Systems and methods for prediction of increase in sea surface temperature

1 . A computer-implemented method of predicting an increase in average sea surface temperature, the method including:

setting, by a processor, a value of a first Boolean flag to true when at least one of an eclipse season start date and an eclipse season end date in a particular calendar year occurs in a first predefined number of days preceding a timestamp for a particular partition in a plurality of partitions segmenting the particular calendar year;

training a machine learning model, wherein the training comprises comparing an output of the machine learning model with ground truth labels:

predicting, using the trained machine learning model, a probability of increase in average sea surface temperature relative to a historical average sea surface temperature for a predefined time interval in the particular calendar year, wherein the predicting is based at least in part on inputting, to the trained machine learning model, the value of the first Boolean flag; and

invoking an external system to predict occurrence of an extreme climate event in the predefined time interval when the probability of increase in average sea surface temperature is above a first predefined threshold.

2 . The method of claim 1 , wherein the eclipse season comprises at least one lunar eclipse and at least one solar eclipse within a predefined eclipse season duration starting from the eclipse season start date and ending at the eclipse season end date.

3 . The computer-implemented method of claim 2 , further including:

determining, a first time interval, between occurrence of the at least one lunar eclipse in the eclipse season and the timestamp for the particular partition and a second time interval, between occurrence of the at least one solar eclipse in the eclipse season and the timestamp for the particular partition;

selecting from amongst the first time interval and the second time interval, at least one time interval as a first selected time interval that is shorter in duration than the other time interval; and

the inputting, further includes providing the first selected time interval as input to the trained machine learning model.

4 . The computer-implemented method of claim 3 , further including:

determining, a third time interval, between occurrence of the at least one lunar eclipse in the eclipse season and the timestamp for the particular partition and a fourth time interval, between occurrence of the at least one solar eclipse in the eclipse season and the timestamp for the particular partition;

selecting from amongst the third time interval and the fourth time interval, at least one time interval as a second selected time interval that is shorter in duration than the other time interval; and

the inputting, further includes providing the second selected time interval as input to the trained machine learning model.

5 . The computer-implemented method of claim 2 , further including:

determining, for the particular calendar year, a plurality of distances between the Moon and the Earth corresponding to occurrence of plurality of the at least one lunar eclipse and the at least one solar eclipse within the predefined eclipse season in the particular calendar year; and

the inputting further includes providing, as input to the trained machine learning model, the determined plurality of distances between the Moon and the Earth.

6 . The method of claim 1 , wherein the trained machine learning model is trained using a plurality of time series data, a particular time series data in the plurality of time series data comprises (i) a plurality of timestamps corresponding to respective plurality of solar eclipses and a plurality of timestamps corresponding to respective plurality of lunar eclipses in an eighteen-year Saros Cycle, (ii) a plurality of timestamps corresponding to the plurality of partitions segmenting respective plurality of calendar years in the eighteen-year Saros Cycle and (iii) a plurality of average sea surface temperatures corresponding to the plurality of calendar years in the eighteen-year Saros Cycle.

7 . The computer-implemented method of claim 6 , further including:

inputting a plurality of predicted probability of increase in the average sea surface temperature as predicted by the trained machine learning model corresponding to a plurality of calendar years in the eighteen-year Saros Cycle to a second trained machine learning model;

wherein the second trained machine learning model is trained using the plurality of time series data to predict a probability of occurrence of an extreme climate event when the plurality of predicted probability of increase in the average sea surface temperature as predicted by the trained machine learning model corresponding to the plurality of calendar years are above a second predefined threshold; and

predicting, using the second trained machine learning model, the probability of occurrence of the extreme climate event following the plurality of calendar years in the eighteen-year Saros Cycle.

8 . The computer-implemented method of claim 7 , wherein the predicting the probability of occurrence of the extreme climate event is performed for the predefined time interval following the plurality of calendar years in the eighteen-year Saros Cycle.

9 . The computer-implemented method of claim 1 , wherein the setting of the value of the first Boolean flag is performed when at least one of the eclipse season start date and the eclipse season end date in the particular calendar year occurs in a second predefined number of days preceding the timestamp for the particular partition in the plurality of partitions segmenting the particular calendar year.

10 . The computer-implemented method of claim 9 , wherein the trained machine learning model is trained to assign at least a double weightage to the value of the first Boolean flag for the predicting of the probability of increase in the average sea surface temperature when a difference between the eclipse season start date and the timestamp for the particular partition in the plurality of partitions is less than a third predefined number of days.

11 . The computer-implemented method of claim 10 , further including:

setting a value of a second Boolean flag when at least one of the eclipse season start date and the eclipse season end date in the particular calendar year occurs in the first predefined number of days succeeding a timestamp for the particular partition in the plurality of partitions segmenting the particular calendar year; and

the inputting to the trained machine learning model further includes, inputting the second Boolean flag.

12 . The computer-implemented method of claim 11 , wherein the setting the value of the second Boolean flag is performed when the at least one of the eclipse season start date and the eclipse season end date in the particular calendar year occurs in the second predefined number of days succeeding the timestamp for the particular partition in the plurality of partitions segmenting the particular calendar year.

13 . The method of claim 12 , wherein the first predefined number of days range between twenty five and thirty four, the second predefined number of days range between thirty five and forty five and the third predefined number of days are less than ten.

14 . The computer-implemented method of claim 1 , wherein a duration of the predefined time interval ranges from one minute to one year.

15 . The computer-implemented method of claim 1 , wherein the plurality of partitions segmenting the particular calendar year include at least a March equinox, a June solstice, a September equinox and a December solstice.

16 . The computer-implemented method of claim 1 , further including:

determining, for the particular calendar year, respective distances between the Moon and the Earth for each new-moon phase and each full-moon phase; and

the inputting further includes providing, as input to the trained machine learning model, the determined respective distances between the Moon and the Earth for each new-moon phase and each full-moon phase in the particular calendar year and respective timestamps for each new-moon phase and each full-moon phase.

17 . The computer-implemented method of claim 1 , wherein the extreme climate event is at least one of an El Niño event indicating above average temperatures in Niño 3.4 region of equatorial Pacific Ocean and a La Niña event indicating below average temperatures in Niño 3.4 region of equatorial Pacific Ocean.

18 . The computer-implemented method of claim 1 , wherein the extreme climate event indicates occurrence of an interaction across a lower atmosphere of the Earth and an upper atmosphere of the Earth causing an extreme weather event.

19 . A system including one or more processors coupled to memory, the memory loaded with computer instructions to predict an increase in average sea surface temperature, the instructions, when executed on the processors, implement actions comprising:

setting, by a processor, a value of a first Boolean flag to true when at least one of an eclipse season start date and an eclipse season end date in a particular calendar year occurs in a first predefined number of days preceding a timestamp for a particular partition in a plurality of partitions segmenting the particular calendar year;

training a machine learning model, wherein the training comprises comparing an output of the machine learning model with ground truth labels;

predicting, using the trained machine learning model, a probability of increase in average sea surface temperature relative to a historical average sea surface temperature for a predefined time interval in the particular calendar year, wherein the predicting is based at least in part on inputting, to the trained machine learning model, the value of the first Boolean flag; and

invoking an external system to predict occurrence of an extreme climate event in the predefined time interval when the probability of the increase in average sea surface temperature is above a first predefined threshold.

20 . A non-transitory computer readable storage medium impressed with computer program instructions to predict an increase in average sea surface temperature, the instructions, when executed on a processor, implement a method, comprising:

setting, by the processor, a value of a first Boolean flag to true when at least one of an eclipse season start date and an eclipse season end date in a particular calendar year occurs in a first predefined number of days preceding a timestamp for a particular partition in a plurality of partitions segmenting the particular calendar year;

training a machine learning model, wherein the training comprises comparing an output of the machine learning model with ground truth labels;

predicting, using the trained machine learning model, a probability of increase in average sea surface temperature relative to a historical average sea surface temperature for a predefined time interval in the particular calendar year, wherein the predicting is based at least in part on inputting, to the trained machine learning model, the value of the first Boolean flag; and

invoking an external system to predict occurrence of an extreme climate event in the predefined time interval when the probability of increase in average sea surface temperature is above a first predefined threshold.