Success Index Research Articles

Can coaching bridge the gap for incoming Latinx graduate students?

ABSTRACT Attrition rates in the first two years of graduate school are high and ∼50% higher for underrepresented students. Here we evaluate an online group coaching intervention using a controlled and an experimental group to determine its impact on skills that are likely predictors of student success, namely the Hallmark of Success indices developed by the National Institutes of Health (NIH). Our intervention includes eight online coaching sessions (four prior to the first year of graduate school and four during the first year of graduate school). Coaching group sessions, led by certified coaches, address Resilience, Risk and Protective Factors, Accountability, Leadership, Teamwork, Professional Networking, and Self-Assessment. We evaluated the change in Hallmark of Success indices with a pre- and post-questionnaire of 44 items used to assess students’ self-evaluation in nine areas. Students in the experimental coaching group showed a significant increase in the Hallmark of Success indices in seven out of nine areas of student performance between the pre- and post-questionnaires compared to the null model of no changes between time periods. In addition, in all of the nine indices of success, we noted significant differences between the pre- and post-surveys for the experimental group. Our experimental design shows that our cost-effective coaching model improves student performance as perceived by the students.

Evaluation of Daily and Hourly Performance of Multi-Source Satellite Precipitation Products in China’s Nine Water Resource Regions

Satellite precipitation products (SPPs) are of great significance for water resource management and utilization in China; however, they suffer from considerable uncertainty. While numerous researchers have evaluated the accuracy of various SPPs, further investigation is needed to assess their performance across China’s nine major water resource regions. This study used the latest precipitation dataset of the China Meteorological Administration’s Land Surface Data Assimilation System (CLDAS-V2.0) as the benchmark and evaluated the performance of six SPPs—GSMaP, PERSIANN, CMORPH, CHIRPS, GPM IMERG, and TRMM—using six indices: correlation coefficient (CC), root mean square error (RMSE), mean absolute error (MAE), probability of detection (POD), false alarm rate (FAR), and critical success index (CSI), at both daily and hourly scales across China’s nine water resource regions. The conclusions of this study are as follows: (1) The performance of the six SPPs was generally weaker in the west than in the east, with the Continental Basin (CB) exhibiting the poorest performance, followed by the Southwest Basin (SB). (2) At the hourly scale, the performance of the six SPPs was weaker compared to the daily scale, particularly in the high-altitude CB and the high-latitude Songhua and Liaohe River Basin (SLRB), where observing light precipitation and snowfall presents significant challenges. (3) GSMaP, CMORPH, and GPM IMERG demonstrated superior overall performance compared to CHIRPS, PERISANN, and TRMM. (4) CMORPH was found to be better suited for application in drought-prone areas, showcasing optimal performance in the CB and SB. GSMaP excelled in humid regions, displaying the best overall performance in the remaining seven basins. GPM IMERG serves as a complementary precipitation data source for the first two.

A Google Earth Engine Platform to Integrate Multi-Satellite and Citizen Science Data for the Monitoring of River Ice Dynamics

This study introduces a new automated system that blends multi-satellite information and citizen science data for reliable and timely observations of lake and river ice in under-observed northern regions. The system leverages the Google Earth Engine resources to facilitate the analysis and visualization of ice conditions. The adopted approach utilizes a combination of moderate and high-resolution optical data, along with radar observations. The results demonstrate the system’s capability to accurately detect and monitor river ice, particularly during key periods, such as the freeze-up and the breakup. The integration citizen science data showed added values in the validation of remote sensing products, as well as filling gaps whenever satellite observations cannot be collected due to cloud obstruction. Moreover, it was shown that citizen science data can be converted to valuable quantitative information, such as the case of ice thickness, which is very useful when combined with ice extent derived from remote sensing. In this study, citizen science data were employed for the quantitative assessment of the remote sensing product. Obtained results showed a good agreement between the product and observed river status, with a Critical Success Index of 0.82. Notably, the system has shown effectiveness in capturing the spatial and temporal evolution of snow and ice conditions, as evidenced by its application in analyzing specific ice jam events in 2023. The study concludes that the developed system marks a significant advancement in river ice monitoring, combining technological innovation with community engagement.

Dense Time Series Generation of Surface Water Extents through Optical–SAR Sensor Fusion and Gap Filling

Surface water is a vital component of the Earth’s water cycle and characterizing its dynamics is essential for understanding and managing our water resources. Satellite-based remote sensing has been used to monitor surface water dynamics, but cloud cover can obscure surface observations, particularly during flood events, hindering water identification. The fusion of optical and synthetic aperture radar (SAR) data leverages the advantages of both sensors to provide accurate surface water maps while increasing the temporal density of unobstructed observations for monitoring surface water spatial dynamics. This paper presents a method for generating dense time series of surface water observations using optical–SAR sensor fusion and gap filling. We applied this method to data from the Copernicus Sentinel-1 and Landsat 8 satellite data from 2019 over six regions spanning different ecological and climatological conditions. We validated the resulting surface water maps using an independent, hand-labeled dataset and found an overall accuracy of 0.9025, with an accuracy range of 0.8656–0.9212 between the different regions. The validation showed an overall false alarm ratio (FAR) of 0.0631, a probability of detection (POD) of 0.8394, and a critical success index (CSI) of 0.8073, indicating that the method generally performs well at identifying water areas. However, it slightly underpredicts water areas with more false negatives. We found that fusing optical and SAR data for surface water mapping increased, on average, the number of observations for the regions and months validated in 2019 from 11.46 for optical and 55.35 for SAR to 64.90 using both, a 466% and 17% increase, respectively. The results show that the method can effectively fill in gaps in optical data caused by cloud cover and produce a dense time series of surface water maps. The method has the potential to improve the monitoring of surface water dynamics and support sustainable water management.

Implementasi Kurikulum Program Pembinaan Startup di Balikpapan

Indonesia has the opportunity to create an entrepreneurial nation by utilizing digital technology which will make us hosts in our own country. This is supported by the large potential of the digital industry in Indonesia. Managing adequate financial reports is the key to startup success in achieving access to capital, which unfortunately is a problem for most startups in the city of Balikpapan. This community service program was carried out for startup representatives in the city of Balikpapan with direct training in creating financial dashboards and digital marketing. The success index of this activity can be assessed by the percentage increase in participants' desire to be able to implement a financial dashboard without assistance by 72% and implement digital marketing without assistance by 55.6%.

Development of a Full-Scale Connected U-Net for Reflectivity Inpainting in Spaceborne Radar Blind Zones

Abstract CloudSat’s Cloud Profiling Radar is a valuable tool for remotely monitoring high-latitude snowfall, but its ability to observe hydrometeor activity near the Earth’s surface is limited by a radar blind zone caused by ground clutter contamination. This study presents the development of a deeply supervised U-Net-style convolutional neural network to predict cold season reflectivity profiles within the blind zone at two Arctic locations. The network learns to predict the presence and intensity of near-surface hydrometeors by coupling latent features encoded in blind zone-aloft clouds with additional context from collocated atmospheric state variables (i.e., temperature, specific humidity, and wind speed). Results show that the U-Net predictions outperform traditional linear extrapolation methods, with low mean absolute error, a 38% higher Sørensen–Dice coefficient, and vertical reflectivity distributions 60% closer to observed values. The U-Net is also able to detect the presence of near-surface cloud with a critical success index (CSI) of 72% and cases of shallow cumuliform snowfall and virga with 18% higher CSI values compared to linear methods. An explainability analysis shows that reflectivity information throughout the scene, especially at cloud edges and at the 1.2-km blind zone threshold, along with atmospheric state variables near the tropopause, are the most significant contributors to model skill. This surface-trained generative inpainting technique has the potential to enhance current and future remote sensing precipitation missions by providing a better understanding of the nonlinear relationship between blind zone reflectivity values and the surrounding atmospheric state. Significance Statement Snowfall is a critical contributor to the global water–energy budget, with important connections to water resource management, flood mitigation, and ecosystem sustainability. However, traditional spaceborne remote monitoring of snowfall faces challenges due to a near-surface radar blind zone, which masks a portion of the atmosphere. In this study, a deep learning model was developed to fill in missing data across these regions using surface radar and atmospheric state variables. The model accurately predicts reflectivity, with significant improvements over conventional methods. This innovative approach enhances our understanding of reflectivity patterns and atmospheric interactions, bolstering advances in remote snowfall prediction.

Podcasting for Public Knowledge: A Multiple Case Study of Scholarly Podcasts at One University

Through a lens for engaged scholarship (Boyer in Journal of Public Service and Outreach, 1(1), 11–20, 1996) this multiple case study (Merriam, 1996) explores the potential of scholarly podcasts for public knowledge dissemination, highlighting the misalignment of university impact metrics with this medium. Our team collected qualitative and numerical data from six podcasters across our university system. We identify metrics for assessing scholarly podcast value, offer recommendations for institutional communication, and share our insights and challenges. Data analysis suggests that a Listen Score (Listen Notes, ND) and an increasing Podcast Success Index (Singh et al. JMIR Medical Education, 2(2), 1–10, 2016) may be consistent with a wider reach. Consistent production and promotion are key and infrastructure support for scholarly podcasters is necessary.

A tuning-free moderate-scale burned area detection algorithm — a case study in chornobyl-contaminated region

ABSTRACT The predefined threshold is one of the main fundamental and methodological challenges underlying the uncertainty and delayed release of existing burned area (BA) products. To improve the accuracy and timeliness of BA mapping, a tuning-free moderate-scale BA detection algorithm (AT&RF algorithm) was proposed to carry out rapid BA mapping for the immediate post-fire assessment. A case study was carried out at the Chornobyl Exclusion Zone (ChEZ) to assess the performance of the algorithm. The evaluation results indicated that the algorithm successfully detected the BA in 2015 and 2020 in ChEZ, which outperforms the existing BA products by significantly decreasing the omission error by 18.6% and 15.1% and improving the Kappa coefficient and critical success index by 9.4% and 4.0% on average for MCD64A1 and FireCCI51 products, respectively. It is also possible to transfer the algorithm quickly to other regions without complicated adjustments and significantly improve the timeliness and accuracy compared to the existing global BA products for assessing environmental impact and scope.

Threshold-based flood early warning in an urbanizing catchment through multi-source data integration: Satellite and citizen science contribution

An effective flood early warning system is vital to take action to save lives and protect properties in urban areas which are increasingly prone to flooding. Despite substantial progress in flood early warning systems, limited available and accessible data often impede their advancement and reliability. Engaging communities affected by flooding can help address data and information gaps in flood early warning systems, facilitated by appropriate methods. This study developed and evaluated a flood threshold combination method to support a community-based flood early warning system in the Akaki catchment, home to Addis Ababa, the capital city of Ethiopia. Various flood threshold combinations were formulated, calibrated and validated by integrating multiple sources of data: rainfall, antecedent precipitation index estimates, Sentinel-1 Synthetic Aperture Radar satellite time series of flood extent, long-term simulated streamflow, citizen science data, river water level and three days lead-time numerical weather prediction rainfall forecast. During validation, the rainfall and river water level threshold combination outperformed other threshold combinations with probability of detection, false alarm ratio, and critical success index estimates of 0.74, 0.18 and 0.63, respectively. The flood threshold combination showed high detection performance for most flooding conditions. Flood forecasts with a 1-day lead-time exhibited a high likelihood in detecting historical severe flood events. The study provides a tested methodology for selecting suitable flood threshold-combinations, enhance the engagement of citizen scientists in a community–based flood early warning system in urban communities.

Effect of Cyclotorsion Compensation in Small Incision Lenticule Extraction Surgery for the Correction of Myopic Astigmatism: A Systematic Review and Meta-Analysis.

Small incision lenticule extraction (SMILE) has made notable advancements in addressing myopic astigmatism. Nevertheless, the potential impact of cyclotorsion on surgical outcomes cannot be overlooked. This study aims to assess the effectiveness of cyclotorsion compensation technology in SMILE surgery for the correction of myopic astigmatism, examining its influence on postoperative visual quality. A systematic review and meta-analysis were conducted. A comprehensive literature search was performed using databases, including PubMed, Web of Science, EMBASE, Cochrane Library, EBSCO, Scopus, CNKI, VIP, and Wan Fang. Studies meeting the criteria were selected and included. Data were independently extracted by three authors. Clinical outcome parameters were analyzed using Review Manager version 5.3. This meta-analysis included ten studies. The results showed that, compared with the control group (cyclotorsion compensation was not performed in SMILE), the following indicators in the cyclotorsion compensation group were: residual astigmatism (RA) [weighted mean difference (MD) = 0.73, 95% confidence interval (CI) + 0.26 to + 1.19, P = 0.002], spherical equivalent (SE) (MD = 1.99, 95% CI + 0.77 to + 3.21, P = 0.001), coma (MD = -0.06, 95% CI -0.08 to -0.04, P < 0.00001), higher-order aberrations (HOAs) (MD = -0.04, 95% CI -0.06 to -0.02, P < 0.0001), follow-up 6-month angle of error (AE) (MD = -2.67, 95% CI -3.71 to -1.63, P < 0.00001), and follow-up 6-month uncorrected distance visual acuity (UDVA) (MD = -0.05, 95% CI -0.08 to -0.01, P = 0.005), and the differences in results were statistically significant. However, the differences among correction index, index of success (IOS), targeted induced astigmatism (TIA), magnitude of error (ME), and spherical aberration (SA) were not statistically significant. Cyclotorsion compensation proves to be effective and predictable for correcting myopic astigmatism. The cyclotorsion compensation group demonstrated advantages over the control group in terms of postoperative residual astigmatism, and it induced fewer coma aberrations. Whether cyclotorsion compensation can lead to better visual quality remains to be seen, and further research on correcting myopic astigmatism through cyclotorsion compensation is warranted.

Evaluation of Retrospective National Water Model Soil Moisture and Streamflow for Drought‐Monitoring Applications

AbstractThe National Oceanic and Atmospheric Administration (NOAA)’s National Water Model (NWM) provides analyses and predictions of hydrologic variables relevant to drought monitoring and forecasts at fine time and space scales (hourly, 0.25–1 km). We present results exploring the potential for NWM soil moisture and streamflow analyses to inform operational drought monitoring. Both agricultural and hydrologic drought monitoring rely either explicitly or implicitly on an accurate representation of anomalous soil moisture values. Much of our analysis focuses on comparisons of soil moisture anomalies in the NWM to those from in‐situ observations. To establish benchmarks for NWM soil moisture skill, we also include other gridded data sets currently used to inform the US Drought Monitor, specifically those from the North American Land Data Assimilation System phase 2 (NLDAS‐2) land surface models. We then compare NWM streamflow low flows with ∼500 stream gauges from the United States Geological Survey (USGS) Hydro‐Climatic Data Network of undisturbed basins. The NWM soil moisture simulation’s skill parallels that from NLDAS‐2. The accuracy of drought condition identification from NWM streamflow exceeds that based on soil moisture as determined by Critical Success Index scores for extreme dry percentiles. Different meteorological forcings are used in the operational NWM cycles than those used in this retrospective analysis. This forcing disconnect, together with concerns about current‐generation land surface model soil moisture‐transport schemes, inhibit its current operational use for drought monitoring.

Investigating the Human Reliability in the Healthcare Sector Using the Fuzzy Analytic Network Process and the Success Likelihood Index Method.

We aimed to identify the factors contributing to human error in hospital emergency departments using scientific methods. We used the Fuzzy Analytical Network Process (FANP) and Success Likelihood Index Method (SLIM) to investigate human reliability in 54 hospital emergency departments in 15 provinces of Iran from 2021 to 2022. The study classified 17 general factors affecting human errors in hospital emergency departments. Organizational (0.349), occupational (0.330), and personal factors (0.320) had the most significant impact on human error. Based on a matrix of paired comparisons for nine emergency tasks using the probability of success index method, "checking test results and diagnosis" had the highest probability of error when referring patients to intensive care or discharge. Although the study prioritized patients, there was still a cumulative probability of human error before disease diagnosis at 0.01332, highlighting the need for further training to minimize these risks. The FANP and SLIM were effective in identifying the factors contributing to human error in hospital emergency departments. Doctors and nurses working in these departments require more knowledge, experience, and responsibility to avoid errors. By identifying factors influencing the occurrence of human error and finding solutions to reduce risks, hospitals can improve the quality of their care and prevent errors.

A D-vine copula-based quantile regression towards merging satellite precipitation products over rugged topography: a case study in the upper Tekeze–Atbara Basin

Abstract. Precipitation is a vital key element in various studies of hydrology, flood prediction, drought monitoring, and water resource management. The main challenge in conducting studies over remote regions with rugged topography is that weather stations are usually scarce and unevenly distributed. However, open-source satellite-based precipitation products (SPPs) with a suitable resolution provide alternative options in these data-scarce regions, which are typically associated with high uncertainty. To reduce the uncertainty of individual satellite products, we have proposed a D-vine copula-based quantile regression (DVQR) model to merge multiple SPPs with rain gauges (RGs). The DVQR model was employed during the 2001–2017 summer monsoon seasons and compared with two other quantile regression methods based on the multivariate linear (MLQR) and the Bayesian model averaging (BMAQ) techniques, respectively, and with two traditional merging methods – the simple modeling average (SMA) and the one-outlier-removed average (OORA) – using descriptive and categorical statistics. Four SPPs have been considered in this study, namely, Tropical Applications of Meteorology using SATellite (TAMSAT v3.1), the Climate Prediction Center MORPHing Product Climate Data Record (CMORPH-CDR), Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG v06), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN-CDR). The bilinear (BIL) interpolation technique was applied to downscale SPPs from a coarse to a fine spatial resolution (1 km). The rugged-topography region of the upper Tekeze–Atbara Basin (UTAB) in Ethiopia was selected as the study area. The results indicate that the precipitation data estimates with the DVQR, MLQR, and BMAQ models and with traditional merging methods outperform the downscaled SPPs. Monthly evaluations reveal that all products perform better in July and September than in June and August due to precipitation variability. The DVQR, MLQR, and BMAQ models exhibit higher accuracy than the traditional merging methods over the UTAB. The DVQR model substantially improved all of the statistical metrics (CC = 0.80, NSE = 0.615, KGE = 0.785, MAE = 1.97 mm d−1, RMSE = 2.86 mm d−1, and PBIAS = 0.96 %) considered compared with the BMAQ and MLQR models. However, the DVQR model did not outperform the BMAQ and MLQR models with respect to the probability of detection (POD) and false-alarm ratio (FAR), although it had the best frequency bias index (FBI) and critical success index (CSI) among all of the employed models. Overall, the newly proposed merging approach improves the quality of SPPs and demonstrates the value of the proposed DVQR model in merging multiple SPPs over regions with rugged topography such as the UTAB.

Improving Precipitation Forecasting through Early Fusion and Spatiotemporal Prediction: A Case Study Using the MultiPred Model

Due to the complexity and uncertainty of meteorological systems, traditional precipitation forecasting methods have certain limitations. Therefore, based on the common characteristics of meteorological data, a precipitation forecasting model named MultiPred is proposed, with the goal of continuously predicting precipitation for 4 h in a specific region. This model combines the multimodal fusion method with recursive spatiotemporal prediction models. The training and testing process of the model roughly involves using spatial feature extraction networks and temporal feature extraction networks to generate preliminary predictions for multimodal data. Subsequently, a modal fusion layer is employed to further extract and fuse the spatial features of the preliminary predictions from the previous step, outputting the predicted precipitation values for the target area. Experimental tests and training were conducted using ERA5 multi-meteorological modal data and GPM satellite precipitation data from 2017 to 2020, covering longitudes from 110° to 122° and latitudes from 20° to 32°. The training set used data from the first three years, while the validation set and test set each comprised 50% of the data from the fourth year. The initial learning rate for the experiment was set to 1 × 10−4, and training was performed for 1000 epochs. Additionally, the training process utilized a loss function composed of Mean Absolute Error (MAE), Mean Squared Error (MSE), and Structural Similarity Index (SSIM). The model was evaluated using the Critical Success Index (CSI), Probability of Detection (POD), and the Heidke Skill Score (HSS). Experimental results demonstrate that MultiPred excels in precipitation forecasting, particularly for light precipitation events with amounts greater than or equal to 0.1 mm and less than 2 mm. It achieves optimal performance in both light and heavy precipitation forecasting tasks.

The influence of Urmia Lake desiccation on an extreme snowfall event: A case study using the WRF-Lake model

Urmia Lake in the northwest Iran has undergone significant desiccation in recent decades, shrinking to a much smaller size than its original. This study implements the WRF-Lake coupled modeling system to examine the impacts of the observed physical changes over the Lake during a heavy snowfall event occurred February 1–2, 2017. Modifications are introduced to represent the shrunken, salt-encrusted lake surface and surrounding arid environment through updating land use, soil characteristics, lake size and depth, and albedo data. Experiments incorporating lake-atmosphere coupling and land surface refinements (Lake_CTL, Lake_LULC, Lake_AL) outperform simulations without a lake (WRF_noLake) in reproducing observed precipitation patterns at stations surrounding the Urmia Lake. All simulations underestimate the extreme 290 mm snowfall recorded at the Urmia station; however, the Lake_LULC and Lake_AL improve snowfall representation at other stations around the lake. Regarding snow cover validation, Lake_LULC and Lake_AL exhibit a higher correlation with satellite observations and show a significant improvement in snow detection near Urmia Lake. They have a Probability of Detection (POD) of 0.94 and a Critical Success Index (CSI) of 0.91, compared to a POD of 0.79–0.80 and a CSIs of ∼0.77 for the default lake model (Lake_CTL) and the WRF_noLake experiments, respectively. Additionally, adjusting the initial lake surface temperature to match observations substantially reduces cold biases in near-surface air temperatures over Urmia Lake. The lack of lake temperature updating in WRF-Lake poses ongoing challenges in the model, though. In summary, this study demonstrates that refining the representations of desiccated lakes and their surroundings in high-resolution coupled models would improve simulations of meteorological processes influenced by lake-atmosphere interactions.

On the Dependence of the Critical Success Index (CSI) on Prevalence.

The critical success index (CSI) is an established metric used in meteorology to verify the accuracy of weather forecasts. It is defined as the ratio of hits to the sum of hits, false alarms, and misses. Translationally, CSI has gained popularity as a unitary outcome measure in various clinical situations where large numbers of true negatives may influence the interpretation of other, more traditional, outcome measures, such as specificity (Spec) and negative predictive value (NPV), or when unified interpretation of positive predictive value (PPV) and sensitivity (Sens) is needed. The derivation of CSI from measures including PPV has prompted questions as to whether and how CSI values may vary with disease prevalence (P), just as PPV estimates are dependent on P, and hence whether CSI values are generalizable between studies with differing prevalences. As no detailed study of the relation of CSI to prevalence has been undertaken hitherto, the dataset of a previously published test accuracy study of a cognitive screening instrument was interrogated to address this question. Three different methods were used to examine the change in CSI across a range of prevalences, using both the Bayes formula and equations directly relating CSI to Sens, PPV, P, and the test threshold (Q). These approaches showed that, as expected, CSI does vary with prevalence, but the dependence differs according to the method of calculation that is adopted. Bayesian rescaling of both Sens and PPV generates a concave curve, suggesting that CSI will be maximal at a particular prevalence, which may vary according to the particular dataset.

Validating GOES Radar Estimation via Machine Learning to Inform NWP (GREMLIN) Product over CONUS

Abstract Geostationary Operational Environmental Satellites (GOES) Radar Estimation via Machine Learning to Inform NWP (GREMLIN) is a machine learning model that outputs composite reflectivity using GOES-R Series Advanced Baseline Imager (ABI) and Geostationary Lightning Mapper (GLM) input data. GREMLIN is useful for observing severe weather and initializing convection for short-term forecasts, especially over regions without ground-based radars. This study expands the evaluation of GREMLIN’s accuracy against the Multi-Radar Multi-Sensor (MRMS) System to the entire contiguous United States (CONUS) for the entire annual cycle. Regional and temporal variation of validation metrics are examined over CONUS by season, day of year, and time of day. Since GREMLIN was trained with data in spring and summer, root-mean-square difference (RMSD) and bias are lowest in the order of summer, spring, fall, and winter. In summer, diurnal patterns of RMSD follow those of precipitation occurrence. Winter has the highest RMSD because of cold surfaces mistaken as precipitating clouds, but some of these errors can be removed by applying the ABI clear-sky mask product and correcting biases using a lookup table. In GREMLIN, strong echoes are closely related to the existence of lightning and corresponding low brightness temperatures, which result in different error distributions over different regions of CONUS. This leads to negative biases in cold seasons over Washington State, lower 30-dBZ critical success index caused by high misses over the Northeast, and higher false alarms over Florida that are due to higher frequency of lightning.

Improved Diagnosis of Precipitation Type with LightGBM Machine Learning

Abstract Existing precipitation-type algorithms have difficulty discerning the occurrence of freezing rain and ice pellets. These inherent biases are not only problematic in operational forecasting but also complicate the development of model-based precipitation-type climatologies. To address these issues, this paper introduces a novel light gradient-boosting machine (LightGBM)-based machine learning precipitation-type algorithm that utilizes reanalysis and surface observations. By comparing it with the Bourgouin precipitation-type algorithm as a baseline, we demonstrate that our algorithm improves the critical success index (CSI) for all examined precipitation types. Moreover, when compared with the precipitation-type diagnosis in reanalysis, our algorithm exhibits increased F1 scores for snow, freezing rain, and ice pellets. Subsequently, we utilize the algorithm to compute a freezing-rain climatology over the eastern United States. The resulting climatology pattern aligns well with observations; however, a significant mean bias is observed. We interpret this bias to be influenced by both the algorithm itself and assumptions regarding precipitation processes, which include biases associated with freezing drizzle, precipitation occurrence, and regional synoptic weather patterns. To mitigate the overall bias, we propose increasing the precipitation cutoff from 0.04 to 0.25 mm h−1, as it better reflects the precision of precipitation observations. This adjustment yields a substantial reduction in the overall bias. Finally, given the strong performance of LightGBM in predicting mixed precipitation episodes, we anticipate that the algorithm can be effectively utilized in operational settings and for diagnosing precipitation types in climate model outputs. Significance Statement Freezing rain can have significant impacts on transportation and infrastructure, making accurate prediction of precipitation types crucial. In this study, we use a machine learning method known as LightGBM to predict precipitation types. We show that the new algorithm performs better than the existing methods for all precipitation types examined. Additionally, we compute a freezing-rain climatology over the eastern United States. Although the resulting climatology pattern corresponds well to observations, the algorithm overpredicts freezing-rain occurrence. We argue that this bias can be substantially reduced by increasing the precipitation cutoff from 0.04 to 0.25 mm h−1. Overall, this work highlights the potential of the LightGBM algorithm for both weather forecasting and diagnosing precipitation types in climate models.

Accounting for uncertainties in forecasting tropical-cyclone-induced compound flooding

Abstract. Tropical-cyclone impacts can have devastating effects on the population, infrastructure, and natural habitats. However, predicting these impacts is difficult due to the inherent uncertainties in the storm track and intensity. In addition, due to computational constraints, both the relevant ocean physics and the uncertainties in meteorological forcing are only partly accounted for. This paper presents a new method, called the Tropical Cyclone Forecasting Framework (TC-FF), to probabilistically forecast compound flooding induced by tropical cyclones, considering uncertainties in track, forward speed, and wind speed and/or intensity. The open-source method accounts for all major relevant physical drivers, including tide, surge, and rainfall, and considers TC uncertainties through Gaussian error distributions and autoregressive techniques. The tool creates temporally and spatially varying wind fields to force a computationally efficient compound-flood model, allowing for the computation of probabilistic wind and flood hazard maps for any oceanic basin in the world as it does not require detailed information on the distribution of historical errors. A comparison of TC-FF and JTWC operational ensembles, both based on DeMaria et al. (2009), revealed minor differences of &lt;10 %, suggesting that TC-FF can be employed as an alternative, for example, in data-scarce environments. The method was applied to Cyclone Idai in Mozambique. The underlying physical model showed reliable skill in terms of tidal propagation, reproducing the storm surge generation during landfall and flooding near the city of Beira (success index of 0.59). The method was successfully applied to forecasting the impact of Idai with different lead times. The case study analyzed needed at least 200 ensemble members to get reliable water levels and flood results 3 d before landfall (&lt;1 % flood probability error and &lt;20 cm sampling errors). Results showed the sensitivity of forecasting, especially with increasing lead times, highlighting the importance of accounting for cyclone variability in decision-making and risk management.

Prior-Guided gated convolutional networks for rainstorm forecasting

Accurate rainstorm forecasting is crucial for the sustainable development of human society. Recently, machine learning-based rainstorm prediction methods have shown promising results. However, these methods often fail to adequately consider the prior knowledge of rainstorms and do not explicitly account for the dynamic spatio-temporal patterns of rainstorm events. This study introduces a novel end-to-end prior-informed rainstorm forecasting model that incorporates both fundamental physical priors and the spatio-temporal development patterns of rainstorms. The model utilizes a gated convolutional encoder-decoder network to effectively represent the spatio-temporal patterns of rainstorm events. A key component of the representation network is the Substantial Derivative-GuIded gated convolutional Unit (SDGiU), which updates latent states under the constraints of physical priors. Additionally, an integrated loss function is designed to minimize reconstruction errors on multiple scales and facilitate the generation of forecasts that reproduce the actual spatio-temporal patterns of rainstorm formation, development and dissipation. Experimental results on two reanalysis datasets show that the proposed forecasting model outperforms competing state-of-the-art baselines by at least 19.7% (15.0%) in overall Critical Success Index (Heidke Skill Score). Qualitative analysis indicates that the proposed model can generate predictions that are both physically consistent and spatially-temporally coherent.