Ensemble sensitivity analysis of an extreme rainfall event over the Himalayas in June 2013

https://doi.org/10.1016/j.dynatmoce.2021.101202Get rights and content

Highlights

  • Forecast Sensitivity of an extreme rainfall event over Himalayas.

  • Higher ensemble sensitivity to precipitation near midtropospheric trough.

  • The initial condition error over the maximum sensitive region impacts precipitation.

  • ESA on convection-permitting ensembles indicates non-convective precipitation.

Abstract

Forecast Sensitivity of an extreme rainfall event over the Uttarakhand state located in the Western Himalayas is investigated through Ensemble-based Sensitivity Analysis (ESA). ESA enables the assessment of forecast errors and its relation to the flow fields through linear regression approach. The ensembles are initialized from an Ensemble Kalman Filter (EnKF) Data Assimilation in Weather Research and Forecast (WRF) model. ESA is then applied to evaluate the dynamics and predictability at two different days of the extreme precipitation episode. Results indicate that the precipitation forecast over Uttarakhand is sensitive to the mid-tropospheric trough and moisture fields for both the days, in general. The day 1 precipitation shows negative sensitivity to the trough over upstream regions of the storm location while in day 2, the sensitive region is found to be located over the southward intruded branch of the mid–tropospheric trough. Perturbations introduced in the initial conditions (IC) over the most sensitive region over the west of the storm location indicate significant variations in the forecast location of precipitation. IC perturbed experiments show that the perturbation amplitude is correlated linearly with predicted change in precipitation, which becomes nonlinear as the forecast length increases. ESA performed on convection-permitting ensembles show that precipitation over the Uttarakhand is mostly non-convective. However, when the location of the response function box is moved north-westward of the Uttarakhand, the sensitivity patterns show signs of convection.

Graphical abstract

Ensemble sensitivity (shading) of 24-h accumulated precipitation averaged over the square box to the geopotential height at 500 hPa valid at 1200 UTC 16 June 2013. Contours (every 10 m) represents 500 hPa geopotential height. The black box depicts the region of response function used in the computations. The location of Uttarakhand is represented by a black dot.

  1. Download : Download high-res image (233KB)
  2. Download : Download full-size image

Introduction

The prediction of extreme rainfall events using Numerical Weather Prediction (NWP) models is challenging due to the uncertainties associated with the initial conditions and the models. Statistically reliable ensemble predictions from the different realization of initial conditions of the atmosphere are found to be robust in forecasting heavy rainfall events (Klasa et al., 2018; Mittermaier and Csima, 2017; Schumacher et al., 2013; Zhang et al., 2003). Additionally, the forecast ensembles can be employed for understanding the predictability and dynamics of extreme weather events using sensitivity analysis (Bednarczyk and Ancell, 2015; Berman et al., 2017; Zhang and Meng, 2018; KU et al., 2020).

Though the deterministic adjoint-based method (Langland et al., 1995; Zhu and Gelaro, 2008) provides a platform for sensitivity analysis, it is limited by the assumptions of linear error growth and simplified physics. Torn and Hakim (2008) proposed an ensemble-based approach for estimating the relationship between the initial conditions and model forecast fields using ensemble sensitivity analysis (ESA). ESA uses ensemble statistics to estimate the sensitivity of initial conditions to forecast metric, which is widely applied to investigate the dynamics of weather events such as extratropical cyclones (Torn and Hakim, 2009; Zheng et al., 2013; Chang et al., 2013), tropical cyclones (Gombos et al., 2012; Torn et al., 2018; Xie et al., 2013), and heavy rainfall events (Berman et al., 2017; Zhang and Meng, 2018).

Extreme precipitation events over and near the Himalayas are often associated with large-scale synoptic conditions such as the southward intrusion of the upper-level westerly trough and northward propagating moisture-laden low-level circulation (Vellore et al., 2016). The heavy rainfall that occurred over the Uttarakhand state in June 2013 is one such event with strong synoptic forcing that caused massive destruction to life and properties. The major synoptic features that lead to such an event are as follows. A midtropospheric trough embedded with the westerly flow is developed by June 10, 2013. At 700 hPa level, dry air from northwest is advected to the Uttarakhand location from northwest whereas the wind at 850 hPa level has transported moist air from the Arabian Sea. The presence of dry air over the upper levels, and warm, moist air over the lower levels produced favourable conditions for intense convection, which has contributed heavy convective precipitation to the Uttarakhand that has moistened the soil of Himalayan escarpment. The actual flood event occurred four days later and the prevailing synoptic conditions were different from the convective event. The short-wave trough on the westerlies has extended southwards by 13 June 2013, which finally merged with the closed cyclonic circulation that migrated westward from the Bay of Bengal. This has resulted in extensive north-eastward flow that has transported moisture into the regions of flood that produced unprecedented rainfall and flooding over the Uttarakhand (Kumar et al., 2016). Although the key factors responsible for the Uttarakhand storm have been recognized in the literature (Chevuturi and Dimri, 2016; Dobhal et al., 2013; Dube et al., 2014; Houze et al., 2017; Kotal et al., 2014; Ranalkar et al., 2016; Krishnamurti and Kumar, 2017), the uncertainties in the storm producing factors, its relative importance in the development of the storm, and the predictability aspects have not been investigated.

In the present study, the Uttarakhand heavy rainfall and flooding event of June 2013 will be examined using the ESA approach. In the context of climate change, it has been reported that the unprecedented extreme precipitation episodes and flooding events are on the rise over the Himalayas during the past few decades (Priya et al., 2017). Hence, it is important to advance our understanding of the key processes that lead to such events and quantify its predictability characteristics. Fig. 1 shows the spatial distribution of accumulated precipitation from the TRMM satellite and ensemble mean forecast using the Weather Research and Forecast (WRF) model valid at 1200UTC of 16 and 17, June 2013 (hereafter D1 and D2). As can be seen from Fig. 1, the location and intensity errors associated with the forecasted precipitation are substantial and the location of precipitation maxima in the model forecast is shifted northward as compared to the TRMM observations, for both the days. The objective of the present study is to identify the uncertainties in the factors that lead to errors in the location and intensity of precipitation maxima using an ensemble framework. Further, the study will also address the predictability aspects of the event in synoptic as well as in convective scales by applying ESA.

The paper is outlined as follows. Section 2 provides an overview of the case considered in this study. A description of the model, data assimilation system, and ESA are provided in Section 3. The results obtained for both synoptic-scale and convective-scale are discussed in Section 4 while Section 5 summarizes the study.

Section snippets

Case overview

The state of Uttarakhand is located at the foothills of the Himalayas in the Indian subcontinent. This region comes under the influence of monsoonal as well as large-scale extratropical circulation and hence vulnerable to the intense precipitation episodes. In the year 2013 the Indian Summer Monsoon advanced rapidly towards northern India and covered the entire country by June 16, after its onset over Kerala (Kumar and Krishnamurti, 2016). The Uttarakhand and its adjoining areas experienced a

Model and data assimilation system

Numerical experiments are performed using the Advanced Weather Research and Forecast (ARW-WRF) model (Skamarock et al., 2005) of version 3.8.1. The WRF is a non-hydrostatic, fully compressible model with an ensemble of parameterization schemes. The parameterization schemes used in this study are as follows: WRF-Single Moment five-class for microphysics (Hong et al., 2004), Kain-Fritsch for cumulus (Kain, 2004), Dudhia for shortwave radiation (Dudhia, 1989), Yonsei University (YSU) for the

Results

This section investigates the synoptic and convective scale features associated with the Uttarakhand heavy rainfall using the ESA approach.

Summary and conclusions

This study utilizes the ensemble forecasts initialized from an EnKF DA system using the WRF model to understand the dynamics and predictability of a heavy rainfall event over the Uttarakhand, India, during 14–17 June 2013. ESA is employed in synoptic and convective scale ensembles to identify the multi-scale aspect of the weather event. Further, predictability aspects of the heavy rainfall event is explored by applying perturbations to the analysis ensembles in the most sensitive regions and

CRediT authorship contribution statement

Babitha George: Software, Validation, Formal analysis, Investigation, Writing - original draft, Visualization. Govindan Kutty: Conceptualization, Methodology, Resources, Writing - review & editing, Supervision.

Acknowledgments

We acknowledge the Aaditya High-Performance Computer at the Indian Institute of Tropical Meteorology, Pune for providing the computing resources for this work. The GFS data, the observations in PREPBUFR format and the TRMM datasets were obtained from their website at https://www.ncdc.noaa.gov/data-access/model-data/model-datasets/global-forcast-system-gfs, https://rda.ucar.edu/datasets/ds337.0/ and https://trmm.gsfc.nasa.gov/, respectively.

References (43)

  • A. Dube et al.

    Forecasting the heavy rainfall during Himalayan flooding—June 2013

    Weather Clim. Extremes

    (2014)
  • J.L. Anderson

    An ensemble adjustment Kalman filter for data assimilation

    Mon. Weather Rev.

    (2001)
  • J. Anderson

    Spatially and temporally varying adaptive covariance inflation for ensemble filters

    Tellus A

    (2009)
  • D. Barker et al.

    The weather research and forecasting model’s community variational/ensemble data assimilation system: WRFDA

    Bull. Am. Meteorol. Soc.

    (2012)
  • C.N. Bednarczyk et al.

    Ensemble sensitivity analysis applied to a southern plains convective event

    Mon. Weather Rev.

    (2015)
  • J.D. Berman et al.

    Sensitivity of Northern Great Plains convection forecasts to upstream and downstream forecast errors

    Mon. Weather Rev.

    (2017)
  • E.K.M. Chang et al.

    Medium-range ensemble sensitivity analysis of two extreme pacific extratropical cyclones

    Mon. Weather Rev.

    (2013)
  • F. Chen et al.

    Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: model implementation and sensitivity

    Mon. Weather Rev.

    (2001)
  • A. Chevuturi et al.

    Investigation of Uttarakhand (India) disaster-2013 using weather research and forecasting model

    Nat. Hazards

    (2016)
  • D. Dobhal et al.

    Kedarnath disaster: facts and plausible causes

    Curr. Sci.

    (2013)
  • J. Dudhia

    Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model

    J. Atmos. Sci.

    (1989)
  • L. Garcies et al.

    Ensemble sensitivities of the real atmosphere: application to Mediterranean intense cyclones

    Tellus A

    (2009)
  • G. Gaspari et al.

    Construction of correlation functions in two and three dimensions

    Q. J. R. Meteorol. Soc.

    (1999)
  • D. Gombos et al.

    Ensemble statistics for diagnosing dynamics: tropical cyclone track forecast sensitivities revealed by ensemble regression

    Mon. Weather Rev.

    (2012)
  • S.-Y. Hong et al.

    A revised approach to ice microphysical processes for the bulk parameterization of clouds and precipitation

    Mon. Weather Rev.

    (2004)
  • S.-Y. Hong et al.

    A new vertical diffusion package with an explicit treatment of entrainment processes

    Mon. Weather Rev.

    (2006)
  • R. Houze et al.

    Multiscale aspects of the storm producing the June 2013 flooding in Uttarakhand, India

    Mon. Weather Rev.

    (2017)
  • S. Joseph et al.

    North Indian heavy rainfall event during June 2013: diagnostics and extended range prediction

    Clim. Dyn.

    (2015)
  • J.S. Kain

    The Kain–fritsch convective parameterization: an update

    J. Appl. Meteorol. Climatol.

    (2004)
  • C. Klasa et al.

    An evaluation of the convection‐permitting ensemble COSMO‐E for three contrasting precipitation events in Switzerland

    Q. J. R. Meteorol. Soc.

    (2018)
  • S. Kotal et al.

    Catastrophic heavy rainfall episode over Uttarakhand during 16–18 June 2013–observational aspects

    Curr. Sci.

    (2014)
  • Cited by (5)

    • Simulations of an extreme rainstorm event (1056.7 mm/day) along the South China coast: Cloud microphysical processes and maintenance mechanism of rainstorm

      2023, Atmospheric Research
      Citation Excerpt :

      Severe rainstorm forecasting remains a great challenge (Meng et al., 2019). In recent years, extensive studies have been done on extreme precipitation, which were mostly a discussion of the circulation pattern and triggering mechanism of rainstorms based on synoptic meteorology (Cowan et al., 2019;Ran et al., 2021; Su et al., 2016;Dong et al., 2020; Qian et al., 2021), some studied the mesoscale convective structure of rainstorms (Wang et al., 2020; Chen and Li, 2021), some others explored the sensitivity of severe rainstorms to initial conditions of numerical models and different parameterized schemes (Meng et al., 2019;George and Kutty, 2021; Wang et al., 2021). In our previous studies, we analyzed the synoptic causes of the extreme rainstorm event (1056.7 mm/day) along the South China coast in summer 2018 (Dong et al., 2020), and have also successfully simulated the extreme event using convection-permitting WRF model.

    • Multivariate ensemble sensitivity analysis applied for an extreme rainfall over Indian subcontinent

      2022, Atmospheric Research
      Citation Excerpt :

      The sensitivity analysis thus obtained by regressing the forecast variable independently on each analysis variable is known as univariate ensemble sensitivity (Torn and Hakim, 2008). Univariate ensemble sensitivity is widely applied to understand the dynamics of forecast errors associated with various weather systems such as extratropical cyclones (Garcies and Homar, 2010; Zheng et al., 2013), tropical cyclones (Brown and Hakim, 2015; Rios-Berrios et al., 2016), African easterly waves (Torn, 2010), heavy rainfall events (George and Kutty, 2022, 2021; Shen et al., 2020; Yu and Meng, 2016; Zhang and Meng, 2018) and convective-scale events (Bednarczyk and Ancell, 2015; Berman et al., 2017; Torn and Romine, 2015; Zhang et al., 2020). However, the possibility of univariate ensemble sensitivity to overestimate the forecast responses due to sampling errors cannot be neglected.

    • Ensemble-based estimates of the impact of potential observations

      2023, Quarterly Journal of the Royal Meteorological Society
    View full text