Prudhvi Kumar Reddy
Bio: Prudhvi Kumar Reddy is an academic researcher from Vardhaman College of Engineering. The author has contributed to research in topics: Wavelet transform & Haar wavelet. The author has an hindex of 1, co-authored 1 publications receiving 5 citations.
••05 Jun 2013
TL;DR: A proposed model yields an average accuracy of 89.15 % in the identification of hail based on the square root balance sparsity norm threshold value obtained on compressing and de-noising the satellite image.
Abstract: Weather forecasting is a formidable challenge in the field of science as it depends on multiple parameters which are dynamic and chaotic. The rain, snow and hails are different climatic conditions that depend on the atmospheric parameters and are major forms of precipitation. Hailstorms are measured using traditional radar. Radar based hail measurements face major problems as the signals are weak and face attenuation issues with strong echoes. Hence, satellite or digital images are one of the efficient sources in the prediction of hail. In the process of image acquisition from the satellite imagery it would often find barriers like noise, burrs and so on, obscure or even cover the original image of an area or can reduce the image quality which include lot of noise. Therefore wavelet transform is used to enhance the image or to eliminate striping noise. They have advantages over traditional wavelet methods in analyzing physical situations where the signals are discontinuities. One of the wavelet transform used in the prediction of hail for a satellite or digital image is the haar wavelet transform. Differentiation between rain and hail depends on the square root balance sparsity norm threshold value obtained on compressing and de-noising the satellite image. The proposed model yields an average accuracy of 89.15 % in the identification of hail.
TL;DR: Support Vector Machine (SVM) based Real Time Hand-Written Digit Recognition System to recognize user given handwritten digits in real time is presented.
Abstract: Meanwhile Neural Networks based algorithms have intimated steadfast potential on various visual tasks including the recognition of Digits. This paper presents Support Vector Machine (SVM) based Real Time Hand-Written Digit Recognition System. The system involves two main sections i.e. training and recognition section. SVM classifier is used as the training algorithm and then tested it on MNIST dataset. We achieved a training accuracy of 98.05% and a test accuracy of 97.83% demonstrating that the proposed method can achieve significant and promising performance in digit recognition. Then we implemented our model to recognize user given handwritten digits in real time.
TL;DR: In this article, the performance of the TRMM Precipitation Radar and TRMM Microwave Imager is further investigated for detecting hailstorms using a 16-yr precipitation feature database derived from the Tropical Rainfall Measuring Mission (TRMM) satellite.
Abstract: In previous studies, remote sensing properties of hailstorms have been discussed using various spaceborne sensors. Relationships between hail occurrence and strong passive microwave brightness temperature depressions have been established. Using a 16-yr precipitation-feature database derived from the Tropical Rainfall Measuring Mission (TRMM) satellite, the performance of the TRMM Precipitation Radar and TRMM Microwave Imager is further investigated for hail detection. Detection criteria for hail larger than 19 mm are separately developed from Ku-band radar reflectivity and microwave brightness temperature properties of precipitation features that are collocated with surface hail reports over the southeastern and south-central United States. A threshold of 44 dBZ at −22°C is found to have the highest critical success index and Heidke skill score. The threshold of 230 K at 37 GHz yields the best scores among passive microwave properties. Using these two thresholds, global distributions of possible ...
••15 May 2014
TL;DR: This is the first research study carried on thunderstorm prediction using the clustering and wavelet techniques resulting with higher accuracy, and the proposed model yields an average accuracy of 89.23%.
Abstract: Thunderstorm is a sudden electrical expulsion manifested by a blaze of lightening with a muffled sound. It is one of the most spectacular mesoscale weather phenomena in the atmosphere which occurs seasonally. On the other hand, prediction of thunderstorms is said to be the most complicated task in weather forecasting, due to its limited spatial and temporal extension either dynamically or physically. Every thunderstorm produce lightening, this kills more people every year than tornadoes. Heavy rain from thunderstorm leads to flash flooding, and causes extensive loss to property and other living organisms. Different scientific and technological researches are been carried on for the forecasting of this severe weather feature in advance to reduce damages. In this regard, many of the researchers proposed various methodologies like STP model, MOM model, CG model, LM model, QKP model, DBD model and so on for the detection, but neither of them could provide an accurate prediction. The present research adopted clustering and wavelet transform techniques in order to improve the prediction rate to a greater extent. This is the first research study carried on thunderstorm prediction using the clustering and wavelet techniques resulting with higher accuracy. The proposed model yields an average accuracy of 89.23% in the identification of thunderstorm.
TL;DR: A deep convolutional neural network is developed with a large collection of radar images as input to train and validate a classification model, and then the model is used to detect hailstorm events.
Abstract: With the improvement of sensing and storing technologies, a large amount of weather data become available, and the data size will continue growing as radar imaging instruments continuously acquire data. In this work, we develop a deep convolutional neural network with a large collection of radar images as input to train and validate a classification model, and then we use the model to detect hailstorm events. This is interdisciplinary work between the disciplines of computer science and meteorology. We are primarily interested in what hailstorm features the network learns and how it learns as convolving into deeper iterations. The evaluation results show a high classification accuracy in comparison with existing hailstorm detection approaches. The proposed approach can also be used to detect other types of severe weather events with minimal efforts on variable or parameter changes.
21 Jun 2019
TL;DR: In this article, a neural network model is built based on an image semantic segmentation technology of deep learning; according to different characteristics of real echo data and noise data in the radar echo image at continuous time points, the radar image is divided by using the trained neural network, so that the denoising of the image is realized, the interference is reduced, and the accuracy of meteorological prediction is improved.
Abstract: The embodiment of the invention provides a radar image denoising system and method and computer equipment, and the system comprises a first unit which obtains radar echo images which are continuous intime; a second unit which is used for inputting the radar echo images which are continuous in time into a trained neural network model; And a third unit which is used for dividing real echo data andnoise data based on the spatial characteristics and the time characteristics of the data in the radar echo image and the neural network model. The embodiment of the invention provides a radar image denoising system. a neural network model is built based on an image semantic segmentation technology of deep learning; According to different characteristics of real echo data and noise data in the radar echo image at continuous time points, the radar echo image is divided by using the trained neural network model, so that the denoising of the image is realized, the interference is reduced, and theaccuracy of meteorological prediction is improved.
TL;DR: In this article , the authors used the GOES-16 satellite data collected in the mountainous Córdoba region in collaboration with the citizen science program "Cosecheros de Granizo 2018-2020" including from a record-breaking hail event and from the 2018-2019 RELAMPAGO field campaign.
Abstract: Córdoba Province in Argentina is a global hotspot for deep hail-producing storms. Previous studies of hail formation and detection largely relied on satellite snapshots or modeling studies, but lacked hail validation, relying instead on proxy metrics. To address this limitation, this study used hail collected in the mountainous Córdoba region in collaboration with the citizen science program “Cosecheros de Granizo 2018–2020” including from a record-breaking hail event and from the 2018–2019 RELAMPAGO field campaign. Three cases including a MCS and two supercells, which have verified hail in different environment locations relative to the Sierras de Córdoba, were analyzed for multi-spectral signatures in GOES-16 satellite data. Brightness temperatures decreased over time after convective initiation, reaching values cooler than the tropopause with variations around those values of different magnitudes. Overall, all cases exhibited a slight weakening of the updraft and strong presence of smaller ice crystal sizes just prior to the hail report, especially for the larger hailstones. The results demonstrate promise in using satellite proxies for hail detection in multiple environments for different storm modes. The long-term goal is to better understand hail-producing storms and unique challenges of forecasting hail in this region.