National Centre for Medium Range Weather Forecasting

About: National Centre for Medium Range Weather Forecasting is a government organization based out in Noida, India. It is known for research contribution in the topics: Monsoon & Weather Research and Forecasting Model. The organization has 176 authors who have published 368 publications receiving 4749 citations.

Performance of NCMRWF Model TC Track Forecasts During 2013

Abstract: There are two tropical cyclone (TC) seasons over the North Indian Ocean (NIO), (including the Bay of Bengal (BOB) and the Arabian Sea (AS)), i.e. during the pre-monsoon months (April–early June) and the post-monsoon months (October–December) (Mohanty et al., Mar Geod 33:294–314, 2010). Further the Indian subcontinent happens to be one of the world’s highly vulnerable areas since the coastal population density is very high leading to an extensive damage to life and property. Therefore, forecasting of TC track and landfall location is critical for early warnings and mitigation of disaster. Track forecast errors over the NIO though improved significantly in recent years (Mohapatra et al., J Earth Syst Sci 122:589–601, 2013, J Earth Syst Sci 124:861–874. doi: 10.1007/s12040-015-0581-x, 2015) are still high relative to those over the Atlantic and Pacific Oceans. With advancements in computational power, development of better NWP models (both global and regional), the forecasting capability of meteorologists have greatly increased. Several meteorological centers like NCEP, UKMet office, ECMWF, JMA, JTWC etc give a real time forecast of TC tracks from their global NWP models (deterministic as well as Ensemble Prediction Systems (EPS)) (Hamill et al. Mon Weather Rev 139:3243–3247, 2011; Froude et al. Mon Weather Rev 135:2545–2567, 2007; Buckingham et al. Weather Forecast 25:1736–1754, 2010; Heming et al. Meteorol Appl 2:171–184, 1995; Heming and Radford Mon Weather Rev 126:1323–1331, 1998). TC track prediction from an ensemble forecasting system besides providing a track from each ensemble member also provides the strike probability (Weber Mon Weather Rev 133:1840–1852, 2005). For the TCs of NIO, Mohapatra et al. (J Earth Syst Sci 122:589–601, 2013, J Earth Syst Sci 124:861–874. doi: 10.1007/s12040-015-0581-x, 2015) provided a detailed verification of the official forecast tracks and its improvements in the recent past. This study provides a detailed verification of the NCMRWF NWP model forecasts of 2013 TC cases. Some of the earlier studies (Ashrit et al. Improved track and intensity predictions using TC bogusing and regional assimilation. In: Mohanty UC, Mohapatra M, Singh OP, Bandyopadhyay BK, Rathore LS (eds) Monitoring and prediction of TCs in the Indian ocean and climate change, Springer, Dordrecht, p 246–254, 2014; Chourasia et al. Mausam 64:135–148, 2013 and Mohandas and Ashrit Nat hazard 73:213–235, 2014) focused on the NCMRWF model TC forecasts and the impact of bogusing, assimilation and cumulus parameterisation etc. The present study is focused on the real time operational forecasts provided to India Meteorological Department (IMD). During May–December 2013, there were five TCs observed in the Bay of Bengal namely: Viyaru (May10–17), Phailin (October 4–14), Helen (November 19–23), Lehar (November 19–28) and Madi (December 6–13). This report summarises the performance of the real time prediction of these TC tracks by the NCMRWF Global Forecast Systems.

Validation and assimilation of INSAT atmospheric motion vectors: Case studies for tropical cyclones

Abstract: INSAT-3DR is the latest geostationary satellite launched by the Indian Space Research Organization (ISRO) as a continuation to the INSAT-3D, for enhanced meteorological observations. National Centre for Medium Range Weather Forecasting (NCMRWF) receives INSAT-3DR Atmospheric Motion Vectors (AMVs) through Global Telecommunication System (GTS) along with the AMVs from other satellites. The INSAT-3DR AMVs are validated against the in-situ observations for a period of 3 months, May–July 2020. The validation results are compared with the AMVs from other satellites like INSAT-3D and Meteosat-8 located over the same geographical area and found that the quality of INSAT-3DR AMVs is comparable. After the successful validation, INSAT-3DR AMVs are assimilated in the NCMRWF Global Forecast System (NGFS) for two cyclone cases, formed during May–June 2020 over the North Indian Ocean. Four Observation System Experiments (OSEs) are designed, with the assimilation of individual and combined AMVs from INSAT (3D and 3DR) and Meteosat-8, to see the impact of AMVs during the cyclones Amphan formed over the Bay of Bengal and Nisarga formed over the Arabian Sea. In general, assimilation of AMVs improved the simulation of both the cyclones Amphan and Nisarga formed during May–June 2020. Introduction of INSAT AMVs slowed down the otherwise fast-moving cyclone Amphan simulated due to the assimilation of Meteosat-8 AMVs. Both intensity and track of the cyclones Amphan and Nisarga are better simulated when the AMVs from INSAT and Meteosat-8 are assimilated together.

Role of Space-Borne Sea Surface Winds on the Simulation of Tropical Cyclones Over the Indian Seas

Abstract: The impact of space-based sea surface winds on the simulation of tropical cyclones that formed over two distinct basins of the northern Indian Ocean during October–November 2019 is investigated in this study. Observing system experiments (OSEs) using the National Centre for Medium Range Weather Forecasting (NCMRWF) Unified Model (NCUM) assimilation and forecast system were designed to simulate the characteristics of the cyclones “Kyarr”, which formed over the Arabian Sea (24 October to 3 November 2019), and “Bulbul” over the Bay of Bengal (4–11 November 2019) by including the sea surface winds derived from the scatterometer on board the MetOp-A, MetOp-B, ScatSat-1 and microwave radiometer aboard WindSat satellites. Approximately 3% of the sea surface winds received were assimilated in the NCUM system during both cyclone cases, and the two MetOp scatterometers contributed the most significant fraction. Assimilation of sea surface winds improved the cyclone characteristics in the analysis compared to a baseline experiment, which denied all the sea surface winds. Sea surface wind assimilation improved the surface wind analysis by 2–7% and reduced the root mean square differences by ~ 10% when compared against ERA5 reanalysis. Better simulation of cyclone track in the higher lead time suggests that the sea surface wind information is critical in the analysis during the cyclone's initial stage. The estimated track and intensity errors were larger for the Kyarr than the Bulbul cyclone. This could be due to the nature of the cyclones, Kyarr being a super cyclone which dissipated over the ocean, whereas Bulbul was a severe cyclone that dissipated after landfall. An improvement in the post-landfall track of Bulbul due to the assimilation of sea surface winds is also noted.

Phase inconsistency as a major source of error in NGFS forecast

Abstract: South Asian monsoon exhibits multiscale spatiotemporal variability. Analyzing the nature and behavior of numerical weather forecast error associated with these space-time heterogeneities will eventually help in improving the models. We investigate the spatiotemporal error characteristics of the National Centre for Medium-Range Weather Forecasting (NCMRWF) Global Forecast System (NGFS) model over South Asian land and ocean separately. Although error grows with lead-time, it saturates within 3–5 days of forecast initiation. The saturated error is only about 15–25% higher than that of day-1, indicating that most of the error accumulates within first 24-h of forecast. Increase in error over oceanic regions is due to an increase in the area with high error at all precipitation ranges with large day-to-day variability. However, over land error growth is primarily confined at locations of high mean precipitation. Decomposition of error arising due to intensity and phase variations reveals that about 90% of it arises from the model’s inability to capture phase of precipitation at various timescales. We show that NGFS cannot capture synoptic scale variations ($$<10$$ day) after day-2. Both the high-frequency (10–20 day) and low-frequency (30–60 day) intraseasonal variations are reasonably predicted up to day-3. At diurnal timescale, NGFS forecasts show a peak in precipitation about 3–6 h prior to that observed, both over land and ocean. Surprisingly, this error does not change with lead-time. Lastly, we show that major error characteristics do not depend on the seasonal mean monsoon rainfall.

Improved Prediction of Cyclone Phailin (9-12 October 2013) with 4DVAR Assimilation

Abstract: This correspondence is a brief summary on the verification of National Centre for Medium Range Weather Forecasting (NCMRWF) model forecasts during the recent VSCS (very severe cyclonic storm) Phailin (9–12 October 2013). The study presents quantitative verification of forecast tracks and rainfall (after landfall). The real-time operational tropical cyclone (TC) track forecasts are based on NCMRWF Global Forecast System (NGFS) (T574L64) and NCMRWF Unified Model (NCUM). Both of these are state-of-the-art modelling systems with advanced parameterization schemes for sub-grid scale physical processes. Description of the model configurations implemented at NCMRWF can be found in Prasad et al. for NGFS model and Rajagopal et al. for NCUM model. The NGFS model uses a TC relocation algorithm for realistic representation of the location of the cyclone position in the initial conditions. In NCUM, the initial position of the cyclone is captured in the analysis through four-dimensional variational data assimilation (4DVAR). The 4DVAR assimilation system prepares the analysis at observation time and implicitly generates flow-dependent background errors. The NCUM is able to spin-up the TC without initialization, which can be attributed to combined impacts of improved model configuration, better data coverage (such as satellite data) and data assimilation technique. The data assimilated into the models include the special upper air observations from Visakhapatnam (0600 and 1800 UTC of 11 and 12 October 2013) and Regional ATOVS Retransmission Service (RARS). The verification presented in this study shows the improved track prediction and rainfall forecast (amount and distribution after landfall) in NCUM due to 4DVAR, demonstrating the importance of assimilation of observations at the most appropriate time. Tracking of the TCs in the forecasts uses the TC Vital Database (‘tcvitals’). The tcsvitals is an archive of Cyclone Message Files, which contain information such as cyclone location, intensity, horizontal wind and pressure structure, and depth of convection, created in real time by forecasting centres. These vitals are also used during the vortex relocation and bogusing process in the NGFS (ref. 3). The ‘tcvitals’ generated by the Joint Typhoon Warning Centre (JTWC) is used in this study for relocation as well as verification of the predicted cyclone positions. The TC forecast tracks are derived based on vertical weighted average of the maximum or minimum of several parameters in the vicinity of a vortex in the input first guess and forecasts. A detailed account of the tracking algorithm is presented in Marchok. Briefly, for TC, seven parameters are tracked, including the relative vorticity maximum, geopotential height minimum and wind speed minimum at both 850 and 700 hPa, as well as the minimum in sea-level pressure. The locations based on these parameters are averaged together to provide an average cyclone position at each forecast hour. In order to avoid tracking weak, transient disturbances (either real or artifacts of model noise), two constraints have been added: (1) the storm must live for at least 24 h within a forecast and (2) the storm must maintain a closed mean sea level pressure (MSLP) contour, using a 2 hPa contour interval. Figure 1 a shows the cyclone tracks based on observed positions and positions of the cyclone in the initial conditions from 9 to 12 October 2013. NGFS model features an average initial position error of about 46 km, while the average initial position error in NCUM is 28 km. With initial error of 83 km in NGFS and 76 km in NCUM, both models have the highest initial position errors on