weather modelling, gis, remote sensing in the mountains, north india, army camp, army base, snow capped mountains

Weather Modelling

By: Swati Basu and E N Rajagopal
A seamless approach across spatial and temporal scales using the earth system modelling framework is the key for providing realistic prediction of weather and climate. Ensemble based forecasting and data assimilation techniques are being used for estimating the uncertainty and providing probabilistic forecasts to end users.
Magazine Articles Technology

Weather and climate information play an important role in shaping the country’s economic and strategic activities. India is emerging as global player in various sectors and it has resulted in the increase in demands for accurate weather and climate prediction for applications in various sectors. Weather forecasting research in data assimilation and, weather and climate modelling (short, medium and seasonal) using computer-based Numerical Weather Prediction (NWP) models is a well-established practice now, which is being spearheaded in India by the National Centre for Medium Range Weather Forecasting (NCMRWF). During the last two decades, the accuracy of weather prediction has improved significantly, with better understanding of the underlying physical and dynamic processes, quality and quantity of meteorological observations; especially from satellites and availability of progressively increasing computing power.

Atmospheric modelling

The concept of a unified modelling system for seamless prediction of weather and climate has gained importance and acceptance in recent times. An unified model (UM) and the associated 4D-VAR (four-dimensional variational) data assimilation system have been successfully implemented at NCMRWF. A regional version of the UM has also been implemented and tested for few synoptic cases. More detailed short-range forecasts are provided by this high-resolution regional version which has a more detailed representation of orography and is able to represent certain atmospheric processes more accurately.

Verification of model forecasts

The last eight years record (2005-2013) of the root-mean-square error (RMSE) against observations from Indian Radiosonde stations of 3-day forecasts shows that the prediction skills have improved between 7 and 8 per cent. This improvement resulted from periodic increase in the resolution of the model and the capability to assimilate satellite radiances.

Atmospheric observations

Atmospheric observations are quite heterogeneous in terms of their horizontal, vertical and temporal resolution. On the other hand, NWP models need a regularly spaced and balanced initial condition of the atmosphere to start time integration. Process of converting the information in observations to the initial condition required for NWP models is known as data assimilation (DA). Only observations that qualify stringent quality control checks are accepted by the DA system.

Utilisation of satellite observations

The science of meteorology and the practice of weather forecasting has immensely improved over the years owing to the enhanced ability of the global remote sensing community to observe the 3-D atmosphere. This is significant in view of the limited scope for expansion of conventional data network to meet the input data requirements. As India is surrounded by data sparse oceanic region, there is strong motivation to make optimum use of data from remote sensing observational platforms. The various types of satellite observations/sensors that are used at NCMRWF are described below.

Atmospheric motion vectors: Geostationary satellites orbit earth in a geosynchronous orbit (~36,000 km above earth) with the orbital period same as the earth’s rotation period. Geostationary satellites give the continuous observation of specific area above the earth. Information about the upper air wind is obtained by tracking the displacement of clouds, water vapour, etc., by an automatic pattern recognition technique. This procedure has also been applied to generate data from polar-orbiting satellites (a sun synchronous orbit ~1000 km above earth). This is popularly known as atmospheric motion vectors and widely used in numerical weather prediction.

Ocean surface scatterometer: Satellite sensors operating at microwave frequencies can make measurements of the ocean surface under nearly all-weather conditions. Scatterometers are radar instruments that measure the radar backscatter from a part of the ocean surface, which depends on wind speed, wind direction, and observation geometry. Oceansat-2, which is an Indian Remote Sensing satellite dedicated to ocean research, has a scatterometer on-board.

Satellite Radiances: Satellites do not measure temperature and moisture directly –they measure radiances (amount of radiation emitted into space from different atmospheric layers) in various wave length bands. Vertical profiles of temperature and moisture can be obtained from radiances sensed by different channels using a mathematical model. However modern day NWP systems can assimilate radiances directly instead of using the temperature and moisture profiles.

Global Positioning System Radio Occultation (GPSRO) from Low Earth Orbit satellites: The GPSRO technique involves a low earth orbit (~2000 km above earth) satellite receiving a signal from a GPS satellite. The signal has to pass through the atmosphere and gets refracted along the way, and the magnitude of refraction depends on the temperature and moisture content in the atmosphere.

Megha-Tropiques: The Indo-French Megha-Tropiques satellite’s payload consists of MADRAS (Microwave Analysis and Detection of Rain and Atmospheric Systems – a microwave imager to measure precipitation and cloud properties), SAPHIR (a 6 channel microwave sounding instrument near the absorption band of water vapour at 183 GHz for the retrieval of water vapour vertical profile and horizontal distribution), and ScaRaB (Scanner for Radiation Budget – an optical scanning radiometer devoted to the measurements of radiative fluxes at the top-of-atmosphere in the shortwave and longwave domain).

Global data assimilation system

Data assimilation systems currently used are either in 3D or 4D variational schemes. Observations from all over the globe are assimilated four times a day viz., 00, 06, 12 and 18 UTC. The assimilation scheme utilizes all data collected within ±3 hours of the assimilation time and received within a specified cut-off period.

Recent studies have shown that Ensemble Kalman Filter (EnKF) technique has an advantage in defining flow dependent errors and offers more flexible treatment for model errors aiding a ‘hybrid data assimilation’ which has been adopted by many leading global NWP centres. NCMRWF too plans to shift to hybrid data assimilation soon.

 

Global ensemble forecast system

Weather forecasts’ uncertainty is due to errors in the initial conditions, and model approximations. Together, they limit the skill of a deterministic forecast system. Ensemble forecasting has emerged as the practical way of estimating the forecast uncertainty and making probabilistic forecasts. A Global Ensemble Forecasting System (GEFS) has been implemented at NCMRWF and is running in real time.

Probabilistic Quantitative Precipitation Forecast (PQPF): Ensemble forecast systems allow for prediction of the rainfall in probabilistic terms. Fig 1. shows an example of the PQPF plots depicting the forecast rainfall distribution alongside the spatial distribution of rainfall probabilities in the 1-2 cm/day, 2-5 cm/day and 5-10 cm/day categories. The top left panel shows the mean rainfall in the Day-6 forecast valid for 8th July 2013.

Fig 1
Fig 1: Forecast Vs actual rainfall for 6th August 2013

Ensemble Based Tropical Cyclone Track Forecasts: Ensemble forecast system can be used for forecasting the tropical cyclone track. This module has been implemented at NCMRWF for the first time and is used for tracking the cyclones in Bay of Bengal, Arabian Sea, Indian Ocean and other oceanic regions as well. Fig 2 shows the tracks and strike probabilities forecast for the cyclone ‘Nilam’ on October 31, 2012.

 

Coupled ocean-atmosphere modelling

Monsoon rainfall prediction is crucial for agriculture, food security and economy of India. Ocean, atmosphere and land-surface all interact and play dominant roles simultaneously in the genesis of the monsoon. Only atmosphere model is incapable of capturing the variability – and coupled ocean-atmosphere models require higher computing resources and relevant observations from ocean, land-surface and atmosphere to be assimilated to the modelling system of the earth. In last decade in developed nations due to the availability of computer resources and enhanced earth observations, it became possible to carry out the data assimilation and model runs for coupled system which has shown good potential to capture the monsoon and tropical variability. These coupled models are also being developed with an aim to target seamless prediction suite where the same modelling frame work could be used for both weather (short and medium range) and short-term climate (month and season).

Fig 2: The tracks (left: ensemble members) and strike probabilities (right) forecast for the cyclone 'Nilam' on 31st Oct 2012. The forecast track errors suggest that the GEFS model track errors are significantly lower at all lead times, compared to the deterministic NCMRWF Global Forecast System.
Fig 2: The tracks (left: ensemble members) and strike probabilities (right) forecast for the cyclone ‘Nilam’ on 31st Oct 2012. The forecast track errors suggest that the GEFS model track errors are significantly lower at all lead times, compared to the deterministic NCMRWF Global Forecast System.

Fig 2 Part 2

Future plans

As a centre of excellence in weather and climate modelling, NCMRWF will focus on model development in a seamless frame work across scales. Land-surface, ocean and cryosphere of the earth system will be dealt with more sophistication including proper initialisation by their respective data assimilation systems. The global/regional models and data assimilation systems will be continuously improved and customised for weather/climate systems over Afro-Asian region. The chemical components and dust/aerosol aspects are also being taken up to examine their impact in weather/climate systems over the Indian region. India’s capability to launch satellites is an added advantage, and NCMRWF will continue to use advanced techniques to assimilate various remote sensed data into its modelling system. With the recent developments in parallel processing in computing, cloud computing and networking, NCMRWF is also gearing up to use massively parallel computers to take up cutting edge high resolution modelling and data assimilation. Development of novel applications of weather/climate prediction for different sectors like agriculture, water resources, flood/drought management, wind energy, energy management, environment, disaster mitigation, etc. are equally important and will be carried out at NCMRWF in collaboration with end users.

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *