The Indian Institute of Tropical Meteorology (IITM), Pune is striving to advance prediction of monsoon weather and climate change seamlessly across different scales. Under the umbrella of National Monsoon Mission, a programme of Ministry of Earth Sciences (MoES), IITM made landmark achievement in monsoon forecasting and research. Monsoon Mission also brought working partnership between the academic, research and development organisations, both national and international and the MoES to improve the monsoon forecast skill across the country. Through this path-breaking Mission, IITM has setup a state-of-the-art dynamical modelling framework for improving skills that predict seasonal and extended as well as short and medium range monsoons up to two weeks. The improved models are passed on to India Meteorological Department (IMD) for their operational use.
Short-range weather forecasting system
With the current advancement in weather sciences in India, it would become increasingly difficult to fault the weatherman. The newly developed path-breaking short-range weather forecasting system will not only bring India at par with many developed nations in terms of weather prediction, but also assist well-informed decision-making.
P Mukhopadhyay, a senior scientist at the IITM explains how the newly adopted Global Forecast System (GFS), a model produced by the National Center for Environmental Prediction (NCEP) has helped better forecasting. The GFS models have assisted the establishment of short range global ensemble forecast system (GEFS), foremost amongst which is a high resolution (12 km) deterministic weather forecast in the country, named GFS T1534.
The GFS T1534 forecasting deterministic model was configured at the IITM for an experimental forecast in the monsoon season of 2016. It showed significant improvement in daily weather forecasts. Based on the performance of GFS T1534—a deterministic model, (IITM is yet to run the ensemble models in this resolution) for recent extreme events such as tropical cyclones, cold waves and more, the MoES decided to make it operational in IMD since January 16, 2017. This high resolution model replaced the earlier low resolution (25 km) operational model of IMD for deterministic forecast. Apart from this, the other experimental model that is due to be handed over to IMD in the later part of the year are the GEFS T574 (21 ensemble model running at a resolution of 25-27 km) that forecasts percentage probability of rainfall, temperature etc.
The initial conditions for the high-resolution model are being generated daily by ESSO National Center for Medium Range Weather Forecast (NCMRWF), Noida and the forecast for next 10 days are being generated at the IITM.
In the ensemble prediction system the numerical model provides the percentage probability of the most likely state of the atmosphere in the forecast. This is particularly meaningful as the numerical representation of real atmosphere vis-à-vis the weather forecast model or numerical weather prediction (NWP) models have inherent uncertainties arising from physical parameterisation and initial condition etc.
Winds of change
Dynamical extended range sub-seasonal forecasting
Dam and flood management systems in the country are a huge beneficiary of such forecasting. States can effectively plan release and store regimes of dam water, which would help prevent potential flood scenarios as well as help plan dry seasons. Similarly, the prediction of heat waves and cold waves well in advance with a warning of over 30 days, would help civil administration to manage the supply of rescue and relief materials.
The extended range forecast products are indeed pertinent to guide crop sowing patterns in real-time, based on the forecast of rainy and dry spells across the Indian region.
The IITM uses the National Center for Environmental Prediction (NCEP), USA, climate forecast system version2 model (CFSv2)—an ocean-atmosphere coupled dynamical model, configured to run in two horizontal resolutions (~38km and ~110km). The initial conditions to run the model are generated through a perturbation technique developed in-house using NCEP GODAS data assimilation system. From every initial condition, the model is run for the next 30 days.
Large scale floods during monsoon are a part of sub-seasonal variability. However, extreme events during the monsoon are hard to predict due to chaotic and non-linear nature of the monsoon system. In 2013, the northward progression phase of Indian summer monsoon after its onset over southern tip of the country was very fast. Heavy rainfall was experienced over the northern state of Uttarakhand, on June 16-17, almost in tandem with the rapid advancement of the 2013 monsoon. A K Sahai, a senior scientist, IITM and his team investigated the scientific rationale behind the two events and conducted forecast experiments using the CFSv2 model. They hypothesised that both events were triggered by the interaction between monsoonal flow and the mid-latitude weather systems over the orographic region. The observed rainfall over 200 by 200 km area surrounding Kedarnath Dham was above 100 mm on 17 June and the model was able to capture a large part of it almost two weeks in advance.
These phases are important to understand the starting and ending spells of Indian summer monsoon. For the last four/five years monsoon onset and withdrawal are regularly forecasted based on the extended range forecast products. In cases of early or late onset or early and late withdrawals, these models are potentially providing valuable inputs.
Sahai adds, “The country has now acquired the operational capability to model forecasts to predict such events. We are now able to meet the demand for prediction in the extended range or 2-3 weeks in several sectors depending on water resources, city planning, dam management and health management (e.g. protection against heat death) and more”.
Weather in monsoon
How the Indian summer monsoon will shape each year is the basic concern for India, since the country is highly dependent on rains for its agriculture yield, water resources and energy production. The IMD’s traditional empirical model for seasonal prediction of monsoon was limited in its skill and needed to be upgraded.
IITM selected the CFSv2 model from NCEP, which was using at a lower resolution of around 110 km for making seasonal predictions. The horizontal resolution of the model was enhanced to 38 km to make seasonal forecasts.
According to Suryachandra A Rao, the Project Director, Monsoon Mission and High Performance Computing (HPC), IITM opines that “with this effort, India became the first country in the world to implement high resolution seasonal prediction system”. “The CFSv2 model T382 has a better skill of around 0.55 for the seasonal prediction, far above systems worldwide”.
Along with the seasonal prediction of monsoon rainfall, the system is issuing high resolution seasonal temperature prediction for hot and cold waves over the country during summer and winter seasons as well.
The initial atmospheric conditions are provided by NCMRWF and ocean initial conditions are provided by the Indian National Center for Ocean Information Services (INCOIS). Every month the forecast is provided by the 10th day and the IMD uses them for issuing seasonal forecasts. The model was able to predict the drought events in recent period successfully and its ability to predict drought in 2015 was tested in operational mode.
Severe drought in 2015 was predicted accurately with four months lead-time while majority of the other leading climate centres worldwide were predicting normal monsoon at that time. The dynamical prediction system is now handed over to IMD for making it fully operational.
The stark reality of climate change can be observed in melting glaciers, rising sea levels, changing ecosystems and fatal heat waves. In order to tackle the unpredictability that comes along with long term climate change, the IITM under the Centre for Climate Change Research (CCCR) has developed an Earth System Model (ESM) for studying long-term climate change. IITM-ESM is appropriate for detection, attribution and projection of future climatic changes on global and regional scales.
As per R Krishnan, the Director IITM, the IITM-ESM is a powerful tool to understand interactive feedbacks among the different components of the earth system—the atmosphere, ocean, land surface, cryosphere, biosphere and external forcing (such as human influence) and predict long-term climate variability and change.
The IITM-ESM will be the first climate model from India that would be participating in the upcoming Coupled Model Intercomparison Programme Phase 6 (CMIP6) models for the next Intergovernmental Panel for Climate Change (IPCC) assessment in 2020.
The development of the IITM-ESM was achieved by transforming the CFSv2 seasonal prediction model to a climate model suitable for long-term climate studies.
The in-house development of the model was realised by incorporating a new ocean component (MOM4p1) in the CFSv2 global coupled model. The new ocean model incorporates more realistic ocean processes, leading to alleviation of the cold global temperature bias by nearly 0.8oC. Careful analysis of the IITM-ESM simulations also reveals that many aspects of the observed global and regional climate, the monsoon systems and their variability on different time-scales are well captured by the model. In addition, the oceanic component also provides a great opportunity to investigate the impacts of climate change on sea-level rise as well as marine primary productivity.
In the latest version (IITM ESM2.0), the net energy imbalance at the top of the atmosphere is less than 1 Wm-2 and compares well with state-of-the-art climate models world over. Following the CMIP6 protocols time-varying forcing elements, greenhouse gases (GHG), aerosol optical properties (both natural and anthropogenic aerosols) and their interactions with radiation and climate as well as land-use and land-cover changes (LULC) have been incorporated and rigorously tested in the IITM-ESM2.0.
These model implementations would enable the scientific community to better comprehend and quantify the impact of GHG, aerosols and LULC on the Asian monsoon. Other major value additions from the IITM-ESM2.0 include enhanced realism in representations of polar sea-ice distribution, the Atlantic Ocean Thermohaline Circulation (THC) and mean monsoon rainfall and its variability over South Asia. A 200-year pre-industrial control spin-up simulation has just been completed with the CMIP6-ready version of IITM ESM.2 and the simulations are being further extended to 500+ years.
While the IITM-ESM model will be used to study long term climate change, IITM has also developed models for extended range production that was operationally implemented in 2016.
Despite the fact that the IITM has shown significant progress in its predictive capacity, there still remains a number of challenges. Most importantly, to run the GFS T1534 model in the 21 ensemble mode, a HPC system augmented to 10 Peta Flops from the current 1.2 Peta Flops needs to be urgently put in place. The high resolution (12.5 km) model forecasts initiatives of MoES, as a part of Niti Ayog programme, is likely to make a paradigm shift in the forecasting of weather events across the country.
The author is a Scientist G & Acting Director, Indian Institute of Tropical Meteorology (IITM), Pune. email@example.com