Deep Learning for Spatially Explicit Prediction of Synoptic-Scale Fronts

AbstractThis paper describes the use of convolutional neural nets (CNN), a type of deep learning, to identify fronts in gridded data, followed by a novel postprocessing method that converts probability grids to objects. Synoptic-scale fronts are often associated with extreme weather in the midlatitudes. Predictors are 1000-mb (1 mb = 1 hPa) grids of wind velocity, temperature, specific humidity, wet-bulb potential temperature, and/or geopotential height from the North American Regional Reanalysis. Labels are human-drawn fronts from Weather Prediction Center bulletins. We present two experiments to optimize parameters of the CNN and object conversion. To evaluate our system, we compare the objects (predicted warm and cold fronts) with human-analyzed warm and cold fronts, matching fronts of the same type within a 100- or 250-km neighborhood distance. At 250 km our system obtains a probability of detection of 0.73, success ratio of 0.65 (or false-alarm rate of 0.35), and critical success index of 0.52. These values drastically outperform the baseline, which is a traditional method from numerical frontal analysis. Our system is not intended to replace human meteorologists, but to provide an objective method that can be applied consistently and easily to a large number of cases. Our system could be used, for example, to create climatologies and quantify the spread in forecast frontal properties across members of a numerical weather prediction ensemble.

Download Full-text

Spurious Grid-Scale Precipitation in the North American Regional Reanalysis

Monthly Weather Review ◽

10.1175/mwr3375.1 ◽

2007 ◽

Vol 135 (6) ◽

pp. 2168-2184 ◽

Cited By ~ 33

Author(s):

Gregory L. West ◽

W. James Steenburgh ◽

William Y. Y. Cheng

Keyword(s):

North American ◽

Prediction Models ◽

Weather Prediction ◽

Regional Climate ◽

Potential Temperature ◽

Climate Data ◽

The North ◽

North American Regional Reanalysis ◽

Upper Level ◽

Regional Reanalysis

Abstract Spurious grid-scale precipitation (SGSP) occurs in many mesoscale numerical weather prediction models when the simulated atmosphere becomes convectively unstable and the convective parameterization fails to relieve the instability. Case studies presented in this paper illustrate that SGSP events are also found in the North American Regional Reanalysis (NARR) and are accompanied by excessive maxima in grid-scale precipitation, vertical velocity, moisture variables (e.g., relative humidity and precipitable water), mid- and upper-level equivalent potential temperature, and mid- and upper-level absolute vorticity. SGSP events in environments favorable for high-based convection can also feature low-level cold pools and sea level pressure maxima. Prior to 2003, retrospectively generated NARR analyses feature an average of approximately 370 SGSP events annually. Beginning in 2003, however, NARR analyses are generated in near–real time by the Regional Climate Data Assimilation System (R-CDAS), which is identical to the retrospective NARR analysis system except for the input precipitation and ice cover datasets. Analyses produced by the R-CDAS feature a substantially larger number of SGSP events with more than 4000 occurring in the original 2003 analyses. An oceanic precipitation data processing error, which resulted in a reprocessing of NARR analyses from 2003 to 2005, only partially explains this increase since the reprocessed analyses still produce approximately 2000 SGSP events annually. These results suggest that many NARR SGSP events are not produced by shortcomings in the underlying Eta Model, but by the specification of anomalous latent heating when there is a strong mismatch between modeled and assimilated precipitation. NARR users should ensure that they are using the reprocessed NARR analyses from 2003 to 2005 and consider the possible influence of SGSP on their findings, particularly after the transition to the R-CDAS.

Download Full-text

Prediction of synoptic-scale sea level pressure over the Indian monsoon region using deep learning

10.36227/techrxiv.14377298 ◽

2021 ◽

Author(s):

Aryaman Sinha ◽

Mayuna Gupta ◽

K S S Sai Srujan ◽

Hariprasad Kodamana ◽

Sandeep Sukumaran

Keyword(s):

Deep Learning ◽

Sea Level ◽

Short Term Memory ◽

Weather Forecasting ◽

Weather Prediction ◽

Learning Model ◽

Synoptic Scale ◽

Sea Level Pressure ◽

Level Pressure ◽

Deep Learning Model

<div><div><div><p>The synoptic-scale (3 - 7 days) variability is a dominant contributor to the Indian summer monsoon (ISM) seasonal precipitation. An accurate prediction of ISM precipitation by dynamical or statistical models remains a challenge. Here we show that the sea level pressure (SLP) can be used as a proxy to predict the active-break cycle as well as the genesis of low- pressure-systems (LPS), using a deep learning model, namely, convolutional long short-term memory (ConvLSTM) networks. The deep learning model is able to reliably predict the daily SLP anomalies over Central India and the Bay of Bengal at a lead time of 7 days. As the fluctuations in SLP drive the changes in the strength of the atmospheric circulation, the prediction of SLP anomalies is useful in predicting the intensity of ISM. It is demonstrated that the ConvLSTM possesses better prediction skill compared to a conventional numerical weather prediction model, indicating the usefulness of a physics guided deep learning model in medium range weather forecasting.</p></div></div></div>

Download Full-text

Atmospheric boundary layer dynamics from balloon soundings worldwide: CLASS4GL v1.0

Geoscientific Model Development ◽

10.5194/gmd-12-2139-2019 ◽

2019 ◽

Vol 12 (5) ◽

pp. 2139-2153 ◽

Cited By ~ 4

Author(s):

Hendrik Wouters ◽

Irina Y. Petrova ◽

Chiel C. van Heerwaarden ◽

Jordi Vilà-Guerau de Arellano ◽

Adriaan J. Teuling ◽

...

Keyword(s):

Boundary Layer ◽

Atmospheric Boundary Layer ◽

Mixed Layer ◽

Weather Prediction ◽

Potential Temperature ◽

Specific Humidity ◽

Weather Extremes ◽

Coupled Models ◽

Layer Height ◽

Mixed Layer Height

Abstract. The coupling between soil, vegetation and atmosphere is thought to be crucial in the development and intensification of weather extremes, especially meteorological droughts, heat waves and severe storms. Therefore, understanding the evolution of the atmospheric boundary layer (ABL) and the role of land–atmosphere feedbacks is necessary for earlier warnings, better climate projection and timely societal adaptation. However, this understanding is hampered by the difficulties of attributing cause–effect relationships from complex coupled models and the irregular space–time distribution of in situ observations of the land–atmosphere system. As such, there is a need for simple deterministic appraisals that systematically discriminate land–atmosphere interactions from observed weather phenomena over large domains and climatological time spans. Here, we present a new interactive data platform to study the behavior of the ABL and land–atmosphere interactions based on worldwide weather balloon soundings and an ABL model. This software tool – referred to as CLASS4GL (http://class4gl.eu, last access: 27 May 2018) – is developed with the objectives of (a) mining appropriate global observational data from ∼15 million weather balloon soundings since 1981 and combining them with satellite and reanalysis data and (b) constraining and initializing a numerical model of the daytime evolution of the ABL that serves as a tool to interpret these observations mechanistically and deterministically. As a result, it fully automizes extensive global model experiments to assess the effects of land and atmospheric conditions on the ABL evolution as observed in different climate regions around the world. The suitability of the set of observations, model formulations and global parameters employed by CLASS4GL is extensively validated. In most cases, the framework is able to realistically reproduce the observed daytime response of the mixed-layer height, potential temperature and specific humidity from the balloon soundings. In this extensive global validation exercise, a bias of 10.1 m h−1, −0.036 K h−1 and 0.06 g kg−1 h−1 is found for the morning-to-afternoon evolution of the mixed-layer height, potential temperature and specific humidity. The virtual tool is in continuous development and aims to foster a better process understanding of the drivers of the ABL evolution and their global distribution, particularly during the onset and amplification of weather extremes. Finally, it can also be used to scrutinize the representation of land–atmosphere feedbacks and ABL dynamics in Earth system models, numerical weather prediction models, atmospheric reanalysis and satellite retrievals, with the ultimate goal of improving local climate projections, providing earlier warning of extreme weather and fostering a more effective development of climate adaptation strategies. The tool can be easily downloaded via http://class4gl.eu (last access: 27 May 2018) and is open source.

Download Full-text

Atmospheric boundary layer dynamics from balloon soundings worldwide: CLASS4GL v1.0

10.5194/gmd-2019-24 ◽

2019 ◽

Author(s):

Hendrik Wouters ◽

Irina Y. Petrova ◽

Chiel C. van Heerwaarden ◽

Jordi Vilà-Guerau de Arellano ◽

Adriaan J. Teuling ◽

...

Keyword(s):

Boundary Layer ◽

Atmospheric Boundary Layer ◽

Prediction Models ◽

Climate Adaptation ◽

Weather Prediction ◽

Potential Temperature ◽

Specific Humidity ◽

Climate Projections ◽

Weather Extremes ◽

Coupled Models

Abstract. The coupling between soil, vegetation and atmosphere is thought to be crucial in the development and intensification of weather extremes, especially meteorological droughts, heatwaves and severe storms. Therefore, understanding evolution of the atmospheric boundary layer (ABL) and the role of land–atmosphere feedbacks is necessary for earlier warnings, better climate projection and timely societal adaptation. However, this understanding is hampered by the difficulties to attribute cause–effect relationships from complex coupled models, and the irregular space–time distribution of in situ observations of the land–atmosphere system. As such, there is a need for simple deterministic appraisals that systematically discriminate land–atmosphere interactions from observed weather phenomena over large domains and climatological time spans. Here, we present a new interactive data platform to study the behaviour of the ABL and land–atmosphere interactions based on worldwide weather balloon soundings and an ABL model. This software tool – referred to as CLASS4GL (http://class4gl.eu) – is developed with the objectives to (a) mine appropriate global observational data from over 2 million weather balloon soundings since 1981 and combine them with satellite and reanalysis data, and (b) constrain and initialize a numerical model of the daytime evolution of the ABL that serves as a tool to interpret these observations mechanistically and deterministically. As a result, it fully automises extensive global model experiments to assess the effects of land and atmospheric conditions on the ABL evolution as observed in different climate regions around the world. The suitability of the set of observations, model formulations and global parameters employed by CLASS4GL is extensively validated. In most cases, the framework is able to realistically reproduce the observed daytime response of the ABL height, potential temperature and specific humidity from the balloon soundings. In this extensive global validation exercise, a bias of 0.2 m h−1, −0.052 K h−1 and 0.07 g kg−1 h−1 is found for the morning-to-afternoon evolution of the ABL height, potential temperature and specific humidity. The virtual tool is in continuous development, and aims to foster a better process-understanding of the drivers of the ABL evolution and their global distribution, particularly during the onset and amplification of weather extremes. Finally, it can also be used to scrutinize the representation of land–atmosphere feedbacks and ABL dynamics in Earth system models, numerical weather prediction models, atmospheric reanalysis, and satellite retrievals, with the ultimate goal to improve local climate projections, provide earlier warning of extreme weather, and foster a more effective development of climate adaptation strategies. The tool can be easily downloaded via http://class4gl.eu and is open source.

Download Full-text

Objective Classification of Tornadic and Nontornadic Severe Weather Outbreaks

Monthly Weather Review ◽

10.1175/2009mwr2897.1 ◽

2009 ◽

Vol 137 (12) ◽

pp. 4355-4368 ◽

Cited By ~ 32

Author(s):

Andrew E. Mercer ◽

Chad M. Shafer ◽

Charles A. Doswell ◽

Lance M. Leslie ◽

Michael B. Richman

Keyword(s):

Statistical Methods ◽

Lead Time ◽

Weather Prediction ◽

The United States ◽

Severe Weather ◽

Probability Of Detection ◽

Support Vector ◽

Synoptic Scale ◽

Skill Scores ◽

Objective Classification

Abstract Tornadoes often strike as isolated events, but many occur as part of a major outbreak of tornadoes. Nontornadic outbreaks of severe convective storms are more common across the United States but pose different threats than do those associated with a tornado outbreak. The main goal of this work is to distinguish between significant instances of these outbreak types objectively by using statistical modeling techniques on numerical weather prediction output initialized with synoptic-scale data. The synoptic-scale structure contains information that can be utilized to discriminate between the two types of severe weather outbreaks through statistical methods. The Weather Research and Forecast model (WRF) is initialized with synoptic-scale input data (the NCEP–NCAR reanalysis dataset) on a set of 50 significant tornado outbreaks and 50 nontornadic severe weather outbreaks. Output from the WRF at 18-km grid spacing is used in the objective classification. Individual severe weather parameters forecast by the model near the time of the outbreak are analyzed from simulations initialized at 24, 48, and 72 h prior to the outbreak. An initial candidate set of 15 variables expected to be related to severe storms is reduced to a set of 6 or 7, depending on lead time, that possess the greatest classification capability through permutation testing. These variables serve as inputs into two statistical methods, support vector machines and logistic regression, to classify outbreak type. Each technique is assessed based on bootstrap confidence limits of contingency statistics. An additional backward selection of the reduced variable set is conducted to determine which variable combination provides the optimal contingency statistics. Results for the contingency statistics regarding the verification of discrimination capability are best at 24 h; at 48 h, modest degradation is present. By 72 h, the contingency statistics decline by up to 15%. Overall, results are encouraging, with probability of detection values often exceeding 0.8 and Heidke skill scores in excess of 0.7 at 24-h lead time.

Download Full-text

The Role of NWP Filter for the Satellite Based Detection of Cumulonimbus Clouds

10.20944/preprints201711.0194.v1 ◽

2017 ◽

Author(s):

Richard Mülller ◽

Stephane Haussler ◽

Matthias Jerg

Keyword(s):

Weather Prediction ◽

Convective Available Potential Energy ◽

Probability Of Detection ◽

False Alarms ◽

Brightness Temperatures ◽

Critical Success ◽

High Clouds ◽

Cumulonimbus Clouds

The study investigates the role of NWP filtering for the remote sensing of Cumulonimbus Clouds (Cbs) by implementation of 14 different experiments, covering Central Europe. These experiments compiles different stability filter settings as well as the use of different channels for the InfraRed (IR) brightness temperatures. As stability filter parameters from Numerical Weather Prediction (NWP) are used. The brightness temperature information results from the IR SEVIRI instrument on-board of Meteosat Second Generation satellite and enables the detection of very cold and high clouds close to the tropopause. The satellite only approaches (no NWP filtering) result in the detection of Cbs with a relative high probability of detection, but unfortunately combined with a large False Alarm Rate (FAR), leading to a Critical Success Index (CSI) below 60 %. The false alarms results from other types of very cold and high clouds. It is shown that the false alarms can be significantly decreased by application of an appropriate NWP stability filter, leading to the increase of CSI to about 70 % . A brief review and reflection of the literature clarifies that the this function of the NWP filter can not be replaced by MSG IR spectroscopy. Thus, NWP filtering is strongly recommended to increase the quality of satellite based Cb detection. Further, it has been shown that the well established convective available potential energy (CAPE) and the convection index (KO) works well as stability filter.

Download Full-text

Assessment of the GOES-16 Clear Sky Mask Product over the Contiguous USA Using CALIPSO Retrievals

Remote Sensing ◽

10.3390/rs12101630 ◽

2020 ◽

Vol 12 (10) ◽

pp. 1630

Author(s):

Pedro A. Jiménez

Keyword(s):

Weather Prediction ◽

Wrf Model ◽

Probability Of Detection ◽

Clear Sky ◽

The North ◽

Attractive Option ◽

Spatio Temporal ◽

Sky Conditions ◽

Cloud Mask ◽

Horizontal Grid

Cloud initialization is a challenge in numerical weather prediction. Probably the most relevant observations for this task come from geostationary satellites. These satellites provide the cloud mask with high spatio-temporal resolution and low latencies. The low latency is an attractive option for nowcasting systems such as the solar irradiance nowcasting model MAD-WRF. In this study we examine the potential of using the cloud mask from the GOES-16 satellite over the contiguous U.S. for this particular application. With this aim, the GOES-16 cloud mask product is compared against CALIPSO retrievals during a two year period. Both the GOES-16 data and the CALIPSO retrievals are interpolated to a grid that covers the contiguous U.S. at 9 km of horizontal grid spacing that is being used in MAD-WRF nowcasts. Results indicate a probability of detection, or accuracy, of all sky conditions of 86.0%. However, the accuracy is higher for cloud detections, 90.9% than for clear sky detections 74.8%. The lower performance of clear sky retrievals is a result of missdetections during daytime. This is especially clear for summer, and for regions to the north of parallel 36 during winter. However, regions to the south of parallel 36 show acceptable performance during both daytime and nighttime. It is over these regions wherein the cloud mask product should show its largest potential to enhance the cloud initialization in the MAD-WRF model.

Download Full-text

Synoptic-Scale Precursors to Tropical Cyclone Rapid Intensification in the Atlantic Basin

Advances in Meteorology ◽

10.1155/2015/814043 ◽

2015 ◽

Vol 2015 ◽

pp. 1-16 ◽

Cited By ~ 6

Author(s):

Alexandria Grimes ◽

Andrew E. Mercer

Keyword(s):

Tropical Cyclone ◽

Large Scale ◽

Potential Temperature ◽

Vertical Shear ◽

Specific Humidity ◽

Support Vector ◽

Rapid Intensification ◽

Synoptic Scale ◽

Atlantic Basin ◽

And Cluster Analysis

Forecasting rapid intensification (hereafter referred to as RI) of tropical cyclones in the Atlantic Basin is still a challenge due to a limited understanding of the meteorological processes that are necessary for predicting RI. To address this challenge, this study considered large-scale processes as RI indicators within tropical cyclone environments. The large-scale processes were identified by formulating composite map types of RI and non-RI storms using NASA MERRA data from 1979 to 2009. The composite fields were formulated by a blended RPCA and cluster analysis approach, yielding multiple map types of RI’s and non-RI’s. Additionally, statistical differences in the large-scale processes were identified by formulating permutation tests, based on the composite output, revealing variables that were statistically significantly distinct between RI and non-RI storms. These variables were used as input in two prediction schemes: logistic regression and support vector machine classification. Ultimately, the approach identified midlevel vorticity, pressure vertical velocity, 200–850 hPa vertical shear, low-level potential temperature, and specific humidity as the most significant in diagnosing RI, yielding modest skill in identifying RI storms.

Download Full-text

Prediction of synoptic-scale sea level pressure over the Indian monsoon region using deep learning

10.36227/techrxiv.14377298.v1 ◽

2021 ◽

Author(s):

Aryaman Sinha ◽

Mayuna Gupta ◽

K S S Sai Srujan ◽

Hariprasad Kodamana ◽

Sandeep Sukumaran

Keyword(s):

Deep Learning ◽

Sea Level ◽

Short Term Memory ◽

Weather Forecasting ◽

Weather Prediction ◽

Learning Model ◽

Synoptic Scale ◽

Sea Level Pressure ◽

Level Pressure ◽

Deep Learning Model

Download Full-text

The Impacts of Soil Moisture Initialization on the Forecasts of Weather Research and Forecasting Model: A Case Study in Xinjiang, China

Water ◽

10.3390/w12071892 ◽

2020 ◽

Vol 12 (7) ◽

pp. 1892

Author(s):

Hailiang Zhang ◽

Junjian Liu ◽

Huoqing Li ◽

Xianyong Meng ◽

Ablimitijan Ablikim

Keyword(s):

Soil Moisture ◽

Land Surface ◽

Weather Prediction ◽

Potential Temperature ◽

Wrf Model ◽

Specific Humidity ◽

Water Evaporation ◽

Weather Research And Forecasting ◽

Soil Moisture Initialization ◽

Weather Research

Soil moisture is a critical parameter in numerical weather prediction (NWP) models because it plays a fundamental role in the exchange of water and energy cycles between the atmosphere and the land surface through evaporation. To improve the forecast skills of the Weather Research and Forecasting (WRF) model in Xinjiang, China, this study investigated the impacts of soil moisture initialization on the WRF forecasts by performing a series of simulations. A group of simulations was conducted using the single-column model (SCM) from 1200 UTC on 15 to 18 August 2019, at Urumchi, Xinjiang (43.78° N, 87.6° E); another was performed using the WRF model for a real weather case in Xinjiang from 0000 UTC 15 August to 1200 UTC 18 August 2019, which included an episode of heavy precipitation and gales. Our most notable findings are as follows. Specific humidity increases and potential temperature decreases persistently when soil moisture increases because of soil water evaporation. Soil moisture initialization could impact the energy budget and modulate the partition of the total available energy at the land surface significantly through evaporation and the greenhouse effect. Replacing the soil moisture with a proper multiple of the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) soil moisture data could significantly improve the critical success index (CSI) and frequency bias (FBIAS) of precipitation and the root-mean-squared errors (RMSEs) of 2-m specific humidity and 2-m temperature. These findings indicate the prospect of a new way to improve the forecast skills of WRF in Xinjiang or other similar regions.

Download Full-text