Amit P. Kesarkar

Appraisal of recent theories to understand cyclogenesis pathways of tropical cyclone Madi (2013)

The present study aims to examine the new understanding of cyclogenesis by analyzing the genesis sequence of formation of a very severe cyclonic storm Madi (6–13 December 2013) that occurred over the Bay of Bengal. We have generated a high-resolution (18 km, 6 km, and 2 km) analysis using three-dimensional variational data assimilation technique and Weather Research and Forecasting model. The genesis sequence of Madi cyclone is analyzed using the concepts in the marsupial theory and other theories of tropical cyclone formation. Major results are as follows: the developed analysis is found useful for tracking the movement of westward moving parent disturbance from 15 days prior to the genesis; identifying developed pouch region in the Lagrangian frame of reference; understanding the evolution of the pouch and convection within the pouch region and for the study of intensification inside the pouch region. Also, large-scale priming of environment concurs with the hypotheses of the marsupial theory of tropical cyclogenesis. The analysis of dynamical and thermodynamical processes within the pouch region showed gradual moistening, uplifting of moisture, diabatic heating causing buoyant convection in the vorticity-rich environment followed by vortex tube stretching, development of convection, heavy precipitation, strengthening of lower level convergence, and hence spin-up during a day or two preceding the genesis of Madi cyclone. In general, it is concluded that intensification within pouch region during the cyclogenesis phase followed the marsupial paradigm and bottom-up mechanism.

BATAL: The Balloon Measurement Campaigns of the Asian Tropopause Aerosol Layer

AbstractWe describe and show results from a series of field campaigns that used balloonborne instruments launched from India and Saudi Arabia during the summers 2014–17 to study the nature, formati...

Embarrassingly Parallel GPU Based Matrix Inversion Algorithm for Big Climate Data Assimilation

Attemptstoharnessthebigclimatedatathatcomefromhigh-resolutionmodeloutputandadvanced sensorstoprovidemoreaccurateandrapidly-updatedweatherprediction,callforinnovationsinthe existingdataassimilationsystems.Matrixinversionisakeyoperationinamajorityofdataassimilation techniques.Hence,thisarticlepresentsout-of-coreCUDAimplementationofaniterativemethod ofmatrixinversion.Theresultsshowsignificantspeedupforevensquarematricesofsize1024X 1024andmore,withoutsacrificingtheaccuracyoftheresults.Inasimilartestenvironment,the comparisonofthisapproachwithadirectmethodsuchastheGauss-Jordanapproach,modifiedto processlargematricesthatcannotbeprocesseddirectlywithinasinglekernelcallshowsthatthe formeristwiceasefficientasthelatter.Thisaccelerationisattributedtothedivision-freedesignand theembarrassinglyparallelnatureofeverysub-taskofthealgorithm.Theparallelalgorithmhasbeen designedtobehighlyscalablewhenimplementedwithmultipleGPUsforhandlinglargematrices. KEywoRDS Big Climate Data, Convergence Rate, GPU, Iterative Method, Matrix Type Identification, Numerical Weather Prediction, Parallel Matrix Inverse, Parallel Reduction

Impact of Representing Model Error in a Hybrid Ensemble-Variational Data Assimilation System for Track Forecast of Tropical Cyclones over the Bay of Bengal

Uncertainties in the numerical weather prediction models are generally not well-represented in ensemble-based data assimilation (DA) systems. The performance of an ensemble-based DA system becomes suboptimal, if the sources of error are undersampled in the forecast system. The present study examines the effect of accounting for model error treatments in the hybrid ensemble transform Kalman filter—three-dimensional variational (3DVAR) DA system (hybrid) in the track forecast of two tropical cyclones viz. Hudhud and Thane, formed over the Bay of Bengal, using Advanced Research Weather Research and Forecasting (ARW-WRF) model. We investigated the effect of two types of model error treatment schemes and their combination on the hybrid DA system; (i) multiphysics approach, which uses different combination of cumulus, microphysics and planetary boundary layer schemes, (ii) stochastic kinetic energy backscatter (SKEB) scheme, which perturbs the horizontal wind and potential temperature tendencies, (iii) a combination of both multiphysics and SKEB scheme. Substantial improvements are noticed in the track positions of both the cyclones, when flow-dependent ensemble covariance is used in 3DVAR framework. Explicit model error representation is found to be beneficial in treating the underdispersive ensembles. Among the model error schemes used in this study, a combination of multiphysics and SKEB schemes has outperformed the other two schemes with improved track forecast for both the tropical cyclones.