Piyush Garg

Mr. Piyush Garg

University of Illinois at Champaign Urbana
orcid

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 January 2026 in Science Advances
Verified
Journal Article
Open access
JPJaideep Pathak
Jaideep Pathak
YCYair Cohen
Yair Cohen
Piyush GargPiyush Garg
Piyush Garg
PHPeter Harrington
Peter Harrington
NBNoah Brenowitz
Noah Brenowitz
MPMichael Pritchard
Michael Pritchard
Storm-scale convection-allowing models (CAMs) explicitly resolve convective dynamics within the atmosphere to predict the evolution of thunderstorms and mesoscale convective systems that result in damaging extreme weather. Deep learning models have, thus far, not proven skillful in this regime of kilometer-scale atmospheric simulation, despite being competitive at coarser resolutions with state-of-the-art global, medium-range weather forecasting. We present a generative diffusion model called StormCast, which emulates the High-Resolution Rapid Refresh (HRRR)—National Oceanic and Atmospheric Administration’s state-of-the-art 3-kilometer operational CAM. StormCast autoregressively predicts 99 state variables at the kilometer scale using a 1-hour time step, with dense vertical resolution in the atmospheric boundary layer, conditioned on 26 synoptic variables. We show successfully learned kilometer-scale dynamics including competitive 1- to 6-hour forecast skill for composite radar reflectivity alongside physically realistic convective cluster evolution, moist updrafts, and cold pool morphology. These results present opportunities for improving kilometer-scale regional ML weather prediction and future climate hazard dynamical downscaling.

28 October 2025 in Journal of Advances in Modeling Earth Systems
Verified
Journal Article
Open access
PMPeter Manshausen
Peter Manshausen
YCYair Cohen
Yair Cohen
PHPeter Harrington
Peter Harrington
JPJaideep Pathak
Jaideep Pathak
MPMike Pritchard
Mike Pritchard
NBNoah Brenowitz
Noah Brenowitz
Data assimilation of observations into full atmospheric states is essential for weather forecast model initialization. Recently, methods for deep generative data assimilation have been proposed which allow for using new input data without retraining the model. They could also dramatically accelerate the costly data assimilation process used in operational regional weather models. Here, in a central US testbed, we demonstrate the viability of Score‐based Data Assimilation (SDA) in the context of realistically complex km‐scale weather. We train an unconditional diffusion model to generate snapshots of a state‐of‐the‐art km‐scale analysis product, the High Resolution Rapid Refresh. Then, using SDA to incorporate sparse weather station data, the model produces maps of precipitation and surface winds. The generated fields display physically plausible structures, such as gust fronts, and sensitivity tests confirm learned physics through multivariate relationships. Preliminary skill analysis shows the approach already outperforms a naive baseline of the High‐Resolution Rapid Refresh system itself. By incorporating observations from 40 weather stations, 10% lower RMSEs on left‐out stations are attained. Despite some lingering imperfections such as insufficiently disperse ensemble DA estimates, we find the results overall an encouraging proof of concept, and the first at km‐scale. It is a ripe time to explore extensions that combine increasingly ambitious regional state generators with an increasing set of in situ, ground‐based, and satellite remote sensing data streams.

1 October 2017 in Quarterly Journal of the Royal Meteorological Society
Verified
Journal Article
Closed access
PDF file
WFWendilyn J. Flynn
Wendilyn J. Flynn
SNStephen Nesbitt
Stephen Nesbitt
AAAlison M. Anders
Alison M. Anders
Piyush GargPiyush Garg
Piyush Garg

28 February 2020 in Journal of Geophysical Research: Atmospheres
Verified
Journal Article
Open access
PDF file
Piyush GargPiyush Garg
Piyush Garg
SNStephen Nesbitt
Stephen Nesbitt
Timothy LangTimothy Lang
Timothy Lang
GPGeorgios Priftis
Georgios Priftis
TCThemis Chronis
Themis Chronis
DHDeanna A. Hence
Deanna A. Hence
Convective cold pools over oceans contribute to the air‐sea exchange of heat and moisture. Beyond cold pools’ physical properties, cold pools and their intersections are important mechanisms controlling the life cycle of tropical convective systems. However, few observational techniques are available to routinely observe the properties of cold pools and their association with convection. The primary goal of this study is to demonstrate a new technique using gradients in remotely sensed ocean vector winds as a proxy for the structure and characteristics of tropical oceanic cold pools. Using Advanced Scatterometer (ASCAT) vector wind retrievals with field of view (FOV) of 12.5 km, regions of enhanced gradients in surface winds are identified as gradient features (GFs). A 3‐km Weather Research and Forecasting (WRF) model simulation was carried out to illustrate the technique over the Indian Ocean during the 2011 DYNAmics of Madden‐Julian Oscillation (DYNAMO) field campaign. A forward operator was applied to the model wind fields, allowing direct comparisons of wind gradients and virtual temperature‐defined cold pool thermodynamic and kinematic properties. In situ buoy measurements of air temperature were also used to validate ASCAT‐identified cold pools. Quantitative skill metrics demonstrate that the technique has a low false alarm rate (< 10%), high critical success index (>85%) and a low bias (~1). The GF technique was then applied to ASCAT‐retrieved ocean vector winds, revealing global variations in cold pool properties. These properties are shown to be related to the properties of global convective systems observed by the Tropical Rainfall Measurement Mission (TRMM).

31 March 2021 in Remote Sensing
Verified
Journal Article
Open access
PDF file
GPGeorgios Priftis
Georgios Priftis
Timothy LangTimothy Lang
Timothy Lang
Piyush GargPiyush Garg
Piyush Garg
SNStephen Nesbitt
Stephen Nesbitt
RLRichard D. Lindsley
Richard D. Lindsley
TCThemistoklis Chronis
Themistoklis Chronis
Outflow boundaries induced by cold-pools are a key characteristic of convective systems related to microphysical and kinematic processes during the mature stage of their lifecycle. Over the ocean, such kinematic processes are associated with low-level wind modulations that are captured by scatterometers. This study investigates the ability of the Advanced Scatterometer (ASCAT) wind retrievals to detect the outflow boundary associated with an oceanic mesoscale convective system (MCS). Leveraging a new technique to identify cold pools that is based on features that enclose elevated magnitude of the gradient of the wind, termed as ‘Gradient Feature’ (GF), wind retrievals at 50-, 25- and 7-km spatial resolution were utilized to explore how the characteristics of the outflow boundary vary with resolution. Ground-based radar retrievals were also implemented to assess and correct, when possible, the performance of the ASCAT retrievals. The magnitude of the gradient of the wind for the coarser resolution was an order of magnitude smaller (10−4 s−1) than the finer ones (10−3 s−1). An increase in the magnitude of the gradient wind field associated with the outflow boundary was captured by all resolutions and a respective feature was identified by the GF method. The location of the features relative to the distance from the front edge of the MCS decreased with resolution, indicating the importance of the high resolution ASCAT product to capture their extent, as well as additional smaller scale features. The effect of the background wind field on the selection of the final wind field during the ambiguity removal process for the high-resolution product is also discussed.

31 August 2021 in Journal of Climate
Verified
Journal Article
Closed access
PDF file
Piyush GargPiyush Garg
Piyush Garg
SNStephen W. Nesbitt
Stephen W. Nesbitt
TLTimothy J. Lang
Timothy J. Lang
GPGeorge Priftis
George Priftis
Tropical convection regimes range from deep organized to shallow convective systems. Mesoscale processes such as cold pools within tropical convective systems can play a significant role in the evolution of convection over land and open ocean. Although cold pools are widely observed, their diurnal properties are not well understood over tropical oceans and land. The oceanic cold pool identification metric applied herein uses the gradient feature (GF) technique and is compared with diurnally-resolved buoy-identified thermal cold pools. This study provides a first-ever diurnal climatology of GF number, area, and attributed TRMM 3B42 precipitation using a space-borne scatterometer (RapidScat). Buoy data over the Pacific, Atlantic, and Indian Ocean have been used to validate and examine the RapidScat-identified diurnal cycle of GF number and precipitation. Buoy-observed cold pool duration, precipitation, temperature, and wind speed is analyzed to understand thein situcold pool properties over tropical oceans. GF- and buoy-observed cold pool number and precipitation exhibits a similar bimodal diurnal variability with a morning and afternoon maxima, thus establishing confidence in using GF as a proxy to observe cold pools over tropical oceans. The morning peak is attributed to cold pools associated with deep moist convection while the afternoon peak is related to shallower clouds in relatively drier environments resulting in smaller cold pools over global tropical oceans.