Engineering Papers⌕ Search

Engineering topics

Cecil, Dan

Publications and source records attributed to Cecil, Dan.

18 records
All sourcesNASA NTRSDOE OSTI
NASA NTRS2019

Machine Learning Lifecycle for Earth Science Application: A Practical Insight into Production Deployment

Earth science domain presents unique sets of problems that are increasingly being solved using data driven approaches. The availability of big Earth science data offers immense potential for Machine learning (ML) as evident from numerous research publications lately. However, many of these publications are not ending up as production applications mainly because the data scientists who develop the ML models are now expected to complete the ML lifecycle by deploying and scaling the models in production. We introduce ML lifecycle to the Earth science community including the opportunities and challenges that lie ahead in each phase of the lifecycle. We demonstrate the lifecycle using an Earth science problem that we used ML to address and transitioned to production.

Maskey, Manil↗
NASA NTRS2018

Deepti: Deep Learning-Based Tropical Cyclone Intensity Estimation

We present the development of a deep learning model for objective estimation of tropical cyclone intensity at a higher temporal frequency, deployment of the model in production, design and implementation of the tropical cyclone monitoring and intensity estimation system and development of an interactive portal for situational awareness and evaluation of intensity estimation.

Maskey, Manil↗
NASA NTRS2018

Monitoring Intense Thunderstorms in the Hindu-Kush Himalayan Region

SERVIR Goal: Use NASA EOS (Earth Observing System) assets to build early warning capabilities and facilitate timely disaster response for high impact weather events in the HKH (Hindu-Kush Himalayan) region. Objectives: 1. High Impact Weather Assessment Toolkit(HIWAT) for the HKH region; 2. Jointly develop HIWAT capabilities/training with ICIMOD (International Centre for Integrated Mountain Development); 3. Demonstrate capability in end-user environment; 4. Transition HIWAT to ICIMOD.

thunderstorms↗
NASA NTRS2018

Tropical Cyclone Intensity Estimation Using Deep Convolutional Neural Networks

Estimating tropical cyclone intensity by just using satellite image is a challenging problem. With successful application of the Dvorak technique for more than 30 years along with some modifications and improvements, it is still used worldwide for tropical cyclone intensity estimation. A number of semi-automated techniques have been derived using the original Dvorak technique. However, these techniques suffer from subjective bias as evident from the most recent estimations on October 10, 2017 at 1500 UTC for Tropical Storm Ophelia: The Dvorak intensity estimates ranged from T2.3/33 kt (Tropical Cyclone Number 2.3/33 knots) from UW-CIMSS (University of Wisconsin-Madison - Cooperative Institute for Meteorological Satellite Studies) to T3.0/45 kt from TAFB (the National Hurricane Center's Tropical Analysis and Forecast Branch) to T4.0/65 kt from SAB (NOAA/NESDIS Satellite Analysis Branch). In this particular case, two human experts at TAFB and SAB differed by 20 knots in their Dvorak analyses, and the automated version at the University of Wisconsin was 12 knots lower than either of them. The National Hurricane Center (NHC) estimates about 10-20 percent uncertainty in its post analysis when only satellite based estimates are available. The success of the Dvorak technique proves that spatial patterns in infrared (IR) imagery strongly relate to tropical cyclone intensity. This study aims to utilize deep learning, the current state of the art in pattern recognition and image recognition, to address the need for an automated and objective tropical cyclone intensity estimation. Deep learning is a multi-layer neural network consisting of several layers of simple computational units. It learns discriminative features without relying on a human expert to identify which features are important. Our study mainly focuses on convolutional neural network (CNN), a deep learning algorithm, to develop an objective tropical cyclone intensity estimation. CNN is a supervised learning algorithm requiring a large number of training data. Since the archives of intensity data and tropical cyclone centric satellite images is openly available for use, the training data is easily created by combining the two. Results, case studies, prototypes, and advantages of this approach will be discussed.

tropical cyclone intensity↗
NASA NTRS2016

A New Tropical Cyclone Dynamic Initialization Technique Using High Temporal and Spatial Density Atmospheric Motion Vectors and Airborne Field Campaign Data

Background: Initialization of tropical cyclones in numerical weather prediction (NWP) systems is a great challenge: Mass-wind field balance; Secondary circulation and heating; Asymmetries. There can be large adjustments in structure and intensity in the first 24 hours if the initial vortex is not in balance: Spurious gravity waves; Spin-up (model and physics). Existing mesoscale NWP model TC (Tropical Cyclone) initialization strategies: Bogus vortex, cold start from global analyses; 3DVAR or 4DVAR, possibly with synthetic observations; EnKF (Ensemble Kalman Filter); Dynamic initialization. Dynamic initialization allows vortex to have improved balance and physics spin-up at the initial time (e.g., Hendricks et al. 2013, 2011; Nguyen and Chen 2011; Fiorino and Warner 1981; Hoke and Anthes 1976). Himawari-8 geostationary satellite has capability of continuous imagery (10-minutes) over the full disk: New GOES-R satellites will have same capability. This will allow for unprecedented observations of tropical cyclones. However, current data assimila1on systems are not capable of ingesting such high temporal observations (Atmospheric Mo1on Vectors - AMVs). Hourly AMVs are produced, and thinned to 100-kilometer spacing in the horizontal. An entirely new data assimilation concept is required to utilize these observations.

remote sensing↗
NASA NTRS2015

A Prototype Hail Detection Algorithm and Hail Climatology Developed with the Advanced Microwave Sounding Unit (AMSU)

In previous studies published in the open literature, a strong relationship between the occurrence of hail and the microwave brightness temperatures (primarily at 37 and 85 GHz) was documented. These studies were performed with the Nimbus‐7 SMMR, the TRMM Microwave Imager (TMI) and most recently, the Aqua AMSR‐E sensor. This lead to climatologies of hail frequency from TMI and AMSR‐E, however, limitations include geographical domain of the TMI sensor (35 S to 35 N) and the overpass time of the Aqua satellite (130 am/pm local time), both of which reduce an accurate mapping of hail events over the global domain and the full diurnal cycle. Nonetheless, these studies presented exciting, new applications for passive microwave sensors. Since 1998, NOAA and EUMETSAT have been operating the AMSU‐A/B and the MHS on several operational satellites: NOAA‐15 through NOAA‐19; MetOp‐A and ‐B. With multiple satellites in operation since 2000, the AMSU/MHS sensors provide near global coverage every 4 hours, thus, offering a much larger time and temporal sampling than TRMM or AMSR‐E. With similar observation frequencies near 30 and 85 GHz and additionally three at the 183 GHz water vapor band, the potential to detect strong convection associated with severe storms on a more comprehensive time and space scale exists. In this study, we develop a prototype AMSU‐based hail detection algorithm through the use of collocated satellite and surface hail reports over the continental U.S. for a 12‐year period (2000‐2011). Compared with the surface observations, the algorithm detects approximately 40 percent of hail occurrences. The simple threshold algorithm is then used to generate a hail climatology that is based on all available AMSU observations during 2000‐11 that is stratified in several ways, including total hail occurrence by month (March through September), total annual, and over the diurnal cycle. Independent comparisons are made compared to similar data sets derived from other satellite, ground radar and surface reports. The algorithm was also applied to global land measurements for a single year and showed close agreement with other satellite based hail climatologies. Such a product could serve as a prototype for use with a future geostationary based microwave sensor such as NASA's proposed PATH mission.

Ferraro, Ralph↗
NASA NTRS2014

Impact of Vertical Wind Shear on Tropical Cyclone Rainfall

While tropical cyclone rainfall has a large axisymmetric component, previous observational and theoretical studies have shown that environmental vertical wind shear leads to an asymmetric component of the vertical motion and precipitation fields. Composites consistently depict a precipitation enhancement downshear and also cyclonically downwind from the downshear direction. For consistence with much of the literature and with Northern Hemisphere observations, this is subsequently referred to as "Downshear-Left". Stronger shear magnitudes are associated with greater amplitude precipitation asymmetries. Recent work has reinforced the prior findings, and explored details of the response of the precipitation and kinematic fields to environmental vertical wind shear. Much of this research has focused on tropical cyclones away from land, to limit the influence of other processes that might distort the signal related to vertical wind shear. Recent evidence does suggest vertical wind shear can also play a major role in precipitation asymmetries during and after landfall.

Cecil, Dan↗
NASA NTRS2014

Hurricane Karl's Structure and Some Thoughts for 2014 Strategy

Karl has interesting data, but far below the standards we should be able to achieve in HS3. We need to fly AV-1 over hurricanes in 2014. Most of the cold cloud shield in the inner core of hurricanes should be safe for AV-1 to fly. Significant convection occupies a small region, but we sometimes unnecessarily apply the 5000-ft separation rule to the entire cold cloud shield.

Cecil, Dan↗
NASA NTRS2007

Response of Global Lightning Activity Observed by the TRMM/LIS During Warm and Cold ENSO Phases

This paper investigates the response of global lightning activity to the transition from the warm (January February March-JFM 1998) to the cold (JFM 1999) ENSO phase. The nine-year global lightning climatology for these months from the Tropical Rainfall Measuring Mission (TRMM) Lightning Imaging Sensor (LIS) provides the observational baseline. Flash rate density is computed on a 5.0x5.0 degree lat/lon grid within the LIS coverage area (between approx.37.5 N and S) for each three month period. The flash rate density anomalies from this climatology are examined for these months in 1998 and 1999. The observed lightning anomalies spatially match the documented general circulation features that accompany the warm and cold ENSO events. During the warm ENSO phase the dominant positive lightning anomalies are located mostly over the Western Hemisphere and more specifically over Gulf of Mexico, Caribbean and Northern Mid-Atlantic. We further investigate specifically the Northern Mid-Atlantic related anomaly features since these show strong relation to the North Atlantic Oscillation (NAO). Furthermore these observed anomaly patterns show strong spatial agreement with anomalous upper level (200 mb) cold core cyclonic circulations. Positive sea surface temperature anomalies during the warm ENSO phase also affect the lightning activity, but this is mostly observed near coastal environments. Over the open tropical oceans, there is climatologically less lightning and the anomalies are less pronounced. Warm ENSO related anomalies over the Eastern Hemisphere are most prominent over the South China coast. The transition to the cold ENSO phase illustrates the detected lightning anomalies to be more pronounced over East and West Pacific. A comparison of total global lightning between warm and cold ENSO phase reveals no significant difference, although prominent regional anomalies are located over mostly oceanic environments. All three tropical "chimneys" (Maritime Continent, Central Africa, and Amazon Basin) do not show any particular response to this transition.

Chronis, Themis G.↗
NASA NTRS2007

The ENSO Effect on the Temporal and Spatial Distribution of Global Lightning Activity

The recently reprocessed (1997-2006) OTD/LIS database is used to investigate the global lightning climatology in response to the ENSO cycle. A linear correlation map between lightning anomalies and ENSO (NINO3.4) identifies areas that generally follow patterns similar to precipitation anomalies. We also observed areas where significant lightning/ENSO correlations are found and are not accompanied of significant precipitation/ENSO correlations. An extreme case of the strong decoupling between lightning and precipitation is observed over the Indonesian peninsula (Sumatra) where positive lightning/NINO3.4 correlations are collocated with negative precipitation/NINO3.4 correlations. Evidence of linear relationships between the spatial extent of thunderstorm distribution and the respective NINO3.4 magnitude are presented for different regions on the Earth. Strong coupling is found over areas remote to the main ENSO axis of influence and both during warm and cold ENSO phases. Most of the resulted relationships agree with the tendencies of precipitation related to ENSO empirical maps or documented teleconnection patterns. Over the Australian continent, opposite behavior in terms of thunderstorm activity is noted for warm ENSO phases with NINO3.4 magnitudes with NINO3.4>+l.08 and 0<NqNO3.4<I.08. Finally, we investigate the spatial distribution of areas that consistently portrayed enhanced lightning activity during the main warm/cold (El Nino/La Nina) ENSO episodes of the past decade. The observed patterns show no spatial overlapping and identify areas that in their majority are in agreement with empirical precipitation/ENSO maps. The areas that appear during the warm ENSO phase are found over regions that have been identified as anomalous Hadley circulation ENSO-related patterns. The areas that appear during the cold ENSO phase are found predominantly around the west hemisphere equatorial belt and are in their majority identified by anomalous Walker circulation.

Chronis, Themis G.↗
NASA NTRS2002

The Relationship of Tropical Cyclone Convective Intensity to Passive Microwave Observations

During the 1998 and 2001 hurricane seasons, the Advanced Microwave Precipitation Radiometer (AMPR) was flown aboard the National Aeronautics and Space Administration (NASA) ER-2 high altitude aircraft as part of the Third Convection And Moisture EXperiment (CAMEX-3) and the Fourth Convection And Moisture Experiment (CAMEX-4). Several hurricanes and tropical storms were sampled during these experiments. The passive microwave observations of these tropical cyclones collected at frequencies of 10.7, 19.35, 37.1, and 85.5 GHz will be presented to explain differences in precipitation features of the hurricanes. In particular, the relationship of the passive microwave signatures of precipitation-sized ice to vertical updraft strength will be examined as a possible indicator of future convective intensity. Correlated aircraft radar, lightning, visible and infrared information will also be examined to provide further insight.

Hood, Robbie E.↗