Cloudburst assignment PDF

Preview

Summary

assignment on cloudburst for meteorology ...

Description

CHAPTER – 5 CLOUDBURST

CHAPTER 5 – CLOUDBURST 5.1. Introduction There have been few efforts to study extreme precipitation in the Himalayan region to better understand the various processes associated with them; specifically for the devastating Leh Cloudburst in 2010. Ashrit (2010) investigated the same to understand the evolution and intensification of the cloudburst. Kumar et al. (2012) concluded that wind circulation fields were better simulated with the WSM3 microphysics scheme in comparison with other microphysics schemes used in the study. Thayyen et al. (2013) studied the cloudburst event along with the associated flashfloods through atmospheric modelling and hydrological analysis and estimated the precipitation using a reverse algorithm based on flood measurement. Hobley et al. (2012) used geophysical changes that occurred during the Leh cloudburst event to reconstruct the precipitation pattern during the storm. Rasmussen and Houze Jr. (2012) and Kumar et al. (2014) studied the detailed mesoscale convective system occurring during the Leh flashfloods using observational data and models such as the WRF and Land Information System. Focussing on the modelling approach to simulate high intensity precipitating events, there is a need to consider the variable orography of the region (Leung and Ghan, 1995). With this consideration, higher resolution simulation would provide a clearer idea of the localized meteorological processes. Towards finer scales the convection can be resolved explicitly by grid-scale processes. This is because, the parameterization statistically calculates convection intensity instead of sub-grid level cloud, but as the resolution becomes finer the clouds attain grid scale level (Arakawa, 2004). Thus, convection is resolved explicitly for high resolutions, Gomes and Chou (2010) state that

158

CHAPTER 5 – CLOUDBURST on horizontal resolutions smaller than 3km the precipitation is produced by the cloud microphysics scheme. Review of various experiments on the subject in the literature suggests that parameterization shows errors in grid spacings between 3-25km, and hybrid methods must be applied for resolving convection (Molinari and Dudek, 1992). Weisman et al. (1997) and Done et al., (2004) show that 4km resolution with explicit physics is sufficient in mesoscale models to predict processes within a convective system. In simulation with explicitly resolved convection unfortunately no sub-grid scale convection is accounted for and fails when there is a lack of grid-scale forcing or intense convective instability (Molinari and Dudek, 1992).

Over Himalayas explicit representation of

convection in model simulation is still not explored. Thus, raising the question of effectiveness of current cumulus parameterization schemes over such a variable orographic region. The focus of this study is a heavy precipitation event that occurred over the district of Rudraprayag in Uttarakhand, India. The event affected villages in the Ukhimath (or sometimes written as “Okhimath”) Tehsil of the Rudraprayag district. Fig. 5-1a shows the daily precipitation for the month of Sep2012 (DMMC, 2012). The event took place around 13Sep2012 1930UTC (14Sep2012 0100IST); that is on the night of 13Sep2012 (Sphere India, 2012). High intensity precipitation was recorded in the region which further triggered the landslide and debris flow in the early hours of 14Sep2012. The event was associated with loss of life and also severe damage to infrastructure and property, with reports of up to 34 villages being affected (DMMC, 2012). Fig. 5-1b shows the DWR based reflectivity observed over the region during 13Sep2012 1930UTC (14Sep2012 0100IST). The convective activity associated with the Rudraprayag high

159

CHAPTER 5 – CLOUDBURST intensity precipitation event is depicted in the figure as a localized storm cell demarcated by a red circle. Fig. 5-1c-d displays the image of the Meteosat-5 satellite infrared brightness temperature. The infrared brightness temperature is a measure of the received radiation intensity which depicts cloud cover and storm development. The storm intensity peaks around 13Sep2012 1800UTC and intensity reduced after 13Sep2012 2200UTC. Low values of brightness temperature seen over the Rudraprayag region indicate presence of high cloud tops, which in turn represent higher convective activity over the said region. Various sources (e.g., Sphere India, 2012) reported this as a cloudburst event even though it does not exactly fit the criteria described above of a cloudburst event, but more accurately is an intense precipitation event. The region around Rudraprayag is a disaster-prone region due to its geotectonic configuration and weather conditions, with frequent occurrences of disasters like earthquakes, thunderstorms, flashfloods, and landslides. This event, however, is considered as one of the biggest tragedies in terms of number of lives lost since the formation of the state in 2000 (DMMC, 2012). The Himalayan region is prone to such high intensity precipitation, which wreak havoc over the region. There is a high propensity of secondary impacts associated with heavy precipitation over Himalayan terrain. With the region having sparse observational network, analysing such events using observational data is not possible. But there is a need to better understand such events in greater detail to mitigate the severe impacts of such commonly occurring meteorological disasters. Thus, the goal is to provide a framework for using numerical weather prediction techniques for simulation of such weather events over this region.

160

CHAPTER 5 – CLOUDBURST

(a)

(b)

(c)

(d)

Fig. 5-1: (a) Daily rainfall (mm) for 01-20Sep2012 at Ukhimath (location reported for cloudburst event) (Source: DMMC, 2012) (b) Observed reflectivity by IMD Patiala Radar

for

13Sep2012,

1930UTC

(Source:

IMD

Radar

output

archived

at

http://ddgmui.imd.gov.in/storm2012/) with Ukhimath region is marked by red circle. (c) Meteosat-5 IR-Tb (Brightness Temperature) for 13Sep2012 1800UTC and (d) same as (c) but at 13Sep2012 2200UTC.

161

CHAPTER 5 – CLOUDBURST Due to the steep variation in orography of the region, specific emphasis is given for identifying various atmospheric triggers initiating convection in relation to the orographic forcings of the Himalayan terrain. To capture the orographic effect, the event is simulated at fine scale resolution, and for simulating at high resolutions (specifically around 3km) use of cumulus parameterization is not recommended by many studies as discussed above. However, resolving convection explicitly may not be able to capture onset and magnitude of the convection accurately. 5.2. Objectives  To understand a cloudburst event over the Himalayan region with description of the large scale flow and associated precipitation during such a storm.  The study aims to explain the sensitivity of model for cumulus parameterization at higher resolutions in regions of complex topography.  Using numerical simulation to understand the various dynamical characteristics associated with the intense precipitation event, with the objective to comprehend the development of storm and analyse model resolved grid scale processes like circulation and convection. 5.3. Methodology, Experimental Design and Data In this study, the WRF model (version 3.4.1) with the ARW dynamic solver is used to simulate the intense precipitation event. The model domain is set up over the Uttarakhand region with a central domain point at Ukhimath (30°30’N 79°15’E), where the high intensity precipitation was reported, with triple nests of 27 km, 9 km, and 3 km, respectively. Fig. 5-2 shows the extent of these domains with the topography of the region.

162

CHAPTER 5 – CLOUDBURST

Fig. 5-2: Model domain and topography (×103 m; shaded). Shaded region corresponds to model domain 1 (27 km horizontal model resolution), and fan-shaped boxes with black lines indicate model domain 2 (9 km horizontal model resolution), and domain 3 (3 km horizontal model resolution) as marked in the box itself. Detailed topography (×103 m; shaded) of domain 3 is shown below. Plus sign indicates location of Ukhimath.

163

CHAPTER 5 – CLOUDBURST Model simulations are analyzed for 72 hrs starting with 11 Sep 0000UTC, for developing an understanding of all meteorological processes contributing to the storm development and propagation. In the current study, not only model performance in simulating the precipitation event is evaluated, but the large scale atmospheric processes leading to the weather event are also analyzed. With such an aim, a three day simulation is deemed necessary for the study. The datasets used in this study are listed below: 

GFS-FNL data as ICBC for model simulation



USGS geographical data as terrain input data



Datasets used for verification of precipitation field are –TRMM; IMD-NCMRWF



NOAA interpolated daily mean OLR is for verification of OLR model simulated field. Due to the coarser resolution of the FNL dataset a 12-hour spin-up period has

been incorporated before starting the model integration (Skamarock, 2004), thus initializing the model at 10 Sep 1200UTC. The model configuration and details are provided in Table 5-1. This model configuration has been chosen in this study according to previous work done on the Leh cloudburst by Ashrit (2010), Kumar et al. (2012), and Thayyen et al. (2013). Also, two experiments with different cumulus parameterization settings were performed for the innermost nest. One simulation was run without cumulus parameterization at 3km resolution (hereafter termed as explicit physics simulation) and other simulation included Kain Fritsch cumulus scheme (Kain, 2004) (hereafter termed as parameterized physics simulation) for the same.

164

CHAPTER 5 – CLOUDBURST Table 5-1: Description of model configuration Model

WRF Version 3.4.1

Map Projection

Lambert Conformal

Central Point of Domain

30°30’N 79°15’E

Horizontal Resolution

Triple Nest: 27km, 9km and 3km

Horizontal Grid Scheme

Arakawa C-grid

Time Integration Scheme

Time-split integration using 2nd-order Runge-Kutta scheme

Time Step

108 s

Land surface model

Noah Land Surface Model

Surface layer model

MM5 Similarity Model

Radiation Scheme

Shortwave – Dudhia Scheme Longwave - RRTM

Microphysics

WRF Single Moment 6 Class (Hong and Lim, 2006)

Planetary boundary layer

Yonsei University Scheme (Hong et al., 2006)

Cumulus Parametrization

A. Explicit Physics B. Parameterized Physics (Kain-Fritsch Scheme) (Kain, 2004)

165

CHAPTER 5 – CLOUDBURST 5.4. Results and Discussion 5.4.1. Comparison with observations On analyzing the daily precipitation output, it is observed that high intensity precipitation is observed on 14Sep2012 for model simulation with explicit physics as well as the corresponding daily merged TRMM dataset, which matches with the reported date of the event. Model simulated 6 hourly accumulated precipitation for 13Sep2012 1800UTC for the 3 km resolution is given in Fig. 5-3 with both model simulations and the corresponding TRMM observations. On comparison of the 6 hourly precipitation patterns, peak rainfall intensity is captured over the Ukhimath region between 13Sep2012 1800UTC and 14Sep2012 0000UTC by the model with explicit physics simulation as well as TRMM dataset. But the TRMM 3B42 V7 dataset shows a relative error of 8.6 mm/hr over Ukhimath at 13Sep2012 1800UTC. The precipitation peak is displaced south east of Ukhimath in the model simulation with explicit physics, comparable to TRMM observations. This location of peak precipitation (at 30°19’49’’N latitude and 79°28’09’’E longitude) is marked in the Fig. 5-3a with a black circle. Due to this displacement, the detailed analysis of the event in this study is performed at this location. With the spatial extent of the event is fixed the temporal extent of the precipitation event is analyzed in detail in the Fig. 5-4 at the location of precipitation peak in the model simulation with explicit physics. For clarity the precipitation intensity is analyzed at the precipitation maxima grid point discussed above and eight surrounding grid points for both model simulations. Heavy precipitation is observed at 13Sep2012 1800UTC for model simulation with explicit physics, with the precipitation exceeding 75 mm. In the

166

CHAPTER 5 – CLOUDBURST

Fig. 5-3: 6 hr accumulated precipitation (mm) for 13Sep2012 1800UTC for model-simulated domain 3 with (a) explicit physics, (b) parameterized physics, and (c) TRMM observational analysis. Plus sign indicates location of Ukhimath. Black circle indicates the location of precipitation maxima (at 30°19’49’’N latitude and 79°28’09’’E longitude).

167

CHAPTER 5 – CLOUDBURST

Fig. 5-4: (a) Time series of 6 hr accumulated precipitation (mm) at location of precipitation maxima at 30°19’49’’N latitude and 79°28’09’’E longitude (red line) and 8 grid points surrounding it (grey lines) for model-simulated domain 3 with (a) explicit physics and (b) parameterized physics.

168

CHAPTER 5 – CLOUDBURST same figure it can be seen that model simulated precipitation shows low intensity precipitation in the earlier time periods also. This indicates formation of small but multiple storm cells over the region. A notable exception is seen in this pattern in model simulation with parameterized physics, showing consistent low values of precipitation on both 13 and 14Sep2012. Although the explicit physics simulation and observation data capture the precipitation pattern as seen in the station data (Fig. 5-1a), the intensity of rainfall is not reproduced but is comparable to TRMM output of 70mm. The parameterized physics simulation is not able to capture the precipitation maxima over the Ukhimath region due to the storm. This difference illustrates that explicitly resolving convection at finer resolutions simulates the storm conditions better than when the convection is parameterized. The peak precipitation amount is captured better by the finer model resolution (domain 3) than the coarser resolutions (domain 1 and 2; figures not shown). Under representation of station observation or resolution difference between model simulation and observation analysis can be the cause of the variation in the magnitude of precipitation. The enhanced impact of orography on precipitation is seen clearly in finer resolutions due to better representation of topography. However, low density of observation stations cannot capture the maxima spatially or variable patterns of convective precipitation over this region of variable orography due to underrepresentation of subgrid features in coarser resolutions (Dimri and Niyogi, 2013). Though exact location of precipitation maxima is elusive even in the observations and needs further investigations. According to the TRMM analysis, the rainfall maxima occurred around 13Sep2012 1800UTC.

The corresponding time of “cloudburst” reported around

169

CHAPTER 5 – CLOUDBURST 13Sep2012 1930UTC or night of 13Sep2012 by local reports (Sphere India, 2012; DMMC, 2012). This reported time of the event matches well with the model simulation with explicit physics. The storm’s maximum intensity was observed around 13Sep2012 1930UTC as per the DWR data as reported by the IMD. This is also corroborated from the information available from Meteosat-5 IR-Tb data shown in Fig. 5-1c-d; where it is seen that the storm intensification is observed on 13Sep2012 1800UTC and strength of the intensity reduces on 2200UTC. Romatschke and Houze Jr. (2010, 2011) and Barros et al. (2004) reported occurrence of such small but strong precipitating convective systems over the Western Himalayan region which have a diurnal cycle with peaks in night and afternoons. 5.4.2. Analysis of the atmospheric processes As discussed in the above section, the precipitation maxima is observed at 13Sep2012 1800UTC. In the following section the various atmospheric processes associated with the weather event are discussed in detail, with a specific emphasis on the period around 13Sep2012 1800UTC. Large-scale synoptic flow at 850hPa is shown in Fig. 5-5a, with the corresponding geopotential height at 27 km resolution. Due to high topography, the model does not simulate variables over the Ukhimath region. But the figure is discussed to elaborate on the monsoonal flow seen during this time period. The figure depicts low pressure belt over the Indo-Gangetic plains. This zone corresponds to the monsoon trough zone that developed over Indian region. This monsoon trough with the well marked low has been reported over the same region by the IMD (2012). ISM generates required convection over the Indian region and causes precipitation along the monsoon trough zone.

170

CHAPTER 5 – CLOUDBURST

Fig. 5-5: (a) Wind circulation (m/s; vector) at 850hPa and geopotential height (×10-2 m; shaded) at 27 km resolution for 13Sep2012 1800UTC, (b) same as (a) but at 700hPa and (c) same as (b) but with vertical integrated moisture transport (kg/m/s; vector) and flux (×10-3 mm; shaded). Plus sign indicates location of Ukhimath. 171

CHAPTER 5 – CLOUDBURST Fig. 5-5b depicts the large flow at700hPa, showing a clear low pressure zone formed over the Ukhimath region. This low pressure zone develops strong wind flow along the Himalayas. In the fringe of this flow, the flow is southeasterly in the Rudraprayag region. Fig. 5-5c shows the spatial distribution of vertically integrated moisture flux (transport) and divergence over 27 km resolution. There is strong moisture convergence over southeast of Ukhimath, in form of small cells of moisture convergence. In the simulation, cellular spatial distribution of the moisture convergence zone is seen. The convergence peak and spatial distribution is similar to the corresponding precipitation maxima and distribution. The moisture convergence zone is accumulated at the windward side of the Himalayan slope and the moisture influx is from the IndoGangetic belt towards the Himalayas. The moisture-laden flow from the Arabian Sea causes the convective storm to intensify. This intensification instability over region is due to the buoyancy produced due to the moisture incursion which causes the cloud formation. Upper level wind circulation at 200hPa for 13Sep2012 1800UTC at 3 km model resolution is shown for explicit physics simulation (Fig. 5-6a) and parameterized physics simulation (Fig. 5-6b). Corresponding outgoing longwave radiation (OLR) shown in Fig. 5-6c-d, has been used as an index to depict the clouding condition during the event. Treating convection explicitly in the model, a strong upper-level 200hPa divergence is visible, indicating a lower-level convergence corresponding to the intense precipitation event (Fig. 5-6a). The circulation intensification begins from 13Sep2012 1200UTC onwards in model simulation with explicit physics (figure not depicted). Similarly, Fig. 5-6c depicts strong cloud formation as shown by reduced OLR on 13Sep2012 1800UTC

172

CHAPTER 5 – CLOUDBURST

Fig. 5-6: Wind circulation (m/s; vector) at 200hPa on 13Sep2012 1800UTC at 3 km resolution (a) with explicit physics, and (b) with parameterized physics. Outgoing longwave radiation (W/m2; shaded) on 13Sep2012 1800UTC at 3 km resolution (c) with explicit physics, and (d) with parameterized physics. Plus sign indicates location of Ukhimath.

17...