In earlier work we have demonstrated that TC clouds are, on average, net cooling in that they increase the upwelling radiation. This is due to a larger increase in daytime short wave reflectivity compared with the corresponding decrease in long wave emission that happens at night. While the trend computed over 20 years was for TC clouds to be cooling, there was approximately a 3:1 ratio between individual storms that were net cooling and net warming. In that earlier work, we noted that late-season storms tended to be preferentially warming. Here we demonstrate that this effect is dominated by local solar angle (which is highly correlated with seasonality) and over TC cloud extent.

We present progress towards an automated system that can segment tropical cyclones (TCs) and their neighboring regions from geostationary infrared brightness temperature images. The purpose of these segmentation maps is to provide an area that can be used to compute the contribution of TCs to the upwelling radiation budget. Our previous work has identified regions of TC clouds, but it is known that TCs impact a larger area than just those covered by their clouds. Hence it is necessary to properly label both the TC clouds and the associated clear-sky regions in the vicinity of the TCs. Here we present a convolutional neural network method that can be used to reproduce cloud masks generated with our earlier, first-principles algorithm. We also discuss our efforts to create an extended training set of TC masks that include both clouds and clear sky.

We demonstrate that a convolutional neural network (CNN) based on the U-Net architecture can be used to create a cloud mask data set that accurately identifies the clouds associated with tropical cyclones (TCs). The CNN can be trained using a single year of cloud masks produced by an earlier first-principles algorithm, and the results are insensitive to the specific year of training data used. These masks were originally created in order to compute the upwelling radiation due to TC clouds, and we show that the predicted masks result in both pixel areas and radiation calculations that are nearly identical to those computed using the earlier masks.

A wind radii model using the deviation angle variance technique was applied to 523 tropical cyclones (TCs) in the Western North Pacific region from 1998 to 2018. The model, which uses high resolution satellite cloud images, was used to investigate the impact of the 34-, 50- and 64-kt winds from landfalling TCs. The wind impact analysis showed that the regions most exposed to TC tracks of Taiwan and the Philippines experienced 34-kt winds annually, and that these winds extended far into the mainland (up to approximately 500 km). The 50- and 64-kt winds occurred less frequently along the length of the Philippines but only extended down to the center of Vietnam and reached around 100 to 200 km inland

The deviation angle variance (DAV) method was developed to objectively estimate tropical cyclone (TC) intensity from geostationary infrared (IR) brightness temperature data. Here, we demonstrate that improvements of 25% root mean square error (RMSE) in major hurricane intensity estimation (relative to best track) can be obtained by considering the pixel-by-pixel satellite view angle in the estimation. Using data from the Chinese Fengyun 2E and 2F satellites for Super Typhoon Soudelor (2015), we demonstrate how the satellite observation angle can reduce the accuracy of intensity estimation, especially for the strongest TCs. Based on these results, an improved DAV estimator is developed using 12 years (2004-2015) of Geostationary Operational Environmental Satellite (GOES)-East satellite IR images over the North Atlantic basin.

Based on 12 years (2004-2015) of GOES-East satellite infrared (IR) imagery over the North Atlantic basin, the diurnal cycle of the tropical cyclone (TC) Deviation Angle Variation (DAV) value is analysed, and a backward 24-hr time average DAV is selected to filter the noise in the DAV-TC intensity estimation. The effect of satellite observation angle on the DAV-TC intensity estimation is analysed in theory, in a case study, and in the longer-term statistics. Based on these results, an improvement to the DAV-TC intensity estimation is presented and evaluated in this study. The results show that, after taking the observation angle into account, the new DAV-TC intensity estimation is shown to produce smaller errors and higher correlations than the previous versions, especially for major hurricanes.

This study aims to quantify the portion of the Earth's outgoing radiation that is attributable to tropical cyclones (TCs). To accomplish this, we have developed a method that starts with an image processing algorithm which labels cloud pixels associated with a TC, based on the time series of brightness temperature images and best-track data. The labels attributable to the TC are then combined with radiation data to obtain the TC-related radiation throughout its lifetime. Preliminary results are shown for the North Atlantic Ocean in 2012 and 2013: In 2012, the average TC shortwave and longwave radiation contributed 0.039 PW (or 0.35%) and 0.099 PW (or 0.34%), respectively, to the total regional radiation; In 2013, the contribution due tot TCs decreased to 0.022 PW (or 0.19%) for SW, and 0.059 PW (or 0.20%) for LW radiation.

A multiple linear regression wind radii model developed in the North Atlantic basin based on the deviation angle variance technique was applied in the Australian region. The model was used to improve the historic database of 34-, 50- and 64-kt tropical cyclone (TC) wind radii estimates (R34, R50, and R64) for 374 TCs during the geostationary satellite era. Results during 2010-2016 produced quadrant mean absolute errors ranging between 49 and 61 km for the 34-kt radii, between 24 and 37 km for the 50-kt radii and between 20 and 25 km for the 64-kt radii.

The deviation angle variance (DAV) method is an objective tool for estimating the intensity of tropical cyclones (TCs) using geostationary infrared (IR) brightness temperature data. At early stages in TC development, the DAV signal can be also a robust predictor of tropical cyclogenesis. However, one of the problems with using the DAV method at these early stages is that the operator has to subjectively track potentially developing cloud systems, sometimes before they are clearly identifiable. Here, we present a method that allows us to automatically track the evolution of cloud clusters using only the raw IR imagery and the resulting DAV maps. We have compared our objective method with results manually obtained on a limited data set spanning a 12-day period during the 2010 hurricane season in the western North Pacific and tuned the performance of the method to the manual results. The performance of the method was then tested by comparing the results with best track and invest files produced by the Joint Typhoon Warning Center for the four-year period 2009-2012. The long-term results agree well with the best track and invest files for the disturbances analyzed in terms of start time, end time, and locations of disturbances. The automatic tracking method presented in this letter may be used to reduce the dependence of tropical cyclogenesis DAV analyses on the expertise and ability of the operator.

An objective technique to discriminate developing from nondeveloping cloud clusters during tropical cyclogenesis is described. Since vortices are characterized by high levels of organization or axisymmetry, their detection at early stages of the lifecycle of tropical cyclones makes it possible to determine when they form. To quantify the axisymmetry of a cloud cluster around a predefined radius, a statistical analysis of the orientation of the brightness-temperature gradient is performed. Results show that early detections of axisymmetric structures reliably indicate the cyclone genesis on an average of 0.6 h before an atmospheric disturbance becomes a tropical depression. In addition, the technique shows potential to discriminate nondeveloping from developing cloud clusters. A statistical analysis shows that a true-positive rate of detection of approximately 93% could be achieved with a false-alarm rate of 22%.

Significant advances have been made in developing normal-incidence sensitive quantum-dot infrared photodetectors (QDIPs) that can exhibit spectral responses tunable through the bias voltages applied. This tunability makes it possible to build spectral imaging system in IR range based on single QDIP, without any spectral dispersive device upfront. To achieve such adaptivity, algorithms must be developed to find the optimized operation bias voltages set which maximizes the spectral context inside the output data while reducing the data redundancy. In this paper, we create a new, general defination of signal-to-noise ratio (SNR) in spectral space, based on a geometrical spectral imaging model recently developed. With the new SNR definition, a scene-independent set of bias voltages is selected to maximize the average SNR of the sensor. Then, some bias voltages can be added or removed. This dynamic optimization process is performed throughout the imaging process so that the balance between data information and data volume is always achieved.

The main focus of this paper is a rigorous development and validation of a novel canonical correlation feature- selection (CCFS) algorithm that is particularly well suited for spectral sensors with overlapping and noisy bands. The proposed approach combines a generalized canonical correlation analysis framework and a minimum mean-square-error criterion for the selection of feature subspaces. The latter induces ranking of the best linear combinations of the noisy overlapping bands and, in doing so, guarantees a minimal generalized distance between the centers of classes and their respective reconstructions in the space spanned by sensor bands. To demonstrate the efficacy and the scope of the proposed approach, two different applications are considered. The first one is separability and classification analysis of rock species using laboratory spectral data and a quantum-dot infrared photodetector (QDIP) sensor. The second application deals with supervised classification and spectral unmixing, and abundance estimation of hyperspectral imagery obtained from the Airborne Hyperspectral Imager sensor. Since QDIP bands exhibit significant spectral overlap, the first study validates the new algorithm in this important application context. The results demonstrate that proper postprocessing can facilitate the emergence of QDIP-based sensors as a promising technology for midwave- and longwave-infrared remote sensing and spectral imaging. In particular, the proposed CCFS algorithm makes it possible to exploit the unique advantage offered by QDIPs with a dot-in-a-well configuration, comprising their bias-dependent spectral response, which is attributable to the quantum Stark effect. The main objective of the second study is to assert that the scope of the new CCFS approach also extends to more traditional spectral sensors.

An objective technique for obtaining features associated with the shape and dynamics of cloud structures embedded in tropical cyclones from satellite infrared images is described. As the tropical cyclone develops from an unstructured cloud cluster and intensifies, the cloud structures become more axisymmetric about an identified reference point. Using variables derived from remotely sensed data, the technique calculates the gradient of the brightness temperatures to measure the level of symmetry of each structure, which characterizes the degree of cloud organization of the tropical cyclone. The results presented show that the technique provides an objective measure of both the structure and the intensity of the tropical cyclone from early stages, through intensification, maturity, and dissipation.

A multistage projection pursuit (PP) approach is applied to the classification of tropical cyclones (TCs) during extratropical transition (ET) using 500-hPa Navy Operational Global Atmospheric Prediction System (NOGAPS) geopotential height analyses. PP algorithms reduce the dimensionality of the high-dimensional data while minimizing the loss of information that discriminates among classes of ET types. In this paper, a prediction system is developed for ET TC classification and is applied to 85 western North Pacific storms during 1997-2004 using NOGAPS geopotential height analyses to group them as either intensifiers or dissipators. A classification is developed based on the 1997-2002 analyses, and the forecasting performance of the technique is tested on the storms from 2003 and 2004 in two different ways. In the first, the technique is applied at individual times utilizing just spatial information. In the second, the change in spatial patterns with time is taken into account with spatiotemporal algorithms. The preliminary classification results are slightly less accurate than those of NOGAPS but are found to be promising, and further possible improvements are discussed

Remote sensing of the earth's resources from space-based sensors has evolved in the past forty years from a scientific experiment to a commonly used technological tool. The scientific applications and engineering aspects of remote sensing systems have been studied extensively. However, most of these studies have been aimed at spectral sensors with non- overlapping bands, which means that the spectral responses of different bands have little between-band overlaps. Numerous conventional spectral sensors have dispersive bands which collect independent data in wavelength order [1].

Traditional spatial decompositions such as application of Empirical Orthogonal Functions and Principal Components Analysis (PCA) of meteorological field data have received wide attention in numerous forecasting and analysis applications. The method has been used to identify the important spatial patterns in meteorological fields. Defining the variability between various cases of tropical cyclones is another application of the method. We are utilizing spatial and temporal data as a tool to differentiate the TCs that complete and fail to complete extratropical transition in the western North Pacific. It has been shown that EOF Analysis is a promising tool in the prediction of reintensification of TCs after ET

Pulsed power is a technology that is suited to drive electrical loads requiring very large power pulses in short bursts (high-peak power). Certain applications require technology that can be deployed in small spaces under stressful environments, e.g., on a ship, vehicle, or aircraft. In 2001, the U.S. Department of Defense (DoD) launched a long-range (five-year) Multidisciplinary University Research Initiative (MURI) to study fundamental issues for compact pulsed power. This research program is endeavoring to: 1) introduce new materials for use in pulsed power systems; 2) examine alternative topologies for compact pulse generation; 3) study pulsed power switches, including pseudospark switches; and 4) investigate the basic physics related to the generation of pulsed power, such as the behavior of liquid dielectrics under intense electric field conditions. Furthermore, the integration of all of these building blocks is impacted by system architecture (how things are put together). This paper reviews the advances put forth to date by the researchers in this program and will assess the potential impact for future development of compact pulsed power systems.

A new pseudocolor mapping strategy for use with spectral imagery is presented. This strategy is based on a principal components analysis of spectral data, and it capitalizes on the similarities between three-color human vision and high-dimensional hyperspectral datasets. The mapping is closely related to three-dimensional versions of scatter plots that are commonly used in remote sensing to visualize the data cloud. The transformation results in final images where the color assigned to each pixel is solely determined by the position within the data cloud. Materials with similar spectral characteristics are presented in similar hues, and basic classification and clustering decisions can be made by the observer. Final images tend to have large regions of desaturated pixels that make the image more readily interpretable. The data cloud is shown to be conical in nature, and materials with common spectral signatures radiate from the origin of the cone, which is not (in general) at the origin of the spectral data. A supervised method for locating the origin of the cone based on identification of clusters in the data is presented, and the effects of proper origin orientation are illustrated.