Table 2 Description of the model configuration used Table 1 List of episode selection challenging over the region (Hatwar et al. 2005). Studies on Western Disturbances over North West India using numerical models are limited. Azadi et al. (2001) used NCAR high resolution mesoscale model to simulate active western disturbance during 1997. Dimri et al. (2004) used a mesoscale MM5 model to simulate heavy precipitation associated with an intense WD over Western Himalayas during 21-25 January 1999 when a WD affected NW India. Hatwar et al. (2005) and Dimri and Mohanty (2009) carried out prediction of western disturbance and associated weather systems over Western Himalayas using the IMD’s operational limited area analysis and forecast system and MMS model respectively. Table 3 Microphysics schemes used with (a) combination of other parameterization schemes in 1-5 experiments and (b) summary Model, data, experimental design and evaluation methodology compressible non-hydrostatic equations with hydrostatic option and mass-based terrain following coordinate. The horizontal grid is the Arakawa-C grid. The model can simulate both idealized and real-data cases with various lateral boundary condition options. The model supports one way, two-way, and moving nest options (Skamarock et al. 2008). A brief description of the model configuration and the different nested domains used in this study is given in Table 2 and Fig. 1 respectively. In the present study, three nested domains with resolutions of 81, 27 and 9 km cover northwest and north India. Domain | is the mother domain and Model used The model used in this study is the Weather Research and Forecasting (WRF) model (Version v3.5.1) which is a widely used community mesoscale model. This version of the WRF model uses the Eulerian mass coordinate and is referred to as the Advanced Research WRF (ARW) developed by the National Center for Atmospheric Research (NCAR). It uses the fully and north India. Domain | is the mother domain and Bold indicates the lowest root mean square error in the column Table 4 RMSE and RSR in bracket based on the model forecasts of precipitation (mm/day) (PREC), 500 hpa geopotential height (m) (GPH) an 500 hPa wind (m/s) (CIRC) from different experiments (physics configurations) for (a) WD case 1, 2 and (b) WD case 3, 4 Fig. 2 Daily rainfalls (mm/day) based on a—d IMD’s and e-h model simulated for 15th January 2002, 16th January 2002 of WD case 1 and 7th February 2002, 8th February 2002 of WD case 2 respectively Root mean square error is used to examine model sen- sitivity. It is mathematically defined, (Singh et al. 2004; Wilks 2006) as Root mean square error is used to examine model sen- Results and discussion Sensitivity of the model forecast and to identify best physics options Fig. 3 Daily rainfalls (mm/day) based on a—d IMD’s and e-h model simulated for 18th January 2013, 19th January 2013 of WD case 3 and Sth February 2013, 6th February 2013 of WD case 4 respectively Fig. 3. continued On 15th January 2002, the observed rainfall distribu- tion (see Fig. 2a) shows widespread rainfall over many places over NW India and the model simulation also shows almost the same pattern which can be seen in the observed rainfall (see Fig. 2a, e). On 16th January 2002, the observed rainfall distribution shows, the maxima of precipitation has shifted towards South-East direction (see Fig. 2b). The model could produce the maxima of pre- cipitation which can be seen in the observed rainfall (see Fig. 2b, f). On 7th February 2002, moderate rainfall in the range 1-2 cm over some parts of Western Himalayan region were observed (see Fig. 2c). Model could produce rainfall of the same range over the same region as seen in Figure 3a—d shows daily rainfalls (mm/day) for 18th and 19th January 2013, of WD case 3 and 5th and 6th February 2013 of WD case 4 respectively based on IMD’s gridded rainfall data and Fig. 3(e-f) shows the model simulated rainfall. On 18th January 2013, the observed rainfall distribution (Fig. 3a) shows a belt of moderate to heavy rainfall over the Western Himalayan region with peak centered near lat 32°N and 79°E (Approximately 10 cm) was observed. Figure 3e presents the model forecast for same period. The model could produce moderate to heavy rainfall over Western Himalayan region with peak centered near lat 32°N and 77°E (Ap- proximately 10 cm). The model simulated spatial distri- bution pattern shows widespread light to moderate rain over parts of North West India, which can be clearly discernible in the observed rainfall (see Fig. 3a, e). On 19th January 2013, again Precipitation belt (1-6 cm) was confined mainly over the Western Himalayan region. Also light to moderate rain was observed over parts of eastern India (see Fig. 3b). From Fig. 3f, it is seen that, the model could produce the precipitation belt over the Western Himalayan region, light to moderate rain is seen over Himalayan region, light to moderate rain is seen over parts of eastern India. On 5th February 2013, Precipitation belt (2-9 cm) was confined mainly over the Western Himalayan region with peak centered near lat 33°N and 75°E (>9 cm) light to rather heavy rain over parts of North West India can also be seen (see Fig. 3c). The model simulation could also produce the Precipitation belt over the same region, but over predicted the rainfall over some parts of the Western Himalayan region (see Fig. 3c, g). On 6th February 2013, heavy rainfall persisted over the Western Himalayan region (see Fig. 3d). From Fig. 3h, it is seen that the model could produce the Pre- cipitation belt over same region. Figure 4a—d presents 500 hPa geopotential height (m, contour) and wind (m/s, arrow) based on Era-interim data and region with wind speed more than 25 m/s is depicted in grey shade. Fig- ure 4e—-h shows the model simulated values for 15th and 16th January 2002 of WD case | and 7th and 8th February 2002, of WD case 2 respectively. Geopotential height approximates the actual height of a pressure sur- face above mean sea-level. Since cold air is denser than warm air, it causes pressure surfaces to be lower in colder air masses, while less dense, warmer air allows the pressure surfaces to be higher (Stull 2012). Thus, heights are lower in cold air masses, and higher in warm air masses. A line drawn in the presented figures connecting points of equal height (in meters) is called a height contour. That means, at every point along a given Figure 5a—d presents 500 hPa geopotential height (m, contour) and wind (m/s, arrow) based on Era-interim data and region with wind speed more than 25 m/s is depicted in contour, the values of geopotential height are same. Geopotential height is valuable for locating troughs and ridges which are the upper level counterparts of surface cyclones and anticyclones (Daniel et al 1997). From Fig. 4a, it is seen that on 15th January 2002, a trough lay along longitude about 66°E. Model simulation could produce the trough along longitude about 68°E (see Fig. 4e). Figure 4b shows the trough has deepened and moved rapidly eastwards on 16th January 2002 and lay along longitude about 72°E. Model simulation could produce the trough at the same position (see Fig. 4f). From Fig. 4c, it is seen that on 7th February 2002, the associated cyclonic circulation is over 30°N to 34°N and 68°E to 72°E region. The Model could capture the cyclonic circulation over the same region (see Fig. 4g). The model simulation shows that on 8th February 2002, the cyclonic circulation moved eastward (see Fig. 4h) which can also be seen in observed analysis (see Fig. 4d). Fig. 5 continued grey shade. Figure 5e—h shows the model simulated values for 18th and 19th January 2013 of WD case 3 and 5th and 6th February 2013 of WD case 4 respectively. On 18th January 2013, well intensified trough lay along longitude about 70°E (see Fig. 5a). Model simulation could produce the trough at the same position (see Fig. 5e). The system became less marked on 19th January 2013 (see Fig. 5b) and model simulation could produce the same synoptic conditions (see Fig. 5f). From Fig. 5c, it is seen that the cyclonic circulation is over 30°N to 34°N and 68°E to 72°E region on 6th February 2013. The cyclonic circulation moved eastward and it became less marked on 7th Febru- ary 2013 (Fig. 5d). The Model could produce the same synoptic conditions associated with the system (see Fig. 5c-f). During all the WDs, the model simulated geopotential height shows good accuracy in predicting low pressure area over the region. Figure 6a—h presents the forecast error (model-observation) in 500 hpa wind (m/s, arrow) and the region with bias (forecast error) more than 5 m/s depicted in grey shade for 15th and 16th January 2002 (of WD case 1), 7th and 8th February 2002 (of WD case 2), 18th and 19th January 2013 (of WD case 3) and 5th and 6th February 2013 (of WD case 4) respectively. From Fig. 6a—h it is observed that, the model simulated wind pattern agrees very closely with the observation analysis. However, over western Himalayan region it shows a south westerly bias. The characteristics of the various parameters associated with the WDs have been thoroug the intensity and movement of thi and circulation over the region model. However, it shows a hly studied. It is found that e€ precipitation distribution is well predicted by the bias over the Himalayan region. During all the WDs, the model simulated geopo- tential height shows good accuracy in predicting low pressure areas over the region. 1 [The model simulated wind pattern agrees very closely with the observation analysis. However, over western Himalay an region it shows a south Fig. 6 continued
