Binary classification of rainfall time-series using machine learning algorithms

Summer monsoon rainfall contributes more than 75% of the annual rainfall in India. For the state of Maharashtra, India, this is more than 80% for almost all regions of the state. The high variability of rainfall during this period necessitates the classification of rainy and non-rainy days. While there are various approaches to rainfall classification, this paper proposes rainfall classification based on weather variables. This paper explores the use of support vector machine (SVM) and artificial neural network (ANN) algorithms for the binary classification of summer monsoon rainfall using common weather variables such as relative humidity, temperature, pressure. The daily data, for the summer monsoon months, for nineteen years, was collected for the Shivajinagar station of Pune in the state of Maharashtra, India. Classification accuracy of 82.1 and 82.8%, respectively, was achieved with SVM and ANN algorithms, for an imbalanced dataset. While performance parameters such as misclassification rate, F1 score indicate that better results were achieved with ANN, model parameter selection for SVM was less involved than ANN. Domain adaptation technique was used for rainfall classification at the other two stations of Maharashtra with the network trained for the Shivajinagar station. Satisfactory results for these two stations were obtained only after changing the training method for SVM and ANN.

Short range SW monsoon rainfall forecasting over India using neural networks

Feedforward Neural Networks are used for daily precipitation forecast using several test stations all over India. The six year European Centre of Medium Range Weather Forecasting (ECMWF) data is used with the training set consisting of the four year data from 1985-1988 and validation set consisting of the data from 1989-1990. Neural networks are used to develop a concurrent relationship between precipitation and other atmospheric variables. No attempt is made to select optimal variables for this study and the inputs are chosen to be same as the ones obtained earlier at National Center for Medium Range Weather Forecasting (NCMRWF) in developing a linear regression model. Neural networks are found to yield results which are atleast as good as linear regression and in several cases yield 10 - 20 % improvement. This is encouraging since the variable selection has so far been optimized for linear regression.

Relationship between surface fields over Indian ocean and monsoon rainfall over homogeneous zones of India

The monthly mean atmospheric fields and surface parameters of NCEP/NCAR reanalysis for the period 1948-1998 have been studied to examine the characteristics of monsoon circulation features, sea surface temperature (SST), sea level pressure, surface wind stress and latent heat flux over the Indian Ocean and nearby seas during deficient, normal and excess rain years. The entire period of study has been classified into deficient, normal and excess rain years for all India as well as for each of the five homogeneous zones separately based on the observed seasonal mean rainfall. On the basis of the mean characteristics of the surface fields, the oceanic region covering the Indian Ocean and adjacent seas has been divided into four regional sectors. Using various statistical means the relation between the surface fields over the four regional sectors and the monsoon rainfall over five homogeneous zones of Indian landmass has been examined. Attempt have been made to identify some surface parameters which can be used as predictors for seasonal mean monsoon rainfall over the entire India and also over some homogeneous zones.

Predictive relationships between Indian Ocean sea surface temperatures and Indian summer monsoon rainfall

Monthly sea surface temperature (SST) data of 49 years (1950-98) have been analysed to examine the relationship of SST anomalies in the Indian Ocean with Indian summer monsoon rainfall (ISMR) and to derive useful predictors for long-range forecasts of ISMR. There is significant positive relationship between ISMR and SST anomalies over the Arabian Sea during November to January and also in May. SST anomalies over southeast Indian Ocean during February to March and over North Pacific during May are also positively correlated with ISMR. The composite analysis revealed that in Non-ENSO drought years (1966, 1968, 1974 and 1979) negative SST anomalies are observed over south Indian Ocean from February which slowly spread towards equator during the subsequent months. These negative SST anomalies which persist during the monsoon season may be playing an important role in modulating ISMR especially in non-ENSO years. We have derived two indices, ARBSST (SST anomalies in Arabian Sea averaged over 15o - 25o N, 50o -70o E and November-December-January) and SIOSST (SST anomalies over south Indian Ocean averaged over 15o -30o S, 70o -110o E and February and March) as useful predictors for the long-range forecasts of ISMR. The correlation coefficient (for the period 1950-98) of ARBSST and SIOSST with ISMR is 0.45 and 0.46 respectively which is statistically significant at 99.9 % level. SIOSST index has shown consistently stable relationship with ISMR. However the ARBSST index showed significant correlation with ISMR only after 1976.

Does the El Niño-Southern Oscillation Impact on the Indian Summer Monsoon 1-Dimensional? Quantifying the Role of Antecedent Southwestern Indian Ocean Capacitance on the Variability of Summer Monsoon Rainfall over Homogeneous Regions of India

Recent rapid changes in the global climate and warming temperatures increase the demand for local and regional weather forecasting and analysis to improve the accuracy of seasonal forecasting of extreme events such as droughts and floods. On the other hand, the role of ocean variability is at a focal point in improving the forecasting at different time scales. Here we study the effect of Indian Ocean mean sea level anomaly (MSLA) and sea surface temperature anomalies (SSTA) on Indian summer monsoon rainfall during 1993-2019. While SSTA and MSLA have been increasing in the southwestern Indian Ocean (SWIO), these parameters' large-scale variability and pre-monsoon winds could impact the inter-annual Indian monsoon rainfall variability over homogeneous regions. Similarly, antecedent heat capacitance over SWIO on an inter-annual time scale has been the key to the extreme monsoon rainfall variability from an oceanic perspective. Though both SSTA and MSLA over SWIO have been influenced by El Niño-southern oscillation (ENSO), the impact of SWIO variability was low on rainfall variability over several homogeneous regions. However, rainfall over northeast (NE) and North India (NI) has been moulded by ENSO, thus changing the annual rainfall magnitude. Nevertheless, the impact of ENSO on monsoon rainfall through SWIO variability during the antecedent months is moderate. Thus, the ENSO influence on the atmosphere could be dominating the ocean part in modulating the inter-annual variability of the summer monsoon. Analysis shows that the cooler (warmer) anomaly over the western Indian Ocean affects rainfall variability adversely (favourably) due to the reversal of the wind pattern during the pre-monsoon period.

Influence of Bay o Bengal on winter monsoon rainfall

ABSTRACT. Utilizing marine meteorological data the values of 1 latent heat flux, sea surface temperature (SST) and sea minus air temperature have been computed on a grid mesh of 5° over the Bay of Bengal during September month of the contrasting Winter monsoon years 1987 and 1988. It has been found that the good winter monsoon of 1987 followed (I) higher SSTs over western Bay of Bengal; (ii) very high evaporation rate over the sea area bounded by 10°.20°N. 80°.90oE and (iii) instability in the surface layer over north and adjoining central Bay of Bengal, whereas, the bad winter monsoon of 1988 followed (i) lower SSTs over western Bay of Bengal; (ii) very low evaporation rate over the area I0°.20oN, 80°.90oE and (iii) stability in the surface layer over north and adjoining central Bay of Bengal.

On some aspects of southwest monsoon rainfall over Tamilnadu

The rainfall during southwest monsoon season over Tamilnadu is quite significant from the point of view of water storage in major reservoirs as northeast monsoon rainfall, which is about half of the annual rainfall, is not stable enough due to its large interannual variability. The southwest monsoon rainfall, on the other hand, is more stable. The north-south oriented trough over Tamilnadu and adjoining Bay togetherwith upper air cyclonic circulation/trough in lower tropospheric levels account for three fourths of significant rainfall occurrence during southwest monsoon season. Rainfall during southwest monsoon and northeast monsoon seasons was found to be independent with a small negative correlation of -0.18. This shows that the southwest monsoon rainfall may not be of much use to predict the pattern of northeast monoon rainfall over Tamilnadu.

Spectrum analysis and prediction possibilities of the onset dates of southwest monsoon over Kerala (India)

Maximum Entropy Spectral Analysis of the time series for the onset dates of the southwest monsoon over Kerala (India) revealed several periodicities significant at a 2a a priori level. some at a 3 C a priori level However these contributed only 40-50% to the total variance thus indicating 50-60% as purely random component. Also many of the significant periodicities observed were in the QBO region (T = 2-3 years) which. due to their variable periodicities and amplitudes, are almost equivalent to a random component. Hence predictions were possible only with a limit exceeding 5 days which are probably not very useful for any planning purposes agricultural or otherwise. No relationship was found between onset dates of established monsoon rainfall and the 50 hPa mean monthly equatorial zonal wind for the months of March, April, May or June. However there is a possibility that a relationship may exist between westerly (easterly) winds in May and early (late) onset of the first monsoon (or pre-monsoon ?) rainfall in Kerala. Meager or otherwise.

Variability of southwest monsoon rainfall over West Bengal

Monthly and seasonal variations of southwest monsoon rainfall over the districts of Gangetic and Sub-Himalayan West Bengal are presented and their differences discussed. Latitudinal variations of monsoon rainfall are brought out. Decadal means of seasonal rainfall over plains are compared with those at higher elevations and northern latitudes. An attempt is made to study long term rainfall trends.