitation events in Texas and the model trained using KDAX data from stratiform precipitation events in California render a similar performance. Although ground truth is not available for the real blocked radar data, the repaired observations demonstrated a great potential for improved quantitative applications.

Satellite sensors have been widely used for precipitation retrieval, and a number of precipitation retrieval algorithms have been developed using observations from various satellite sensors. The current operational rainfall rate quantitative precipitation estimate (RRQPE) product from the geostationary operational environmental satellite (GOES) offers full disk rainfall rate estimates based on the observations from the advanced baseline imager (ABI) aboard the GOES-R series. However, accurate precipitation retrieval using satellite sensors is still challenging due to the limitations on spatio-temporal sampling of the satellite sensors and/or the uncertainty associated with the applied parametric retrieval algorithms. In this article, we propose a deep learning framework for precipitation retrieval using the combined observations from the ABI and geostationary lightning mapper (GLM) on the GOES-R series to improve the current operational RRQPE product. Particularly, the proposed deep learning framework is composed of two deep convolutional neural networks (CNNs) that are designed for precipitation detection and quantification. The cloud-top brightness temperature from multiple ABI channels and the lightning flash rate from the GLM measurement are used as inputs to the deep learning framework. To train the designed CNNs, the precipitation product multi-radar multi-sensor (MRMS) system from the National Oceanic and Atmospheric Administration (NOAA) is used as target labels to optimize the network parameters. The experimental results show that the precipitation retrieval performance of the proposed framework is superior to the currently operational GOES RRQPE product in the selected study domain, and the performance is dramatically enhanced after incorporating the lightning data into the deep learning model. Using the independent MRMS product as a reference, the deep learning model can reduce the retrieval uncertainty in the operational RRQPE product by at least 31% in terms of the mean squared error and normalized mean absolute error, and the improvement is more significant in moderate to heavy rain regions. Therefore, the proposed deep learning framework can potentially serve as an alternative approach for GOES precipitation retrievals.

The short-term convective storm forecasting (i.e., nowcasting) mainly relies on weather radar, which can resolve the 3-D structure of convective storms. With the rapid development of numerical models, modern models can produce 3-D reanalysis data, which gives atmospheric background information of convective storms. Current deep learning nowcasting models use only 2-D radar images for nowcasting and often require massive historical data for training. But, it may not be operationally feasible to collect long-term radar data to train a new model. Hence, how to establish a nowcasting model using only a small dataset has become an important issue. In addition, the existing models do not effectively use the state-of-the-art model reanalysis data, which is a shortcoming of these models. To tackle these problems, this article develops a pixelwise convolutional-recurrent neural network (Pixel-CRN) for precipitation nowcasting. It has three key designs: 1) through a concise pixelwise sampling and oversampling technique, Pixel-CRN can be trained using only a small dataset; 2) in spatial learning, Pixel-CRN embeds a spatial convolution subnet into the recurrent unit, which can input raw 3-D radar and model reanalysis data; thus, valuable atmospheric background information can be learned to assist nowcasting; and 3) in spatiotemporal learning, according to the information bottleneck principle, Pixel-CRN builds a heterogeneous encoder–decoder structure to squeeze multichannel 3-D input data into latent space and recurrently generates 30- and 60-min nowcasts. Compared with the existing deep learning nowcasting methods, the experimental results show that the Pixel-CRN can provide skillful results with a rather small training dataset.

Although polarimetric radar measurements are rich in information, traditional approaches of using them can only extract part of the information due to the limitations of the analytical tools. The performance of conventional radar rainfall estimation algorithms is highly dependent on the raindrop size distributions, which vary in different precipitation regions and/or different rainfall systems. It is challenging to remove the inherent parameterization error in “fixed” parametric radar rainfall relations. Recent studies have shown that deep learning techniques are effective in reducing such parameterization error and enhancing radar-based precipitation estimation. However, it is challenging to train a model that is applicable to a broad domain. Often, local rain gauge data would be required to retrain the model obtained for another domain. This study takes advantages of crowdsourced data from the NOAA Meteorological Assimilation Data Ingest System (MADIS), which is the national clearinghouse for weather and hydrologic observations that feed the NWS operational numerical weather and hydrologic prediction models. A convolutional neural network is utilized as benchmark, which incorporates a residual block to address the model degradation caused by the increased model depth. Through manipulation of the training process, the knowledge learned at one location is transferred to other domains characterized by different precipitation properties. The experimental results show that the proposed technique can improve precipitation estimation compared to conventional fixed-parameter rainfall algorithm.

Partial beam blockage (PBB) correction is necessary in complex terrain for radar quantitative applications. Regional KDP-ZHrelationship is established from in situ raindrop size distribution (DSD) measurements and is then utilized to mitigate the PBB effects through combining an attenuation correction approach. The practical performance of this PBB correction technique is evaluated through evaluating the spatial continuity of reflectivity ($Z_{\mathrm{H}}$}) measurements and rainfall estimates based on$R(Z_{\mathrm{H}})$and$R(K_{\text{DP}})$. The results show that with the developed attenuation and PBB correction schemes (i) the spatial continuity of$Z_{\mathrm{H}}$measurements can effectively be enhanced; (ii) rainfall estimates based on$R(Z_{\mathrm{H}})$in the PBB-affected area are incrementally improved with better spatial continuity and its performance tends to be more comparable with$R(K_{\text{DP}})$.

Convective storm is one of the major catastrophic weather events. In recent years, deep learning algorithms fused with various meteorological data are emerging as mainstream techniques in related research areas. However, the black-box nature of deep learning often makes such algorithms hard to interpret from physical point of view. In this study, three interpretation methods are developed to interpret a deep learning nowcasting model, i.e. permutation test, saliency maps and Grad-CAM (Gradient-weighted Class Activation Mapping). After ranking the importance of the eight variables selected in this study by permutation test, it is concluded that the radar reflectivity is the most important factor for the occurrence of convective storm. Based on the results of saliency maps and Grad-CAM, the areas with high perturbation temperature, strong wind and low humidity are also important for short term prediction of convective storms.

Detection of convective initiation (CI) is an important step of early warning of strong convective weather. This study proposes a machine learning approach for CI detection using 18 interesting fields extracted from weather radar, Himawari-8 Advanced Himawari Imager (AHI), and the variational Doppler radar analysis system (VDRAS). The collected features are used to train the machine learning model for CI detection. Therein, the support vector machine (SVM) is used to identify the CI and non-CI. It is concluded that a better result can be achieved by using multiple-source data compared to using satellite data only. The atmospheric boundary layer thermal dynamic information retrieved by VDRAS is proved to be useful for CI detection.

This paper develops a novel deep learning-based model for quantitative precipitation estimation (QPE). In this model, the multi-channel microwave Brightness Temperature (TBVand TBH) and Polarization Difference (PD = TBV - TBH) from the Microwave Radiation Imager (MWRI) aboard the FY-3D satellite are used as predictor variables for QPE. An observational database built from the Integrated Multi-satellite Retrievals for GPM (IMERG) is employed as reference to train the deep learning model. In order to assess the model performance, an independent test dataset is used, and the results show that the model is computationally efficient and the derived QPE is highly consistent with IMERG product. The encouraging results also indicate that PD can enhance the precipitation estimation performance compared to using the TB only.

Passive satellite precipitation sensors suffer from resolving heavy rain and/or shallow orographic precipitation systems, thereby restricting the performance of various composite satellite precipitation products. It is crucial to correct the error patterns of satellite precipitation retrievals for improving hydrometeorology applications at different space-time scales. To achieve this goal, this paper proposes a deep learning system with ensemble optimization strategies to quantify the uncertainties of satellite products with an emphasis on orographic precipitation. In particular, a deep convolutional neural network (CNN) is designed, which utilizes the ground-based Stage IV precipitation estimates as target labels to reduce biases involved in the precipitation product derived from the NOAA/Climate Prediction Center morphing technique (CMORPH). In order to boost the performance of the single model, an ensemble strategy is devised along with the deep learning model. The results show that the accuracy of CMORPH has been significantly improved via the proposed methodology, indicating the great potential of machine learning in improving future satellite precipitation retrievals.

Deep learning methods have been applied to weather forecasting and achieved superior performance than traditional methods. Can we use postprocessing or bias correction methods to further improve the forecasting results? In this study, a frequency-matching correction method is proposed to correct the nowcasting results provided by a deep learning nowcasting method, i.e., TrajGRU. The frequency distribution curve is calculated first using the forecast and radar observation. Then, the correction matrix can be obtained based on this distribution curve. Finally, the forecast is corrected using the correction matrix. The experimental results show that the deep learning model also has bias which needs to be corrected. The proposed method achieves better forecast performance after correction. When using 50 dBZ as the threshold, more obvious forecast improvement can be observed.

In this study, we propose a deep learning nowcasting method called multivariate channel transformer U-net (MCT If-net). It incorporates the channel transformer into the basic U-net structure. Three dual-polarization radar variables, i.e., reflectivity ($Z$), specific differential phase ($K_{dp}$), and differential reflectivity ($Z_{dr}$), are used as inputs to the proposed model. Compared to traditional U-net, MCT U-net has two advantages: 1) We use channel transformer to replace the skip-connection operation to better capture global information. Due to the inability to fuse the semantic features at different scale, a simple skip connection operation cannot capture the global multi-scale context, and this study uses the self-attention mechanism in Transformer to tackle this problem. 2) MCT U-net extracts multi-scale features through down-sampling each variable and then concatenates and sends them to a channel transformer. The combination of multiple variables in higher semantic level helps the model learn each individual modality better. The experimental results show that MCT U-net improves nowcasting performance compared with traditional methods.

Precipitation nowcasting is an important operational service for protecting public property losses and people’s safety. Short-Term Ensemble Prediction System (STEPS) is a probabilistic nowcasting system which has been widely used in the research community (commonly referred as PySTEPS). This study investigates the predictability of PySTEPS during different precipitation systems, i.e., convective and stratiform events. In particular, two study domains, namely, Dallas-Fort Worth (DFW) area in northern Texas and San Francisco Bay Area in northern California, are selected to represent these two typical precipitation patterns, respectively. The experimental nowcasting results show that PySTEPS works well in both areas, especially during stratiform rainfall events in the Bay Area. In addition, PySTEPS exhibits different performance for different precipitation patterns. For convective cases in the DFW area, PySTEPS tends to underestimate rain rate for high-intensity precipitation regions. For stratiform cases in the Bay Area, PySTEPS can predict the precipitation intensity more accurately. With the increase of nowcasting lead time, the qualitative evaluation scores (POD - Probability of Detection, and CSI - Critical Success Index) of PySTEPS decrease slowly during stratiform events compared with convective events, which is also in line with the quantitative evaluation results.

The performance of various composite satellite precipitation products is severely limited by their individual passive microwave (PMW)-based retrieval uncertainties because the PMW sensors have difficulties in resolving heavy rain and/or shallow orographic precipitation systems. Characterizing the error structure of PMW retrievals is crucial to improving precipitation mapping at different space–time scales. To this end, this article introduces a machine learning framework to quantify the uncertainties associated with satellite precipitation products with an emphasis on orographic precipitation. A deep convolutional neural network (CNN) is designed, which utilizes the ground-based Stage IV precipitation estimates as target labels in the training phase, to reduce biases involved in the precipitation product derived using the NOAA/Climate Prediction Center morphing technique (CMORPH). The products before and after bias correction are evaluated using four independent precipitation events over the coastal mountain region in the western United States, and the impact of topography on satellite-based precipitation retrievals is quantified. Experimental results show that the orographic gradients have a strong impact on precipitation retrievals in complex terrain regions. The accuracy of CMORPH is dramatically enhanced after applying the proposed machine learning-based bias correction technique. Using Stage IV data as references, the overall correlation (CC), normalized mean error (NME), and normalized mean absolute error (NMAE) of CMORPH are improved from 0.55, 32%, 63%, to 0.88, −2%, 39%, respectively, after bias correction for the independent case studies presented in this article. Such a machine learning scheme also has great potential for improved fusion of other or future satellite precipitation retrievals.

Convective precipitation nowcasting remains challenging due to the fast change in convective weather. Radar images are the most important data source in nowcasting research area. This study proposes a radar data-based U-Net model for precipitation nowcasting. The nowcasting problem is first transformed into an image-to-image translation problem in deep learning under the U-Net architecture, which is based on convolutional neural networks (CNNs). The input of the model is five consecutive radar images; the output is the predicted radar reflectivity image. The model consists of three operations: upsampling, downsampling, and skip connection. Three methods, U-Net, TREC, and TrajGRU, are used for comparison in the experiments. The experimental results show that both deep learning methods outperform the TREC method, and the CNN-based U-Net can achieve almost the same performance as TrajGRU which is a recurrent neural network (RNN)-based model. With the advantages that U-Net is simple, efficient, easy to understand, and customize, this result shows the great potential of CNN-based models in addressing time-series applications.

Deep learning is emerging as a powerful tool in scientific applications, such as radar-based convective storm nowcasting. However, it is still a challenge to extend the application of a well-trained deep learning nowcasting model, which demands to incorporate the learned knowledge at a certain location to other locations characterized by different precipitation features. This article designs a transfer learning framework to tackle this problem. A convolutional neural network (CNN)-based nowcasting method is utilized as the benchmark, based on which two transfer learning models are constructed through fine-tune and maximum mean discrepancy (MMD) minimization. The base CNN model is trained using radar data in the source study domain near Beijing, China, whereas the transferred models are applied to the target domain near Guangzhou, China, with only a small amount of data in the target area. The influence of a varying number of target data samples on the nowcasting performance is quantified. The experimental results demonstrate that the deep transfer learning models can improve the nowcasting skills.

The dual-frequency precipitation radar (DPR) onboard the global precipitation measurement (GPM) satellite provides valuable measurements of precipitation. In this study, the GPM DPR products (version 6) are validated against a ground-based S-band polarimetric radar in South China based on a volume-matching method. Good consistency is found for the reflectivity factor ($Z$) calibration of the two instruments. From the perspective of microphysics, the mass-weighted mean diameter ($D_{m}$) estimates correspond well with those of the ground-based radar in the inner swath of the normal scan (NS); however, underestimation is found for the raindrop number concentration, indicated by the generalized intercept parameter ($N_{w}$), especially for the intense echoes. Thus, the GPM DPR product may fail to depict the microphysical characteristics of small-to-medium raindrops in high concentration for heavy rainfall in South China. This is attributed to the negative$Z$bias of the DPR caused probably by insufficient correction of attenuation, which also leads to clear underestimation in the liquid water content ($W$) and the rainfall rate ($R$) products for intense echoes. In the outer swath where only single-frequency retrieval is available, overestimation in$D_{m}$exists regardless of echo intensity level, and more underestimation can be found in$N_{w}$,$W$, and$R$especially for intense echoes. In the selected typhoon and squall line cases, better capability in revealing microphysical properties is also found for the inner swath of the NS. After adjusting the scan mode, the performance of the precipitation products in the outer swath can be improved by dual-frequency retrievals in the future.

Recurrent neural networks (RNNs) have been widely used for hyperspectral image (HSI) classification via sequence modeling. However, most of the RNN methods focus on modeling long-range dependencies along the spectral direction, without fully exploring multi-directional dependencies in the joint spectral-spatial domain. To tackle this issue, we propose MSLAN, a two-branch multi-directional spectral-spatial long short-term memory (LSTM) attention network, for HSI classification. In particular, we employ LSTMs to extract six-directional spatial-spectral features which simultaneously capture the spectral-spatial dependencies along different directions. We then design an attention-based feature fuse module to integrate these directional features, followed by a fully connected layer with cross-entropy loss for classification. Additionally, we incorporate an auxiliary branch into our model to enhance the generalization capability. In this branch, random spatial shuffle and a cosine loss are explored for feature consistency learning by taking into account the varying spatial distributions. The resulting two branch networks, sharing the same network structure and weights, are incorporated into a unified deep learning architecture for training. Experiments show the superiority of MSLAN to the state-of-the-art methods for HSI classification with limited training samples.

High spatial resolution precipitation data are important for hydrological modeling and meteorological applications, especially at regional scales. Statistical downscaling methods for satellite precipitation products using the normalized difference vegetation index (NDVI) have been carried out in many regions to provide high spatial resolution precipitation. These methods generally use NDVI and precipitation at the same time, assuming that there is a real-time response of vegetation to precipitation. However, this assumption does not hold in many scenarios. It is known that different vegetation types exhibit different response times to precipitation, i.e., there is a possible lag in the response of vegetation to precipitation depending on the vegetation/landcover type. Therefore, it is not appropriate to estimate precipitation using NDVI collected at the same time. To better represent the relationship between precipitation and vegetation, this article develops a new vegetation index based on adaptive lag phase (VIAL) estimated from a new growth rate that is adaptive to landcover type. Based on VIAL, a new local precipitation downscaling method called LPVIAL is proposed, which essentially considers the nonstationary relationship between precipitation and VIAL. The performance of LPVIAL is assessed by downscaling Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG) from 0.1° to 1-km spatial resolution over the Pearl River Basin in Southern China from 2010 to 2017 at 16-day temporal resolution, and the downscaled products are validated against ground observations. Results indicate that the high-resolution precipitation data obtained from the new downscaling approach perform well, and the accuracy is higher than traditional approaches. With the enhancement of spatial resolution, LPVIAL downscaled products show more detailed spatial information of precipitation with smooth distribution, and the downscaled products have slightly higher accuracy compared with IMERG. It is, therefore, suggested that the adaptive lag phase should be considered in the satellite precipitation product downscaling process.

Traditionally, polarimetric radar-based rainfall estimates are derived through empirical parametric relations obtained from nonlinear regression between rain rates and simulated radar data. The performance of such empirical relations is highly dependent on the variations of raindrop size distribution. In real applications, such empirical algorithms often need to be adjusted for different climate regimes and/or different rainfall types, and we do not have a simple parametric expression linking radar observables to rainfall intensity. Even if one can eliminate all the random errors associated with radar measurements, the parameterization uncertainty inherent in the empirical relations is hard to reduce. In addition, it is difficult to estimate surface rain rates with the radar measurements aloft, especially during the precipitation events characterized by dramatically changing vertical structures (i.e., precipitation particle sizes and distributions are changing during the falling process). In this paper, a convolutional neural network-based machine learning approach is developed to estimate surface rainfall rate from multidimensional polarimetric measurements. This deep learning algorithm can extract the complex relation from high dimensional input space (i.e., radar data) to the target space (i.e., surface rain rate). Polarimetric radar observations and rain gauge data collected near Melbourne, Florida are utilized for demonstration. The trained model is also extended to the whole radar coverage domain to provide complete rainfall mapping. Independent verification results show that this deep learning model has superior performance to conventional fixed-parameter rainfall relations based on either single- or dual-polarization radar measurements.

A water-logging disaster occurred in Eastern China on 1 June 2016 was utilized to investigate the practical performance of six radar rainfall estimators based on horizontal reflectivity ($Z_{\mathrm{H}}$), specific attenuation ($A_{\mathrm{H}}$), specific differential phase ($K_{\text{DP}}$), and multi-variable estimators that integrate differential reflectivity ($Z_{\text{DR}}$), namely$R(Z_{\mathrm{H}},\, Z_{\text{DR}}), R(K_{\text{DP}}, Z_{\text{DR}})$and$R(A_{\mathrm{H}}, Z_{\text{DR}})$. Detailed evaluation with a local rain gauge network and drop size distribution (DSD) measurements shows that (i)$R(A_{\mathrm{H}},\, Z_{\text{DR}})$underestimates the light rain pattern, but it achieves the best scores and performs the best among the various estimators; (ii) each radar rainfall estimator can outperform other estimators at a certain period of time determined by the rainfall patterns; (iii) both the optimal radar rainfall relationship and high calibration accuracy are required to obtain accurate rainfall estimates; (iv) the contamination of melting solid hydrometeors on$A_{\mathrm{H}} \text{and}/\text{or}K_{\text{DP}}$may make them less effective than$Z_{\mathrm{H}}$. In addition, with appropriate adjustment of$\alpha$coefficient,$R(A_{\mathrm{H}})$can perform better than$R(K_{\text{DP}})$.

The axis ratio of fitted ellipse plays a key role for classifying the linear convective precipitation features (convPFs) from others. One of the limitations in identifying convPFs is the discreteness of convective pixels, which could divide a linear convPFs into multiple segments. It would underestimate the occurrence of severe weather system to some extent. A revised unsupervised clustering method based on the density-based algorithm (DBSCAN) is proposed to improve the identification of linear convPFs. Preliminary study shows, the new method is capable of recognizing the linear convPFs with discrete convective pixels correctly.

Convective precipitation nowcasting remains challenging due to the fast change of convective weather. Radar images are the most important data source in nowcasting research area. This study proposes a radar data-based U-Net model for precipitation nowcasting. The input of the model is five consecutive radar images; the output is 30-min prediction of radar reflectivity image. The model consists of three parts: upsampling, downsampling and skip-connection. Two models, U-Net and TrajGRU, are used for comparison in the experiments. Different from U-Net, which is a CNN-based (Convolution Neural Network) model, TrajGRU is an RNN-based (Recurrent Neural Network) model, which is good at time-series processing and has been widely used in precipitation research community. The experimental results show that the CNN-based U-Net can achieve almost the same performance as TrajGRU. This result shows the great potential of CNN-based models in handling time-series applications.

Compared to the legacy GOES Imager, the GOES-16 Advanced Baseline Imager (ABI) provides measurements at much higher resolution in both spatial and temporal dimensions, which benefits many applications such as rainfall rate estimation. However, it is challenging to quantify the water droplets in the cloud and precipitation intensity with only the Infrared (IR) radiation and converted brightness temperature information. Comparison between ground radar rainfall estimates and the operational GOES-16 rain rate products indicates that uncertainty associated with GOES-16 rain rate estimates is significant. This paper proposes an innovative deep learning system for precipitation estimation using measurements from GOES-16/ ABI. Cloud-top brightness temperature information observed by channels 8/10/11/14/15 of the ABI are used as input to this deep learning framework. The rainfall estimates from a ground radar network are used as target labels in the training phase. Independent verification shows that the deep learning-based rainfall system can estimate precipitation time, intensity, and amount very well, especially in heavy rain regions.

Satellite remote sensing observations at high spatial and temporal resolutions have advantages for precipitation retrievals. However, it is still challenging to produce reliable precipitation estimates over complex terrain regions. This paper proposes a deep learning framework to correct the bias associated with satellite revivals of orographic precipitation, focusing on the influence of terrain on precipitation estimation. Using precipitation product derived by the NOAA Climate Prediction Center morphing technique (i.e., CMORPH-CRT) in the western United States as an example, the deep learning model is trained and tested, and the ground-based Stage IV rainfall estimates are used as target labels in the training phase. Independent validation results show that the deep learning-based bias correction model can significantly enhance the performance of satellite precipitation products over complex terrain.

Nowcasting of precipitation associated with landfalling hurricanes refers to the hurricane intensity analysis and forecast for the next few hours. It plays an important role in early weather warning system and real-time decision making for disaster management and risk reduction. However, the skill of traditional nowcasting techniques decreases rapidly in the first hour. In addition, numerical weather prediction (NWP) models have deficiency to provide accurate warning of heavy precipitation on the scale needed for hurricane nowcasting. Therefore, this paper proposes a deep learning-based system for precipitation nowcasting for landfalling hurricanes using multi-radar observations near the coast. The reflectivity imageries from previous 1-hr observations are used as input to the deep learning-based nowcasting model and the actual observations from “future” sequences are used as targets in the training stage. The performance and evaluation results of the nowcasting products with lead time up to 3 hours show that this deep learning system performs well in hurricane nowcasting.

Convective storm nowcasting has long been an important issue and has attracted substantial interest. 3D radar images and 3D re-analysis data contain spatiotemporal information of the convective processes. This paper proposes a multi-channel 3D convolutional recurrent neural network (3D-CRN) for convective storm nowcasting, which aims to learn spatiotemporal information directly from these 3D radar and reanalysis data. 3D-CRN is composed of two sub-networks: three multi-channel 3D convolutional networks are used as the front-end spatial sub-networks, and the convLSTM encoder-decoder is constructed as a back-end temporal sub-network. By combing two subnets into one unified network, 3D-CRN can be jointly trained effectively. In order to give forecasts at different lead time simultaneously, we construct a many-to-many encoder-decoder structure to avoid tedious need to train several models respectively. Experimental results show the effectiveness of the proposed method.

This study presents a variational approach for optimized estimation of specific differential phase ( KDP) for polarimetric radars using a linear forward operator. A cubic B-spline interpolating filter is included to mitigate the impact of measurement error in the total differential phase and ensure the spatial continuity of KDP. For rain, non-negative constraints are introduced to ensure that the KDPestimates are within the physical bounds. The variational approach is flexible to incorporate the background information constructed from the measurements of horizontal reflectivity factor ( ZH) and differential reflectivity ( ZDR) based on the self-consistent relationship of polarimetric variables. The variational approach is evaluated using simulated experiments, as well as real observations from an S-band operational weather radar. Without including background information, the variational approach has slightly better performance compared to the approach based on linear programming (LP), and the background information helps to further improve the performance. In addition, the linear forward operator makes this variational approach computationally efficient. It needs less than 3% computational power required by the approach based on LP, making it more suitable for real-time operational applications.

In this study, as part of an effort to study snowfall characteristics and quantify winter precipitation in East China, we investigated the microphysical properties of snowfall, including size, shape, density, and terminal velocity using a 2-D video disdrometer (2-DVD) and a weighing precipitation gauge in Nanjing (NJ), East China during the winters of 2015-2019. We obtained larger snow density and terminal velocity values than those reported in the literature for this region. Higher snow density could account for higher snowflake terminal velocity, after removing the effects of observation altitude and surface temperature. We then fit the snow particle size distributions (PSDs) to the gamma model and explored the interrelationships among the model parameters and snowfall rate (SR). The relationship between radar reflectivity factor (Ze) and SR was derived based on snow PSD measurements and the snow density relation. Using this Ze-SR relationship, the estimated liquid-equivalent SRs are obtained from S-band NJ radar data collected during several snowfall events. Radar-inferred SRs showed reasonable agreement with those measured on the ground, with a mean absolute error of 16% for the collected snowfall events in NJ.

Remote sensing of precipitation is critical for regional, continental, and global weather, water, and climate research. This study develops a machine learning mechanism to link between point-wise rain gauge measurements, ground-based and spaceborne radar reflectivity observations. Two neural network models are designed to construct a hybrid rainfall system, where the ground radar is used to bridge the scale gaps between rain gauge and satellite. The first model is trained for ground radar using rain gauge data as target labels, whereas the second one is for spaceborne radar using ground radar estimates as training labels. Data from TRMM Precipitation Radar (PR) and GPM Dual-frequency Precipitation Radar (DPR) are utilized to illustrate the application of this hybrid rainfall system. Validation using independent ground-based observations as well as the standard PR and DPR products demonstrates the promising performance and generality of this innovative machine learning algorithm.

Radar quantitative precipitation estimation (QPE) and forecast (QPF) over the coastal mountain area in Northern California can be very challenging due to the complex precipitation microphysics induced by land-ocean interaction in the coastal regions and orographic enhancement in the mountainous areas. A number of previous studies have documented the precipitation characteristics in this complex domain using data collected from remote sensors such as S-band profiling radar (S-PROF) and/or in situ measurements of raindrop size distributions (DSD) at the surface. This study extends the earlier work by including new DSD measurements both from the valley (Santa Rosa) and the mountain (Middletown), to characterize the orographic enhancement of precipitation and diagnose the bulk microphysical properties of rainfall. Two major rainfall types in this area, namely, non-bright band and bright band rain, are identified based on the observations from S-PROF. Several DSD and rainfall parameters are derived from the measured raindrop spectra, and the microphysical characteristics are investigated for different rainfall types and terrains (i.e., valley vs mountain).

A torrential flood event occurred over the complex terrain in Zhejiang on 23 June 2015 is investigated utilizing the first routinely operational C-band polarimetric radar in China. The polarimetric radar quantitative precipitation estimation (QPE) performance is verified using a local rain gauge network, and the microphysical characteristics of the rainstorm are revealed based on polarimetric radar variables. The results show that (i) radar QPE estimator based on KDP performs the best among all estimators; (ii) the rainstorm upon the rainfall center area was dominated by small-moderate sized raindrops with mean volume diameter less than 2 mm, but high raindrop concentration with log10(Nw) exceeding 5 mm-1m-3, resulting in rather high liquid water content; (iii) small hailstones were falling and melting upon the intense rainfall area between gauge stations, which compounded the complex precipitation structure and enhanced raindrop concentration.

Recent advances in low-power millimeter wave low-noise amplifier technologies have enabled the host of high performance atmospheric sounding sensors on ground-based instruments. In this paper, frequencies at K-band (22-31 GHz) and V-band (51-59 GHz) are chosen to retrieve atmospheric temperature and water vapor profiles. The specifications including sensitivity, integral time along with the calibration accuracy for all channels are presented. The key techniques such as calibration method and retrieving algorithm are described. In addition, measurements during two on-site experiments are presented to demonstrate the prototype performance in terms of calibrated brightness temperatures and retrievals (temperature and humidity) using radiosonde data as validation references. Besides the meteorological applications, a potential climate research in different seasons is provided based on a long term of observations.

Precipitation is a critical element in global atmospheric circulation and ecosystem. However, it is challenging to monitor the global or even continental scale precipitation features using ground-based rain gauges due to the spatial coverage limitations. The Geostationary Operational Environmental Satellite-R (GOES-R) series provide new opportunities for continuous observation of precipitation at large scales. This paper presents a detailed cross-comparison of quantitative precipitation estimates (QPE) between GOES-R (GOES-16) and the ground radar-based rainfall product derived from the National Ocean and Atmospheric Administration's (NOAA) multi-radar multi-sensor (MRMS) system over the continental United States (CONUS). The precipitation detectability of GOES-R is investigated using MRMS products as references. In addition, uncertainties associated with the GOES-R precipitation estimates are quantified in different regions over the CONUS.

Precipitation nowcasting is one of the most important applications of weather radars. Essentially, the nowcasting performance is determined by the radar data quality and the implemented nowcasting model. It is well known that the dual-polarization radar can observe more detailed microphysical information about precipitation and significantly enhance the radar data quality compared to traditional single-polarization radars. In this study, we extend an ensemble-based probabilistic precipitation nowcasting scheme for high-resolution polarimetric radar applications focusing on fast-evolving convective precipitation events. In particular, the nowcasting scheme blends an extrapolation technique with a probabilistic nowcasting method that is adapted to work on polarimetric radar-derived precipitation estimates. Case studies based on the S-band Weather Surveillance Radar - 1988 Doppler (WSR-88D) in Fort Worth, Texas (i.e., KFWS radar) show that the nowcasting products agree well with the real observations. However, the accuracy of short-term rainfall nowcasting is greatly influenced by the motion field, and it is still a challenge to predict the initiation of new precipitating clouds.

Space-based precipitation products are often used for regional and/or global hydrologic modeling and climate studies. A number of precipitation products at multiple space and time scales have been developed based on satellite observations. However, their accuracy is limited due to the restrictions on spatiotemporal sampling of the satellite sensors and the applied parametric retrieval algorithms. Similarly, a ground-based weather radar is widely used for quantitative precipitation estimation (QPE), especially after the implementation of dual-polarization capability and urban scale deployment of high-resolution X-band radar networks. Ground-based radars are often used for the validation of various spaceborne measurements and products. This article introduces a novel machine learning-based data fusion framework to improve the satellite-based precipitation retrievals by incorporating dual-polarization measurements from a ground radar network. The prototype architecture of this fusion system is detailed. In particular, a deep learning multi-layer perceptron (MLP) model is designed to produce the rainfall estimates using the geostationary satellite infrared (IR) data and low earth orbit satellite passive microwave (PMW)-based retrievals as inputs. The high-quality rainfall products from the ground radar network are used as the target labels to train this MLP model. An urban scale demonstration study over the Dallas-Fort Worth (DFW) metroplex is presented. In addition, the Climate Prediction Center morphing technique (i.e., CMORPH) is adopted for preprocessing of the satellite observations. Rainfall products from this deep learning system are evaluated using the standard CMORPH products. The results show that the proposed data fusion framework can be used for generating accurate precipitation estimates and could be considered as an alternative tool for developing future satellite retrieval algorithms.

Quantitative precipitation estimation (QPE) using operational weather radars in the western United States is still a challenging issue due to the beam blockage in the mountainous areas and complex rainfall microphysics induced by the orographic enhancement. This article aims to improve operational radar rainfall estimates in complex terrain by incorporating auxiliary remote sensing observations. An innovative vertical profile of reflectivity (VPR) correction scheme is developed for operational radar using observations from multiple vertically pointing profilers to represent the vertical structure of precipitation at various locations. A demonstration study in the Russian River basin in Northern California is detailed. Results show that the QPE performance is significantly improved after VPR correction, and this new VPR correction approach is superior to the conventional approach currently applied in the operational radar rainfall system. The normalized standard error of hourly rainfall estimates for the two precipitation events presented in this article is improved by ~20% after applying the proposed VPR correction scheme.

Effective utilization of the changing precipitation microphysics in real-time radar quantitative precipitation estimation (QPE) is challenging, which requires dynamic adjustment of the radar reflectivity (Z) and rain rate (R) relations. This article develops and demonstrates two dynamic radar rainfall approaches using 16 Doppler weather radars and 4579 surface rain gauges deployed over the Eastern Jiang Huai River Basin (EJRB) in China. Both approaches are derived based on the radar-gauge feedback mechanism. Although the Z-R relations in both approaches are dynamically adjusted within a precipitation system, one is using a single global optimum (SGO) Z-R relation, while the other is using different Z-R relations for different storm cells identified by a storm cell identification and tracking (SCIT) algorithm. Four precipitation events featured by different rainfall characteristics are investigated to evaluate the performances of various QPE methodologies. In addition, the shortterm vertical profile of reflectivity (VPR) clusters is extensively analyzed to resolve the storm-scale characteristics of different storm cells. The evaluation results based on independent gauge observations show that both rainfall approaches with dynamic Z-R relations perform much better than the fixed Z-R relations. The adaptive approach incorporating the SCIT algorithm and real-time gauge measurements has the best performance since it can better capture the spatial variability and evolution of precipitation.

An inverse mapping table (IMT) method is proposed in this article to retrieve the raindrop size distribution (RSD) parameters from X-band polarimetric weather radar data. In the IMT method, a forward mapping database from three parameters of a gamma-type RSD to polarimetric radar variables is first built based on the scattering simulations under ideal atmospheric conditions, and then an inverse mapping database is derived. In particular, given a fixed shape parameter (μ) of RSD, the intersection of horizontal reflectivity (ZH) and differential reflectivity (ZDR) contour lines is first obtained in the domain of total number concentration (NT) and median volume diameter (MVD) D0; and the inverse mapping relationship between ZHand ZDRto NTand D0at a fixed μ value is derived to form a single layer of IMT. Then, the monotonic relationship between μ and the specific differential propagation phase shift (KDP) or backscatter differential phase (δC) can aid in determining μ and a single layer of IMT. Thus, the inverse mapping database from polarimetric observations to the three gamma-type RSD parameters μ, NT, and D0can be established. Demonstration studies during a convective rainfall event and a large-scale rainfall event which occurred in northeastern China are carried out to examine the performance of this IMT method compared to a constrained-gamma (C-G) method that uses empirical relations between RSD parameters. The results show that the IMT method has a better performance in the convective case and similar performance in the large-scale continuous rainfall case.

Quantitative Precipitation Estimation (QPE) is one of the most important applications of weather radars. However, it is difficult to obtain an optimal Z-R relation that can represent the local variability in precipitation microphysics. This study implements an adaptive radar QPE scheme based on the Storm Cell Identification and Tracking (SCIT) algorithm for Doppler radar and rain gauge measurements from Southwest China. The performance of this innovative QPE system is demonstrated using four precipitation events featured by different meteorological characteristics. Evaluation results show that the SCIT-based rainfall approach performs better than conventional Z-R based algorithms.

The evolution process of hurricane Sandy (Oct 20-10.29, 2012) was analyszed using satellite remote sensing observations on polar-orbiting(MWHS on Chinese meteorological satellite FY-3B, GMI on GPM satellite) and simulations on geostationary orbit (designed for FY-4M), conventional datasets and forcasting utilizing the mesoscale Weather Research and Forecasting (WRF) model at temporal and spatial resolution of 15 km and 5 minutes. WRF Data Assimilation (WRFDA) model is utilized to forecast the track and intensity of hurricane Sandy with radiance described above, and the forecast results are cross-compared with the best track to verify the track and intensity of previous forecasts and analysis results, which proposes a kind of novel observations to improve the track and intensity for future hurricanes.

Typhoon is often accompanied by strong convective weather such as high winds and heavy rainfall. The dualpolarization radar observations show a great performance for quantitative precipitation estimation (QPE) over traditional single-polarization radar by gleaning more useful information about precipitation particle size and shape distributions. According to the simulation of raindrop size distribution data from disdrometers in Guangzhou, China, this paper established a localized QPE algorithm for three typhoons landing in southern China. This regional QPE algorithm was compared with various rainfall relations such as the default Z- R relationship (Z = 300R1.4) adopted by the WSR-88D. Results show that the rainfall estimates with localized algorithms approximate to rainfall observations from surface meteorological stations. The localized QPE algorithm is more accurate and reliable than other algorithms for typhoon events over southern China.

A simulator of weather radar signals can be exploited as a useful reference for many applications, such as weather forecasting and nowcasting models or for training artificial intelligence systems designed to optimize the trajectory of aircrafts with the purpose to reduce flight hazard and fuel consumption. However, before being used, it must be accurately examined under different operating conditions, in order to evaluate the consistency of the outputs produced. In this paper, we present a validation procedure for a newly developed polarimetric weather radar simulator (POWERS). The goal is to assess the ability of the simulator to deal with any kind of input data, be they simulated and real raindrop-size distributions, or outputs generated by numerical weather prediction models. Three different approaches are proposed, each providing a connection between meteorological inputs and the radar observables simulated by POWERS. The analysis is carried out in the case of rainfall, both at S- and X-bands.

The Global Precipitation Measurement (GPM) mission Core Observatory satellite, carrying the first space borne dual-frequency precipitation radar (DPR) operating at Ku/Ka band, and the GPM microwave imager (GMI), was launched on 27 February 2014 [1]. The GPM satellite extend the observation range attained by Tropical Rainfall Measuring Mission (TRMM) from tropics to most of the globe and provide accurate measurement of rainfall and snowfall. The observation of global precipitation plays an important role in improving the capabilities of weather, climate, and hydrological predictions.

The NASA dual-frequency, dual-polarization, Doppler radar (D3R) [1] was conceived and developed to support ground validation (GV) operations of the Global Precipitation Measurement (GPM) mission [2]. The D3R operates in the same frequencies bands, Ku- and Ka-band, as GPM's dual-frequency precipitation radar enabling direct comparisons of microphysical observations of precipitation. The D3R radar is shown in Figure 1. To support the GPM GV mission, D3R substantively participated in four field campaigns in North America with diverse geographic features covering both winter and summer conditions.

The Olympic Mountains Experiment (OLYMPEx) is a ground validation (GV) field campaign conducted, in part, to support the GV efforts of the U.S./Japan Global Precipitation Measurement (GPM) mission. The primary goal of OLYMPEx is to validate precipitation measurements in mid-latitude frontal systems moving from ocean to coast to mountains and to determine how remotely sensed measurements of precipitation by GPM can be applied to a range of hydrologic, weather forecasting and climate data. A variety of instruments were deployed for OLYMPEx, including ground radars, disdrometers, rain gauges, and airborne radars. This paper characterizes the microphysical properties of precipitation over the OLYMPEx regime, using observation from NASA dual-polarization radars and disdrometers. Particularly, statistical distribution of raindrop size distribution (DSD) parameters measured by a number of disdrometers is investigated. The dual-polarization radar observations at S- and Ku-band frequencies are used to classify precipitation types, and quantitatively estimate rainfall intensity.

Urban flash floods can occur immediately after heavy rainfall due to urban characteristics including impervious surface and complex drainage systems. Early warnings of such events, even by 5-10 min, are crucial in terms of protecting personal and property safety. Since 2012, the center for Collaborative Adaptive Sensing of Atmosphere (CASA), in collaboration with the National Weather Service (NWS) and North Central Texas Council of Governments (NCTCOG), has initiated the effort to deploy a dense radar network in Dallas-Fort Worth (DFW) for urban weather disaster detection and mitigation [1, 2]. The DFW Metroplex is one of the largest inland metropolitan areas in the U.S., and is also among the fastest growing major urban areas in the country. Every year the DFW area experiences a wide range of natural hazardous weather events including high wind, flash flood, tornado, and hail, etc. [2]. It is an ideal location to demonstrate the application of dense radar network for urban weather sensing and disaster management. As shown in Figure 1, centered in the DFW urban remote sensing network is the deployment of a NWS S-band radar system and eight dual-polarization X-band radars that can provide coverage to most of the 6.5 million people in this region. Based primarily on these radars, a number of product systems have been developed for real time warning operations, including the real-time multiple Doppler wind retrieval system [3], quantitative precipitation estimation (QPE) [4, 5, 6] and nowcast system.

This paper introduces an algorithm of radar quantitative precipitation estimation (RQPE) derived from a S-band dual-polarization weather radar network over southern China. The observations obtained from Zhuhai radar was used to demonstrate the performance of the RQPE algorithm. This initial regional study on RPQE in southern China provides reference on RPQE algorithm over China in near future when all weather radars will be upgraded with dual-polarization capability throughout China.

The San Francisco Bay Area is home to over 7 million people, the fifth largest population center in the United States. This region also supports one of the most prosperous economies in the U.S. A recent report by the State of California's Department of Water Resources has emphasized that the Bay Area is at risk of catastrophic flooding [1]. Monitoring heavy rainfall events and mitigation of their associated negative impacts are critical for protecting life and property in this region. To this end, accurate quantitative precipitation estimation (QPE) and forecast (QPF) are required to provide forecasters with sufficient understanding of rapidly changing weather conditions, and allow them to issue timely watches and warnings.

The NASA D3R was successfully deployed and operated throughout the NASA OLYMPEx field campaign. A differential phase based attenuation correction technique has been implemented for D3R observations. Hydrometeor classification has been demonstrated for five distinct classes using Ku-band observations of both convection and stratiform rain. The stratiform rain hydrometeor classification is compared against LDR observations and shows good agreement in identification of mixed-phase hydrometeors in the melting layer.

How to eliminate the threats from tornadoes and high winds using advanced weather sensing techniques has been pursued for a few decades. However, conventional S-band radar based retrieval system is limited at warning against tornadoes and microbursts due to the combined effects of sampling limitation and Earth's curvature. This paper presents the rationale of using high-resolution X-band Doppler radar network for tornado detection and 3D wind velocity retrieval. The distributed collaborative adaptive sensing (DCAS) paradigm designed for the X-band radar network is able to reconfigure each radar node according to real-time weather changes. The wind retrieval system then selects the best radar pairs for multi-Doppler synthesis based on optimal beam-crossing angles in the target areas. The real-time wind products derived based on radar network observations in “Tornado Alley” have demonstrated the excellent performance of the designed multi-Doppler system. This real-time system is now being used for emergency warnings of severe weather by the local National Weather Service Forecast Offices.

Weather radar signals at high frequencies such as Ku-band are attenuated along the propagation path through rainfall. Hence, reflectivity and differential reflectivity measurements at such frequencies should be corrected for attenuation before any quantitative applications such as the retrieval of raindrop size distribution (DSD), which is a fundamental descriptor of rainfall microphysics. This paper presents the attenuation correction algorithm implemented for NASA Dual-frequency Dual-polarized Doppler Radar (D3R) Ku-band observations. The dual-polarization based correction performance is evaluated with the self-consistency criterion. In addition, the DSD parameters are estimated with the attenuation corrected observations, and the preliminary results are shown.

Quantitative precipitation estimation (QPE) continues to be one of the principal objectives for weather researchers and forecasters. The purpose of this research is to present the development of a regional dual polarization QPE process known as the RAdar Multi-Sensor QPE (RAMS QPE). This scheme applies the dual polarization radar rain rate estimation algorithms developed at Colorado State University into an adaptable QPE system. The methodologies used to combine individual radar scans, and then merge them into a mosaic are described. The implementation and evaluation is performed over a domain that covers South Carolina for a severe rainfall event occurring October 2, 2015 through October 4, 2015. The QPE precipitation fields evaluated in this analysis will stem from the dual polarization radar data obtained from the local NWS WSR-88DP.

A probabilistic attenuation correction technique for a single-polarization networked radar environment is proposed. The proposed technique, based on the Bayesian theory, makes a maximization of a likelihood function of Hitschfeld-Bordan (HB) reflectivity obtained by each radar node. A variance of the HB reflectivity (σHB2) is defined and regarded as instability of each HB reflectivity in the proposed technique. In the X-band simulation based on S-band real radar data, it is revealed that the corrected reflectivity obtained by the proposed technique has good accuracy, and the proposed technique works more stably than the HB technique. The proposed technique is also performed using CASA IP-1 dual-polarization radar network observations, and the corrected reflectivity by the proposed technique has a good agreement with differential phase (ΦDP)-based corrected reflectivity.

In order to investigate how well observations from precipitation-monitoring satellites match up to the best estimate of the true precipitation measured at ground level and how to use the collected precipitation data to evaluate models that describe and predict the hydrology, the Integrated Precipitation and Hydrology Experiment (IPHEx) was conducted in the southern Appalachian Mountains in the eastern United States from May 1 to June 15, 2014. The NASA dual-frequency dual-polarization Doppler radar (D3R), co-located with NASA NPOL radar, was deployed as part of the IPHEx field campaign to characterize precipitation properties at Ku- and Ka-band frequencies. This paper presents the deployment and performance of D3R during the IPHEx field experiment. Sample observations will be presented, with particular attention paid to cross-comparison between D3R and NPOL.

The light rain and snow are critical to the Earth's ecosystem due to the high occurrence rate, especially in middle and high latitude. However, it is challenging to use rainfall gauge to measure light rain due to the limitations of sampling time and bucket volume resolution. This paper presents the characterization and estimation of light rainfall using National Aeronautics and Space Administration (NASA) Dual-frequency Dual-polarization Doppler Radar (D3R) observations collected during the NASA Iowa Flood Studies (IFoodS) and Integrated Precipitation and Hydrology Experiment (IPHEx) field campaigns. Sample rainfall products are shown. Comparisons are performed between radar rainfall products and ground rainfall measurements from rain gauge and disdrometers. It is shown that the radar rainfall measurements agree with the disdrometer observations very well. The excellent performance for light rainfall estimation demonstrates one aspect of the capability of D3R as a ground validation tool for the Global Precipitation Measurement (GPM) satellite precipitation product evaluations.

The light rain (less than or equal to a few mm hr-1) is critical to the Earth's ecosystem due to the high occurrence rate, particularly in middle and high latitude (over 80%). However, it is challenging to use rainfall gauge to measure light rain due to the sampling time and bucket volume resolution. Dual-polarization radar has become an important tool for quantitative precipitation estimation because of its relatively large covering area and ability to fill the sampling void. This paper presents the application of Ku-band dual-polarization radar for light rainfall estimation. The Ku-band radar rainfall algorithms and their error structure are described. The Ku-band observations from the National Aeronautics and Space Administration (NASA) Dual-frequency Dual-polarization Doppler Radar (D3R) during the NASA Iowa Flood Studies (IFoodS) field campaign are used to derive the rainfall products. The comparisons are performed between radar rainfall estimates and ground rainfall measurements from rain gauge and Autonomous Parsivel Unit (APU). It is shown here that the radar rainfall measurements at different timescales (i.e., 5, 10, and 15 min) agree with the APU observations very well. The normalized difference error is about 26.1%, 24.8%, and 23.7%, for 5-min, 10-min, and 15-min rainfall accumulations, respectively. The excellent performance of Ku-band rainfall algorithm for light rain estimation indicates the great potential of using D3R as a ground validation tool for the Global Precipitation Measurement (GPM) satellite precipitation product evaluations.

The Iowa Flood Studies (IFloodS) field experiment was conducted to better understand the strengths and limitations of Global Precipitation Measurement (GPM) mission satellite products in the context of hydrologic applications. The NASA dual-frequency dual-polarization Doppler radar (D3R), designed as part of the GPM ground validation program, participated in the IFloodS field campaign to characterize precipitation properties at Ku- and Ka-band frequencies. This paper presents the deployment of the D3R and summarizes the D3R observations during the IFloodS field campaign. The quality of the D3R measurements is evaluated by comparing with the NASA NPOL S-band radar observations. In addition, the capability for rainfall estimation using the D3R is also described and validated using ground gauge measurements.

Neural network (NN) is a nonparametric method to represent the relation between radar measurements and rainfall rate. The relation is derived directly from a dataset consisting of radar measurements and rain gauge measurements. Tropical Rainfall measuring Mission (TRMM) Precipitation Radar (PR) is known to be the first observation platform for mapping precipitation over the tropics. TRMM measured rainfall makes a significant contribution to the study of precipitation distribution over the globe in the tropics. Ground validation (GV) is a critical component in the TRMM system. However, the ground sensing systems have quite different characteristics from TRMM in terms of resolution, scale, sampling, viewing aspect, and uncertainties in the sensing environments. In this paper a novel hybrid NN model is presented to train ground radars for rainfall estimation using rain gauge data and subsequently the trained ground radar rainfall estimation to train TRMM/PR observation based neural networks. This hybrid NN model provides a mechanism to link between gauges on the ground, the ground radar observations and the TRMM/PR observations. The dual-polarization radar measurements from a ground WSR-88DP site in Dallas-Fort Worth region and local rain gauge data will be used for the demonstration purpose. The performance of the rainfall product derived for TRMM PR is then compared against TRMM standard rainfall products. In addition, a direct gauge comparison study is done to examine the improvement brought in by this hybrid neural networks approach.

Rainfall estimation using dual-polarization radars has shown a number of advantages over traditional single polarization radars. As the building blocks for deriving dual-polarization radar rainfall algorithms, the rain drop size distribution (DSD) has been studied over three decades. In this study, we present the estimation of DSD parameters using dual-polarization radar measurements at S-, X-, and Ku-band frequencies. Various DSD retrieval algorithms are implemented using the data collected by the S-band WSR-88DP/KFWS, X-band CASA/XUTA, and D3R Ku-band radars.

The GPM core satellite will be ready to launch in February 2014, less than 7 months after the end of IGARSS2013 symposium. In the pre-launch era, several international validation experiments such as LPVEX (Light Precipitation Validation Experiment), MC3E (Midlatitude Continental Convective Clouds Experiment), and IFloodS (Iowa Flood Studies) have already generated a substantial set of measurements that continue to contribute to the development and test of pre-launch GPM algorithms. Following launch, it is expected that GPM ground validation will focus on evaluating precipitation data products, generated by a constellation of GPM satellites as well as assumptions made in algorithms. This paper presents simple concepts of dual-polarization radar observation strategies and products that can be generated for the post launch era of the GPM program, especially when there are satellite overpasses. The use of microphysical retrievals from ground-based radars and in-situ observations for validating the retrievals from space based observations is the main focus of this paper.

The use of weather sensing radar measurements along with corresponding gauge data in multisensor applications seek to provide reliable estimates of rainfall rate and accumulation versus single radar. Radar rainfall estimators have a number of advantages over gauges including the ability to observe precipitation over wider areas within shorter timeframes and providing advanced warning of impending precipitation events. The radar reflectivity-rainfall (Z-R) relations are traditionally used for quantitative precipitation estimation (QPE).

Flooding is one of the most catastrophic disasters in the world. Radar rainfall estimation for flash flood forecasting in small, urban catchments is an important accomplishment of CASA. A Kdp(specific differential propagation phase) based rainfall algorithm was developed using the Kdpvalues of X-band radars. The performance of this rainfall algorithm is evaluated for all major rainfall events using a gauge network in the center of CASA IP1 test bed for a 5- year period. The cross comparison with gauge estimates shows great improvement compared to the current state-of-the-art. Since the beginning of 2012, CASA has been involved in developing the first urban weather demonstration network in Dallas-Fort Worth (DFW) area. This paper will summarize the performance of the radar rainfall product in the IP1 network. In addition, the implementation of CASA QPE (quantitative precipitation estimation) system in DFW Urban Demonstration Network will also be presented.