High False Alarm Ratio Research Articles

Convective Initiation Nowcasting in South China Using Physics‐Augmented Random Forest Models and Geostationary Satellites

AbstractConvective initiation (CI) nowcasting in subtropical regions often faces challenges, such as complex physical processes and imbalanced samples of CI events, resulting in a high false alarm ratio (FAR). In this paper, we propose a Storm Warning System with Physics‐Augmentation (SWASP) based on the random forest algorithm and cloud physical conditions, using Himawari‐8 Advanced Himawari Imager data from April to September 2019 in South China. The cloud physical conditions (e.g., cloud‐top cooling rates) were investigated to establish regional thresholds for convection occurrence. Ancillary information, including elevation, satellite zenith angle, and latitude, was also incorporated into the SWASP model. Compared to conventional methods, the SWASP model exhibits an improved probability of detection by 0.11 and 0.08 and a decreased FAR by 0.38 and 0.44 for daytime and nighttime forecasts. Moreover, the SWASP model enables the detection of local convective storm systems about 30 min to 1 hr ahead of radar detection in typical convective storm cases. This study contributes to further advancements of the SWASP model by incorporating physical conditions and emphasizes the potential application of geostationary satellites in convective early warnings.

A Statistical Forecasting Model for Extremes of the Fire Behaviour Index in Australia

The increasing frequency and duration of severe fire events in Australia further necessitate accurate and timely forecasting to mitigate their consequences. This study evaluated the performance of two distinct approaches to forecasting extreme fire danger at two- to three-week lead times for the period 2003 to 2017: the official Australian climate simulation dynamical model and a statistical model based on climate drivers. We employed linear logistic regression to develop the statistical model, assessing the influence of individual climate drivers using single linear regression. The performance of both models was evaluated through case studies of three significant extreme fire events in Australia: the Canberra (2003), Black Saturday (2009), and Pinery (2015) fires. The results revealed that ACCESS-S2 generally underestimated the spatial extent of all three extreme FBI events, but with accuracy scores ranging from 0.66 to 0.86 across the case studies. Conversely, the statistical model tended to overpredict the area affected by extreme FBI, with high false alarm ratios between 0.44 and 0.66. However, the statistical model demonstrated higher probability of detection scores, ranging from 0.57 to 0.87 compared with 0.03 to 0.57 for the dynamic model. These findings highlight the complementary strengths and limitations of both forecasting approaches. Integrating dynamical and statistical models with transparent communication of their uncertainties could potentially improve accuracy and reduce false alarms. This can be achieved through hybrid forecasting, combined with visual inspection and comparison between the statistical and dynamical forecasts. Hybrid forecasting also has the potential to increase forecast lead times to up to several months, ultimately aiding in decision-making and resource allocation for fire management.

Validating IMERG data for diurnal rainfall analysis across the Indonesian maritime continent using gauge observations

The availability of surface rainfall data with high spatial-temporal resolution is needed to understand the diurnal rainfall characteristics in the Indonesian Maritime Continent (IMC) to improve the accuracy of weather and climate models in this region. This study validates the accuracy of the Final Run product of IMERG data version 06B (V06B) and version 07A (V07A), which have a resolution of 0.1°–30 min for diurnal rainfall analysis over IMC. Validation was conducted for precipitation amount (PA), precipitation frequency (PF), and precipitation intensity (PI), by recording 302 automatic rain gauges (RG) every 10 min from January 2014 to September 2021. IMERG V06B and V07A perform well in observing diurnal PA and PF but struggle in observing diurnal PI compared to RG observations. This has been determined by the correlation coefficient (CC) values of IMERG V06B (V07) data, which are 0.76 (0.72) for PA, 0.77 (0.76) for PF, and 0.21 (0.13) for PI. The IMERG data align to an extent with RG observations for the PA value, having a small relative bias (RB). The results also display a systematic PF and PI values error. The high false alarm ratio (FAR) of IMERG data suggests rain detection errors, leading to overestimated PF values. Additionally, the underestimation of PI values is due to the limitation of IMERG data in observing extreme and small-scale rain events. About 86.10% (78.14%) and 81.45% (81.78) of IMERG V06B (IMERG V07A) data show a peak time difference of <3h for PA and PF when compared with RG observations. Overall, IMERG V06B performs better than V07, possibly due to inaccurate orbits of GPROF data, removal of SAPHIR satellite observations, and inter-calibration issues with CORRA and GPCP data. Continuous monitoring and data improvements are necessary to improve the accuracy and reliability of IMERG data in detecting diurnal rainfall patterns in the IMC region.

Blocking intrusion logic using optimized multi-head convolution in wireless sensor network

Wireless sensor nodes (WSN) combine sensing and communication capabilities in the smallest sensor network component. Sensor nodes have basic networking capabilities, such as wireless connection with other nodes, data storage, and a microcontroller to do basic processing. The intrusion detection problem is well analyzed and there exist numerous techniques to solve this issue but suffer will poor intrusion detection accuracy and a higher false alarm ratio. To overcome this challenge, a novel Intrusion Detection via Salp Swarm Optimization based Deep Learning Algorithm (ID-SODA) has been proposed which classifies intrusion node and non-intrusion node. The proposed ID-SODA technique uses the k-means clustering algorithm to perform clustering. The Salp Swarm Optimization (SSO) technique takes into residual energy, distance, and cost while choosing the cluster head selection (CHS). The CHS is given the input to a multi-head convolutional neural network (MHCNN), which will classify into intrusion node and non-intrusion node. The performance analysis of the suggested ID-SODA is evaluated based on the parameters like accuracy, precision, F1 score, detection rate, recall, false alarm rate, and false negative rate. The suggested ID-SODA achieves an accuracy range of 98.95%. The result shows that the suggested ID-SODA improves the overall accuracy better than 6.56%, 2.94%, and 2.95% in SMOTE, SLGBM, and GWOSVM-IDS respectively.

An infrared small target detection method using coordinate attention and feature fusion

Infrared small target detection is a research core of infrared image processing and machine vision that is widely applied to civil and military fields. Although the accuracy and efficiency of infrared small target detection have been significantly improved in recent years, the issues of missing detection and false alarms with complex scenes remain because infrared small targets usually have small sizes and weak signals and are submerged by the backgrounds. To cope with the abovementioned problems, we propose in this paper an infrared small target detection method using coordinate attention and feature fusion, named CAFF-Net. First, we present a deep and shallow feature fusion strategy to improve the detection probability, which captures both the low-level structural and textural features and the high-level semantic information of small targets to avoid missing detection. Second, we adopt the coordinate attention mechanism to highlight the target salience and suppress background interference in the feature map, which is able to decrease the false alarm rate with complex scenes. Third, we contribute a real-world infrared small target dataset, named NCHU-SIRST. The constructed dataset consists of 273 training frames and 317 test frames and offers a comprehensive comparison for various infrared small target detection methods with different scenes. Finally, we run our method on the NCHU-SIRST dataset to conduct a comparison with some existing traditional and deep-learning-based infrared small target detection methods. The experimental results demonstrate that the presented CAFF-Net method performs competitively among the comparable methods, especially with the high detection probability and low false alarm ratio under complex scenes. (The code of the proposed CAFF-Net method and the NCHU-SIRST dataset are available at https://github.com/PCwenyue/CAFF-Net-method-and-the-NCHU-SIRST-dataset).

Evaluation of CMORPH, PERSIANN-CDR, CHIRPS V2.0, TMPA 3B42 V7, and GPM IMERG V6 Satellite Precipitation Datasets in Arabian Arid Regions

Rainfall depth is a crucial parameter in water resources and hydrological studies. Rain gauges provide the most reliable point-based rainfall estimates. However, they do not have a proper density/distribution to provide sufficient rainfall measurements in many areas, especially in arid regions. To evaluate the adequacy of satellite datasets as an alternative to the rain gauges, the Kingdom of Saudi Arabia (KSA) is selected for the current study as a representative of the arid regions. KSA occupies most of the Arabian Peninsula and is characterized by high variability in topographic and climatic conditions. Five satellite precipitation datasets (SPDSs)—CMORPH, PERSIANN-CDR, CHIRPS V2.0, TMPA 3B42 V7, and GPM IMERG V6—are evaluated versus 324 conventional rain-gauges’ daily precipitation measures. The evaluation is conducted based on nine quantitative and categorical metrics. The evaluation analysis is carried out for daily, monthly, yearly, and maximum yearly records. The daily analysis revealed a low correlation for all SPDSs (&lt;0.31), slightly improved in the yearly and maximum yearly analysis and reached its highest value (0.58) in the monthly analysis. The GPM IMERG V6 and PERSIANN-CDR have the highest probability of detection (0.55) but with a high false alarm ratio (&gt;0.8). Accordingly, in arid regions, the use of daily SPDSs in rainfall estimation will lead to high uncertainty in the obtained results. The best performance for all statistical metrics was found at 500–750 m altitudes in the central and northern parts of the study area for all satellites except minor anomalies. CMORPH dataset has the lowest centered root mean square error (RMSEc) for all analysis periods with the best results in the monthly analyses.

Large‐Sample Evaluation of Radar Rainfall Nowcasting for Flood Early Warning

AbstractTo assess the potential of radar rainfall nowcasting for early warning, nowcasts for 659 events were used to construct discharge forecasts for 12 Dutch catchments. Four open‐source nowcasting algorithms were tested: Rainymotion Sparse (RM‐S), Rainymotion DenseRotation (RM‐DR), Pysteps deterministic (PS‐D), and probabilistic (PS‐P) with 20 ensemble members. As benchmark, Eulerian Persistence (EP) and zero precipitation input (ZP) were used. For every 5‐min step in the available nowcasts, a discharge forecast with a 12‐hr forecast horizon was constructed. Simulations using the observed radar rainfall were used as reference. Rainfall and discharge forecast errors were found to increase with both increasing rainfall intensity and spatial variability. For the discharge forecasts, this relationship depends on the initial conditions, as the forecast error increases more quickly with rainfall intensity when the groundwater table is shallow. Overall, discharge forecasts using RM‐DR, PS‐D, and PS‐P outperform the other methods. Threshold exceedance forecasts were assessed by using the maximum event discharge as threshold. Compared to benchmark ZP, an exceedance is, on average, forecast 223 (EP), 196 (RM‐S), 213 (RM‐DR), 119 (PS‐D), and 143 min (PS‐P) in advance. The EP results are counterbalanced by both a high false alarm ratio (FAR) and inconsistent forecasts. Contrarily, PS‐D and PS‐P produce lower FAR and inconsistency index values than all other methods. All methods advance short‐term discharge forecasting compared to no rainfall forecasts at all, though all have shortcomings. As forecast rainfall volumes are a crucial factor in discharge forecasts, a future focus on improving this aspect in nowcasting is recommended.

The Ability of the ICE-T Microphysics Scheme in HARMONIE-AROME to Predict Aircraft Icing

AbstractIn-cloud icing is a major hazard for aviation traffic and forecasting of these events is an important task for weather agencies worldwide. A common tool utilized by aviation forecasters is an icing intensity index based on supercooled liquid water from numerical weather prediction models. We seek to validate the modified microphysics scheme, ICE-T, in the HARMONIE-AROME numerical weather prediction model with respect to aircraft icing. Icing intensities and supercooled liquid water derived from two 3-month winter season simulations with the original microphysics code, CTRL, and ICE-T are compared with pilot reports of icing and satellite retrieved values of liquid and ice water content fromCloudSat–CALIPSOand liquid water path from AMSR-2. The results show increased supercooled liquid water and higher icing indices in ICE-T. Several different thresholds and sizes of neighborhood areas for icing forecasts were tested out, and ICE-T captures more of the reported icing events for all thresholds and nearly all neighborhood areas. With a higher frequency of forecasted icing, a higher false alarm ratio cannot be ruled out, but is not possible to quantify due to the lack of no-icing observations. The increased liquid water content in ICE-T shows a better match with the retrieved satellite observations, yet the values are still greatly underestimated at lower levels. Future studies should investigate this issue further, as liquid water content also has implications for downstream processes such as the cloud radiative effect, latent heat release, and precipitation.

Capability of GPM IMERG Products for Extreme Precipitation Analysis over the Indonesian Maritime Continent

Integrated Multi-satellite Retrievals for GPM (IMERG) data have been widely used to analyze extreme precipitation, but the data have never been validated for the Indonesian Maritime Continent (IMC). This study evaluated the capability of IMERG Early (E), Late (L), and Final (F) data to observe extreme rain in the IMC using the rain gauge data within five years (2016–2020). The capability of IMERG in the observation of the extreme rain index was evaluated using Kling–Gupta efficiency (KGE) matrices. The IMERG well captured climatologic characteristics of the index of annual total precipitation (PRCPTOT), number of wet days (R85p), number of very wet days (R95p), number of rainy days (R1mm), number of heavy rain days (R10mm), number of very heavy rain days (R20mm), consecutive dry days (CDD), and max 5-day precipitation (RX5day), indicated by KGE value &gt;0.4. Moderate performance (KGE = 0–0.4) was shown in the index of the amount of very extremely wet days (R99p), the number of extremely heavy precipitation days (R50mm), max 1-day precipitation (RX1day), and Simple Daily Intensity Index (SDII). Furthermore, low performance of IMERG (KGE &lt; 0) was observed in the consecutive wet days (CWDs) index. Of the 13 extreme rain indices evaluated, IMERG underestimated and overestimated precipitation of nine and four indexes, respectively. IMERG tends to overestimate precipitation of indexes related to low rainfall intensity (e.g., R1mm). The highest overestimation was observed in the CWD index, related to the overestimation of light rainfall and the high false alarm ratio (FAR) from the daily data. For all indices of extreme rain, IMERG showed good capability to observe extreme rain variability in the IMC. Overall, IMERG-L showed a better capability than IMERG-E and -F but with an insignificant difference. Thus, the data of IMERG-E and IMERG-L, with a more rapid latency than IMERG-F, have great potential to be used for extreme rain observation and flood modeling in the IMC.

From TRMM to GPM, how do improvements of post/near-real-time satellite precipitation estimates manifest?

With the end of the Tropical Rainfall Measuring Mission (TRMM) era, its successor of Global Precipitation Measurement (GPM) aims to provide a new generation of satellite precipitation estimates (SPE) with higher spatiotemporal resolution and wider spatial coverage. As the representative precipitation retrieval algorithms for GPM and TRMM eras, it is necessary to compare the Integrated Multi-satellite Retrievals for GPM (IMERG) with the TRMM Multi-satellite Precipitation Analysis (TMPA) to investigate the improvements of SPEs from TRMM to GPM. This study analyzes the post/near-real-time SPEs from TMPA and IMERG over Mainland China from April 2014 to March 2018 at daily scale, using precipitation observations from 510 meteorological stations as reference. A very comprehensive evaluation system, which contains four indicators of continuous metrics, categorical metrics, error components, and systematic and random errors is adopted in this study. Besides, four patterns with spatial distribution, temporal pattern, precipitation intensity distribution, and overall metrics are applied to illustrate each indicator's performance. The results show that IMERG products have lower RB and RMSE and higher CC than TMPA and greatly reduce the three error components (hit bias, missed precipitation, and false precipitation). Furthermore, IMERG products present a higher capability to capture the precipitation events correctly and generally have lower systematic and random errors than TMPA products. However, there are still rooms for IMERG products to improve, such as higher false alarm ratio (FAR) and more false precipitation of IMERG products at relatively low rain rates, the 25% higher systematic error of IMERG-E (IMERG “Early Run” near-real-time SPE) compared to TMPA-RT (TMPA near-real-time SPE). In addition, we find the gauge adjustment from GPCC is not a very effective way to improve the detecting capability and reduce the systematic errors for near-real-time SPEs over Mainland China. This study provides some valuable information on the transition from TMPA to IMERG, which may advance the algorithm's development and the hydrometeorological applications.

Flow Measurement-Based Self-Adaptive Line Segment Clustering Model for Leakage Detection in Water Distribution Networks

Leakage detection of water distribution networks (WDNs) plays a key role in water management. Current leakage detection methods mostly have high requirements for the installation of sensors in WDN. So, there needs an efficient and convenient method when dealing with a new WDN. In this article, a novel, timely, and accurate leakage detection model is proposed, called self-adaptive line segment clustering (SALICT), which only asks for flow measurement data of WDN, reducing the demand for sensors. The proposed method focuses on the identification phase of leakage detection to improve the sensitivity to leakage. To achieve this, SALICT efficiently fuses a density-based clustering analysis method and a line segment clustering model. Furthermore, a series of auxiliary decision functions are introduced and designed to adaptively select parameters, obtain cluster centers, and detect leakage data according to data distribution characteristics. Notably, the proposed method can detect leakages in real-time after learning from historical data. Besides that, the proposed method has the potential to be easily used in different types of WDN. Experimental results show that the proposed method can achieve a high leakage identification ratio and low false alarm ratio and has low time consumption in various water use scenarios.

Operational Forecasting of Tropical Cyclone Rapid Intensification at the National Hurricane Center

Although some recent progress has been made in operational tropical cyclone (TC) intensity forecasting, the prediction of rapid intensification (RI) remains a challenging problem. To document RI forecast progress, deterministic and probabilistic operational intensity models used by the National Hurricane Center (NHC) are briefly reviewed. Results show that none of the deterministic models had RI utility from 1991 to about 2015 due to very low probability of detection, very high false alarm ratio, or both. Some ability to forecast RI has emerged since 2015, with dynamical models being the best guidance for the Atlantic and statistical models the best RI guidance for the eastern North Pacific. The first probabilistic RI guidance became available in 2001, with several upgrades since then leading to modest skill in recent years. A tool introduced in 2018 (DTOPS) is currently the most skillful among NHC’s probabilistic RI guidance. To measure programmatic progress in forecasting RI, the Hurricane Forecast Improvement Program has introduced a new RI metric that uses the traditional mean absolute error but restricts the sample to only those cases where RI occurred in the verifying best track or was forecast. By this metric, RI forecasts have improved by ~20–25% since the 2015–2017 baseline period.

Statistical framework for the evaluation of earthquake forecasting: A case study based on satellite surface temperature anomalies

Abstract There are growing observational evidences that various geophysical anomalies precede large earthquakes. However, the reliability of these anomalies for earthquake forecasting is controversial, and therefore more consistent assessment of forecasting ability is required. A framework for investigating pre-seismic anomaly detection using essential statistical indicators before global earthquakes is proposed. Surface temperature (ST) data from the Atmospheric Infrared Sounder (AIRS) sensor were used to realize this framework. First, seismic-related ST anomalies were identified, and then the statistical characteristics of forecasting ability for three indicators (accuracy, missed detection, and false alarm) were calculated in retrospective and prospective ways. The ST anomalies displayed some aggregation effects. Negative anomalies mainly concentrated on epicenters and to the north, while positive anomalies were found on the periphery; neither were strongly dependent on earthquake magnitude. The temporal evolution of forecasting metrics was relatively stable for the period 2010–2018. Mean accuracy, missed detection, and false alarm ratios were 6.01%, 1.60%, and 92.39%, respectively. Accuracy and missed detection ratios showed some spatial correlation and both peaked in the same area (e.g., eastern Japan); however, most areas showed very high false alarm ratios. Based on our findings, the combination of AIRS ST data and the Z-score anomaly detection algorithm to predict short-term earthquakes is currently not practical; the possibility of earthquake forecasting based on satellite thermal infrared measurements remains a huge challenge. However, our results confirmed the efficiency of this framework for quantitatively evaluating earthquake forecasting ability. This approach could be applied to various geophysical parameters and anomaly detection methods.

A non-invasive and non-wearable food intake monitoring system based on depth sensor

The food intake counting method showed a good significance that can lead to a successful weight loss by simply monitoring the food intake taken during eating. The device used in this project was Kinect Xbox One which used a depth camera to detect the motion of a person’s gesture and posture during food intake. Previous studies have shown that most of the methods used to count food intake device is worn device type. The recent trend is now going towards non-wearable devices due to the difficulty when wearing devices and it has high false alarm ratio. The proposed system gets data from the Kinect camera and monitors the gesture of the user while eating. Then, the gesture data is collected to be recognized and it will start counting the food intake taken by the user. The system recognizes the patterns of the food intake from the user by following the algorithm to analyze the gesture of the basic eating type and the system get an average accuracy of 96.2%. This system can help people who are trying to follow a proper way to avoid being overweight or having eating disorders by monitoring their meal intake and controlling their eating rate.

Comparative Evaluation of the GPM IMERG Early, Late, and Final Hourly Precipitation Products Using the CMPA Data over Sichuan Basin of China

The Global Precipitation Measurement (GPM) mission has generated global precipitation products of improved accuracy and coverage that are promising for advanced hydrological and meteorological studies. This study evaluates three Integrated Multi-satellitE Retrievals for GPM (IMERG) Hourly products, including the Early-, Late-, and Final-run products (IMERG-HE, IMERG-HL, and IMERG-HF, respectively), over Sichuan Basin of China. This highly complex terrain of the steep mountainous region offers further scrutiny on the quality and applicability of the data. The China Meteorological Precipitation Analysis (CMPA) data from January 2016 to December 2018 are used as the reference for the evaluation. Results show that: (1) At grid scale, IMERG-HL and IMERG-HF outperform IMERG-HE in terms of correlation coefficient (CC) and root-mean-square error (RMSE), but IMERG-HL has smaller relative bias (RB) than that of the IMERG-HF (by 21.16%). IMERG-HF presents the highest probability of detection (POD = 0.52) and critical success index (CSI = 0.32), except for high false alarm ratio (FAR) for light precipitation. (2) At regional scale, IMERG-HF outperforms IMERG-HE and IMERG-HL in annual evaluation in all the metrics except for the serious overestimation as shown in RB (20.18%, 3.84%, and 4.97%, respectively). Its accumulative precipitation deviation mainly comes from moderate precipitation events (1–10 mm/h), while better detection capability is seen in light precipitation (&lt;1 mm/h). Seasonally, IMERG-HF performs the best in winter, while IMERG-HL performs the best in the other seasons. (3) IMERG-HF captures the peak precipitation more accurately in all seasons. In reproducing the diurnal cycle, IMERG-HF performs better in winter, while IMERG-HL performs better in summer and autumn, and IMERG-HE in spring. However, all three products overestimate the early morning precipitation (01:00–08:00 local standard time) of the diurnal cycle in spring, summer, and autumn.

Performance of the State-Of-The-Art Gridded Precipitation Products over Mountainous Terrain: A Regional Study over Austria

During the last decade, satellite-based precipitation products have been believed to be a potential source for forcing inputs in hydro-meteorological and agricultural models, which are essential especially over the mountains area or in basins where ground gauges are generally sparse or nonexistent. This study comprehensively evaluates several newly released precipitation products, i.e., MSWEP-V2.2, IMERG-V05B, IMERG-V06A, IMERG-V05-RT, ERA5, and SM2RAIN-ASCAT, at daily and monthly time-scales over Austria. We show that all the examined products are able to reproduce the spatial precipitation distribution over the country. MSWEP, followed by IMERG-V05B and -V06A, show the strongest agreement with in situ observations and perform better than other products with respect to spatial patterns and statistical metrics. Both IMERG and ERA5 products seem to have systematic precipitation overestimation at the monthly time-scale. IMERG-V06A performs slightly better than IMERG-V05B. With respect to heavy precipitation (P &gt; 10 mm/day), MSWEP compare to other products demonstrate advantages in detecting precipitation events with a higher spatial average of probability of detection (POD) and lower false alarm ratio (FAR) scores skill (0.74 and 0.28), while SM2RAIN and ERA5 reveal lower POD (0.35) and higher FAR (0.56) in this precipitation range in comparison with other products. However, the ERA5 and MSWEP indicate robust average POD and FAR values with respect to light/moderate precipitation (10 mm &gt; P ≥ 0.1 mm) with 0.94 and 0.11, respectively. Such robustness of MSWEP may be rooted in applying the daily rain gauges in calibration processes. Moreover, although all products accurately map the spatial precipitation distribution they still have difficulty capturing the effects of topography on precipitation. The overall performance of the precipitation products was lower in the peripheries of the study area where most stations are situated in the mountainous area and was higher over the low altitude regions. However, according to our analysis of the considered products, MSWEP-V2.2, followed by IMERG-V06S and -V05B, are the most suitable for driving hydro-meteorological, agricultural, and other models over mountainous terrain.

Food intake gesture monitoring system based-on depth sensor

Food intake gesture technology is one of a new strategy for obesity people managing their health care while saving their time and money. This approach involves combining face and hand joint point for monitoring food intake of a user using Kinect Xbox One camera sensor. Rather than counting calories, scientists at Brigham Young University found dieters who eager to reduce their number of daily bites by 20 to 30 percent lost around two kilograms a month, regardless of what they ate [1]. Research studies showed that most of the methods used to count bite are worn type devices which has high false alarm ratio. Today trend is going toward the non-wearable device. This sensor is used to capture skeletal data of user while eating and train the data to capture the motion and movement while eating. There are specific joint to be capture such as Jaw face point and wrist roll joint. Overall accuracy is around 94%. Basically, this increase in the overall recognition rate of this system.

Evaluating the hydrological utility of latest IMERG products over the Upper Huaihe River Basin, China

With increasing spatiotemporal resolution and accuracy, satellite-based precipitation estimations (SPEs) have become critical data sources for hydrological applications. Their global scale coverage and short latency perfectly fit the requirements of flood forecasting. However, the comprehensive evaluation of the latest generation of SPEs is an essential step before the confident use of these indirect precipitation estimators. Therefore, the latest releases of the Integrated MultisatellitE Retrievals for Global Precipitation Measurement (IMERG) Version 5 products, including the near-real-time “Early” run and “Later” run products (IMERG-E and IMERG-L, respectively), and the post-real-time “Final” run IMERG product (IMERG-F), are employed in this paper. Through the hydrometeorological assessment of these three IMERG products against dense gauge observations during the period from April 2014 to December 2015 over the Upper Huaihe River Basin (UHRB), the usability of IMERG products in hydrological application is explored via the Variable Infiltration Capacity model. The results indicate that, irrespective of a slight positive bias, the post-real-time IMERG-F product presents the best performance among the three IMERG products at both the grid scale and basin scale. Although all three IMERG products show remarkably high probability of detection (POD, approximately 0.82), the IMERG-E presents a relatively higher false alarm ratio (FAR = 0.37) in comparison with IMERG-L (FAR = 0.32) and IMERG-F (FAR = 0.31), which is primarily due to the significant overestimation in light rain events. In terms of hydrological simulations of streamflow, the dense gauge observations exhibit very good performance for both entire simulation period and flood events, as expected. IMERG-F demonstrates acceptable performance in entire simulation, while both IMERG-E and IMERG-L demonstrate poor performance. Given the clear overestimation in short-terms flooding and the notable underestimation in long-terms flooding, all three IMERG products should be used with caution in flood simulation over UHRB. Overall, this is a systematic assessment of IMERG products. We hope this study can provide useful guidelines for potential hydrologists of IMERG products and further IMERG algorithm development. Also, given the scope of on-going efforts to improve the IMERG products, we believe the subsequent versions of IMERG can be widely used in global or regional hydrological application, particularly in the field of flood forecasting.

R-CNN-Based Ship Detection from High Resolution Remote Sensing Imagery

Offshore and inland river ship detection has been studied on both synthetic aperture radar (SAR) and optical remote sensing imagery. However, the classic ship detection methods based on SAR images can cause a high false alarm ratio and be influenced by the sea surface model, especially on inland rivers and in offshore areas. The classic detection methods based on optical images do not perform well on small and gathering ships. This paper adopts the idea of deep networks and presents a fast regional-based convolutional neural network (R-CNN) method to detect ships from high-resolution remote sensing imagery. First, we choose GaoFen-2 optical remote sensing images with a resolution of 1 m and preprocess the images with a support vector machine (SVM) to divide the large detection area into small regions of interest (ROI) that may contain ships. Then, we apply ship detection algorithms based on a region-based convolutional neural network (R-CNN) on ROI images. To improve the detection result of small and gathering ships, we adopt an effective target detection framework, Faster-RCNN, and improve the structure of its original convolutional neural network (CNN), VGG16, by using multiresolution convolutional features and performing ROI pooling on a larger feature map in a region proposal network (RPN). Finally, we compare the most effective classic ship detection method, the deformable part model (DPM), another two widely used target detection frameworks, the single shot multibox detector (SSD) and YOLOv2, the original VGG16-based Faster-RCNN, and our improved Faster-RCNN. Experimental results show that our improved Faster-RCNN method achieves a higher recall and accuracy for small ships and gathering ships. Therefore, it provides a very effective method for offshore and inland river ship detection based on high-resolution remote sensing imagery.

Efficacy of the Localized Aviation MOS Program in Ceiling Flight Category Forecasts

(1) Background: Flying in instrument meteorological conditions (IMC) carries an elevated risk of fatal outcome for general aviation (GA) pilots. For the typical GA flight, aerodrome-specific forecasts (Terminal Aerodrome Forecast (TAF), Localized Aviation Model Output Statistics Program (LAMP)) assist the airman in pre-determining whether a flight can be safely undertaken. While LAMP forecasts are more prevalent at GA-frequented aerodromes, the Federal Aviation Administration (FAA) recommends that this tool be used as supplementary to the TAF only. Herein, the predictive accuracy of LAMP for ceiling flight categories of visual flight rules (VFR) and instrument flight rules (IFR) was determined. (2) Methods: LAMP accuracy was evaluated for the period of July–December 2018 using aviation-specific probability of detection (PODA), false alarm ratio (FARA) and critical success scores (CSSA). Statistical differences were determined using Chi-Square tests. (3) Results: LAMP forecasts (n = 823) across 39 states were accrued. LAMP PODA for VFR (0.67) and IFR (0.78) exceeded (p &lt; 0.031) the corresponding TAF scores (0.57 and 0.56). For VFR, the LAMP showed a non-significant (p = 0.243) higher FARA (0.25) than the TAF (0.19). For IFR forecasts, the LAMP FARA was lower (p &lt; 0.001) (0.48 and 0.81, respectively). LAMP CSSA scores exceeded the TAF for VFR (p = 0.012) and IFR forecasts (p &lt; 0.001). (4) Conclusion: These findings support the greater integration of LAMP into pre-flight weather briefings.