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.

Tropospheric temperature variation over India and links with the Indian summer monsoon : 1971-2000

In the context of the ever increasing interest in the regional aspects of global warming, understanding the spatio-temporal variations of tropospheric temperature over India is of great importance. The present study, based on the data from 19 well distributed radiosonde stations for the period 1971-2000, examines the seasonal and annual mean temperature variations at the surface and five selected upper levels, viz., 850, 700, 500, 200 and 150 hPa. An attempt has also been made to bring out the association between tropospheric temperature variations over India and the summer monsoon variability, including the role of its major teleconnection parameter, the El Niño/Southern Oscillation (ENSO). Seasonal and annual mean all-India temperature series are analyzed for surface and five tropospheric levels. The mean annual cycles of temperature at different tropospheric levels indicate that the pre-monsoon season is slightly warmer than the monsoon season at the surface, 850 hPa and 150 hPa levels, while it is relatively cooler at all intermediate levels. The mean annual temperature shows a warming of 0.18° C and 0.3° C per 10 years at the surface and 850 hPa, respectively. Tropospheric temperature anomaly composites of excess (deficient) monsoon rainfall years show pronounced positive (negative) anomalies during the month of May, at all the levels. The pre-monsoon pressure of Darwin has significant positive correlation with the monsoon temperature at the surface and 850 hPa.

Impact of Indian and East Asian summer monsoons on the diurnal temperature range of the low-latitude highlands of China in the rainy season

This study investigates the impact of the Indian and East Asian summer monsoons on the diurnal temperature range (DTR) in the low-latitude highlands of China (CLLH) based on in-situ DTR observations, ERA5 reanalysis data, and numerical simulations. Diagnoses indicate that the DTR in the CLLH shows a significant positive correlation with the Indian summer monsoon (ISM), while a negative correlation with the East Asian summer monsoon (EASM). When a strengthened ISM occurs with a weakened EASM, an anomalous anticyclonic circulation with downward motion is excited over the CLLH. This anomalous circulation pattern increases the DTR in the rainy season by reducing the medium and high cloud cover in the CLLH. When a weakened ISM with a strengthened EASM decreases the DTR over the CLLH in the rainy season. Numerical experiments help to verify this crucial physical process linking the variability of the ISM and EASM with the DTR in the CLLH. The model results further indicate that the covariability of ISM and EASM contributes most to the variability of the rainy season DTR in the CLLH, followed by the individual variability of the EASM, and the smallest contribution to the rainy season DTR in the CLLH is the individual variability of the ISM.

Summer monsoon low pressure systems over the Indian region and their relationship with the sub-divisional rainfall

In the present paper performance of the monthly sub-divisional summer monsoon rainfall is studied in association with the position of the Low Pressure System (LPS) over the Indian region. Existence of the LPS over a particular location increases the rainfall activities in certain parts of the country while decreases in some other parts. For this study, the Indian region (5°-35° N and 60° -100° E) is divided into 5° Lat. ´ 5° Long. grids. The duration of LPS is taken in terms of LPS days with respect to the location of LPS in a particular grid. Monthly total number of LPS days in each of the grids are computed during the summer monsoon season, June to September for the period 1891 – 1990. Maximum number of LPS days (more than half of the total) are observed in the latitude belt between 20°-25°N. The percentages of total LPS days in this area are higher in July and August which are peak monsoon months as compared to June and September. When there is a LPS are in the area 20°-25° N and 80°-90° E, there is significant increase in the rainfall activities in the sub-divisions along mean monsoon trough while northeast India and southeast peninsular India experience significant decrease in rainfall in the months of July and August. Owing to the movement of LPS from east to west through central India, most parts of the country, excluding northeast India and south peninsular India get good rainfall activity. Correlation coefficients between monthly LPS days over the different grids and monthly sub-divisional rainfall are computed to study the relationships. The performance of sub-divisional rainfall mostly related with the occurrence of LPS in certain grid- locations. The correlation field maps may give some useful information about rainfall performance due to LPS in a particular grid locations.

Land-sea Thermal Contrast in Relation With Summer Monsoon Onset Over the Chao Phraya River Basin

An earlier onset of the Southeast Asian summer monsoon (SAM) was observed over the Chao Phraya River basin in Thailand using Thai Meteorological Department (TMD)-derived high-resolution merged rainfall from 1981 to 2016. As the SAM is precipitous, its variability depends on many local and global factors, such as thermal conditions over the Bay of Bengal (BoB) and Tibetan Plateau (TbT). Despite tremendous studies in the past, the role of thermal heat contrast over SAM is still not fully understood. Using the observation and reanalysis datasets, it was found that the absolute value of total heat over the BoB was higher. However, the interannual variability in total heat is higher over the TbT. Significant changes in surface temperature (±1.5°C), air thickness (±20 meters) and geopotential height found over the TbT were associated with early (late) SAM onset. The results also suggested that the significant changes in air thickness were influenced by the surface temperature difference over the TbT, and the changes in the integrated apparent heat source and integrated apparent moisture sink were up to ± 100 Wm−2, which resulted in stronger (weaker) convective activities over the BoB and mainland of the Indochina Peninsula during early (late) SAM onset. At the intraseasonal timescale, the instance MJO found over the Indian Ocean and Western Hemisphere at 4 to 10 days span during early SAM onset. An opposite scenario is found for a late SAM onset years with MJO location over Western Pacific and Maritime continent.