Pulak Guhathakurta

Application of Artificial Neural Networks in Weather Forecasting: A Comprehensive Literature Review

To recognize application of Artificial Neural Networks (ANNs) in weather forecasting, especially in rainfall forecasting a comprehensive literature review from 1923 to 2012 is done and presented in this paper. And it is found that architectures of ANN such as BPN, RBFN is best established to be forecast chaotic behavior and have efficient enough to forecast monsoon rainfall as well as other weather parameter prediction phenomenon over the smaller geographical region.

Detecting changes in rainfall pattern and seasonality index vis-à-vis increasing water scarcity in Maharashtra

Knowledge of mean rainfall and its variability of smaller spatial scale are important for the planners in various sectors including water and agriculture. In the present work, long rainfall data series (1901–2006) of districts of Maharashtra in monthly and seasonal scales are constructed and then mean rainfall and coefficient of variability are analyzed to get the spatial pattern and variability. Significant long term changes in monthly rainfall in the district scale are identified by trend analysis of rainfall time series. The seasonality index which is the measure of distribution of precipitation throughout the seasonal cycle is used to classify the different rainfall regime. Also long term changes of the seasonality index are identified by the trend analysis. The state Maharashtra which is to the northwest of peninsular India is highly influenced by the southwest monsoon and the state is facing water scarcity almost every year. This study will help to find out possible reason for the increasing water scarcity in Maharashtra.

A combined deficit index for regional agricultural drought assessment over semi-arid tract of India using geostationary meteorological satellite data

Abstract The untimely onset and uneven distribution of south-west monsoon rainfall lead to agricultural drought causing reduction in food-grain production with high vulnerability over semi-arid tract (SAT) of India. A combined deficit index (CDI) has been developed from tri-monthly sum of deficit in antecedent rainfall and deficit in monthly vegetation vigor with a lag period of one month between the two. The formulation of CDI used a core biophysical (e.g., NDVI) and a hydro-meteorological (e.g., rainfall) variables derived using observation from Indian geostationary satellites. The CDI was tested and evaluated in two drought years (2009 and 2012) within a span of five years (2009–2013) over SAT. The index was found to have good correlation (0.49–0.68) with standardized precipitation index (SPI) computed from rain-gauge measurements but showed lower correlation with anomaly in monthly land surface temperature (LST). Significant correlations were found between CDI and reduction in agricultural carbon productivity (0.67–0.83), evapotranspiration (0.64–0.73), agricultural grain yield (0.70–0.85). Inconsistent correlation between CDI and ET reduction was noticed in 2012 in contrast to consistent correlation between CDI and reduction in carbon productivity both in 2009 and 2012. The comparison of CDI-based drought-affected area with those from existing operational approach showed 75% overlapping regions though class-to-class matching was only 40–45%. The results demonstrated that CDI is a potential indicator for assessment of late-season regional agricultural drought based on lag-response between water supply and crop vigor.

Variability and Trends of Extreme Rainfall and Rainstorms

In this chapter, variability and long-term trends of heavy rainfall events and rainstorms over India during the monsoon season (June to September) are documented. For analyzing extreme rainfall events, rain gauge station and gridded data for the period 1901–2010 have been used. For studying rainstorms, another gridded rainfall data set of 1951–2010 also has been used. The trend analyses revealed increasing trends in the frequency of dry days in most parts of the country during the winter, pre-monsoon, and southwest monsoon seasons. Frequency of very light rain and light-to-moderate rain events has decreased significantly over most of the country. Both the station and gridded data have shown significant increasing trends of very heavy to extremely heavy rainfall events over most parts of the country. Over Central India, extreme rainfall events show significant decadal variations which could be related to variations in sea surface temperatures over the tropical oceans. Over the period 1901–2010, heavy rainfall events show an increasing trend of six percent per decade. Further, an analysis is made on rainstorms over North India where majority of rainstorms cause floods over North India. Frequency of rainstorms has shown an increasing trend of 4 rainstorms (50 % increase) in 65 years (1951–2015). Similarly, the duration of rainstorms has shown an increase of about 15 days (80 % increase) during the period 1951–2015, which is also significant. Both the increases are statistically significant at the 99 % confidence level.

A Broad Literature Survey of Development and Application of Artificial Neural Networks in Rainfall-Runoff Modelling

Rainfall-Runoff (R-R) modelling is one of the most important and challenging work in the real and present world. In all-purpose, rainfall, temperature, soil moisture and infiltration are highly nonlinear and complicated parameters. These parameters have been used in R-R modelling and this modelling requires highly developed techniques and simulation for accurate forecasting. An artificial neural network (ANN) is a successful technique and it has a capability to design R-R model but selection of appropriate architecture (model) of ANN is most important challenge. To determine the significant development and application of artificial neural network in R-R modelling, a broad literature survey last 35 years (from 1979 to 2014) is done and results are presented in this survey paper. It is concluded that the architectures of ANN, such as back propagation neural network (BPN), radial basis function (RBF), and fuzzy neural network (FNN) are better evaluated over the conceptual and numerical method and worldwide recognized to be modelled the R-R.

Trends and variability of meteorological drought over the districts of India using standardized precipitation index

Meteorological drought during the southwest monsoon season and for the northeast monsoon season over five meteorological subdivisions of India for the period 1901–2015 has been examined using district and all India standardized precipitation index (SPI). Whenever all India southwest monsoon rainfall was less than −10% or below normal, for those years all India SPI was found as −1 or less. Composite analysis of SPI for the below normal years, viz., less than −15% and −20% of normal rainfall years indicate that during those years more than 30% of country’s area was under drought condition, whenever all India southwest monsoon rainfall was –15% or less than normal. Trend analysis of monthly SPI for the monsoon months identified the districts experiencing significant increase in drought occurrences. Significant positive correlation has been found with the meteorological drought over most of the districts of central, northern and peninsular India, while negative correlation was seen over the districts of eastern India with NINO 3.4 SST. For the first time, meteorological drought analysis over districts and its association with equatorial pacific SST and probability analysis has been done for the northeast monsoon over the affected regions of south peninsular India. Temporal correlation of all India southwest monsoon SPI and south peninsular India northeast monsoon SPI has been done with the global SST to identify the teleconnection of drought in India with global parameters.

Observational Analysis of Heavy Rainfall During Southwest Monsoon over India

The climate of south Asia is largely dominated by monsoon circulation. Most of the rainfall received in India is during the 4 months of southwest monsoon season. In recent years, India has faced frequent and severe floods that caused havoc in terms of economic loss and loss of human lives. The devastating floods are occurring almost every year but the places are not same for every occurrence. Most of these floods are categorized as flash floods which are generally associated with heavy precipitation. Heavy precipitation with cloud burst also caused disasters, particularly in northern states of the country. It may be mentioned that the information on the changes in extreme weather events is more important than the changes in mean pattern for better disaster management and mitigation. There is also high temporal variability of monsoon rainfall. This causes the extreme years with high monsoon rainfall departure from the long period mean value, the positive departure causing flood and negative departure causing drought. The variability of monsoon rainfall has been studied by many climate scientists and they have also drawn several conclusions. However the study of variability of rainfall is different from the studies of other climate parameters.

Identifying the Changes in Rainfall Pattern and Heavy Rainfall Events During 1871–2010 over Cherrapunji

Cherrapunji holds world record of highest recorded point rainfall for different durations i.e. 1 month, 2 months, 12 months, 2 years (WMO 1994). It is situated in the southern slopes of Khasi and Jayantia Hills of ‘Meghalaya’ province of India overlooking the plains of the Sylhet district in Bangladesh. The meaning of Meghalaya is the abode of cloud. The altitude of Cherrapunji (Fig. 1) is 1,313 m above mean sea level. Rain occurs almost regularly throughout the year over this region due to its peculiar orographic features. Synoptically, this region is the convergence zone of westerly, easterly and moist southerly winds causing lots of rain and therefore the average monthly rainfall is also very high over this region. The monsoon clouds fly unhindered over the plains of Bangladesh for about 400 km. Thereafter, they hit Khasi hills which abruptly erupt out of the plains to reach a height of about 1,370 m above m.s.l. within a short distance of 2–5 km. The orography of the hills, with many deep valleys, channels the low flying (150–300 m) moisture laden clouds from a wide area to converge over Cherrapunji which falls in the middle of the path of this stream. The winds push the rain clouds through these gorges and up the steep slopes. The rapid ascendance of the clouds into the upper atmosphere hastens the cooling and helps vapour to condense. Most of Cherrapunji’s rain is the consequence of air being lifted as a large body of water vapour. Extremely large amount of rainfall at Cherrapunji is perhaps the most well-known feature of orographic rain in northeast India.