1:45 PM - 3:15 PM
[O08-P03] Estimating the “precipitation rate” in the “Oba Misa Diary” during the Edo period
-An attempt to restore weather data using the “detail rate”-
Keywords:Nakamura Heizaemon Diary, Detail rate, Regression analysis, Oba Misa Diary
1 Abstract
Last year, we proved that the precipitation rate in the "Nakamura Heizaemon Diary" written in Kitakyushu during the Edo period was affected by the "detail rate" which measures the accuracy of the diary, and I decided to quantitatively reconstruct the precipitation rate in the Edo period before the observation by the meteorological observatory by using the "precipitation rate in the diary," "detail rate" and "precipitation rate at the meteorological observatory" as the objective variables in a regression analysis(Fig1). This year, we used the "Oba Misa Diary" written in Setagaya from the Edo period to the Meiji period as a reference material.
2 Motivation for research
In the past seven years, we have analyzed seven historical documents and have created a database of the weather recorded in these documents.
3 Purpose of the research
(1)The past seven years have been analyzed in seven ancient documents, and a database of weather records created.(Fig2)
(2)Verify the abnormal rainfall of 1868 in the diary based on the record of precipitation in Yokohama recorded by Hepburn, an American medical missionary.
(3)Using the "detailed rate" that measures the accuracy of the diary records, the "precipitation rate in the diary" and the "precipitation rate at the meteorological observatory," we will use regression analysis to reconstruct the "precipitation rate in the Edo period"before records at the meteorological observatory started.
4 The research
We classified the weather in the historical documents to make the classifications closer to the definitions currently used by Japan Meteorological Agency and the weather was classified as snowy, rainy, cloudy, or sunny.
5 Data processing
The data obtained was 11,478days over 45 years.
6 Accuracy of old document records
We made a graph of the amount of precipitation and number of rainy days which Hepburn recorded(Fig3). Also, in 1868, we found there to be abnormally rainfall in the summer. We examined the rate of precipitation appearance recorded in the Sekiguchi Diary(Fig4) and the Oba Misa Diary(Fig5), which our seniors had created a database for in 2015.
7 Detail rate of the Oba Misa Diary
The accuracy of the records was examined using the "detail rate" from a previous study (Sho et al., 2017).A graph of the Oba Misa Diary detail rate and precipitation rate show a correlation coefficient of 0.74 between the two sets of data, suggesting that the precipitation rate of the diary is affected by the level of detail(Fig6).
8 Simple regression analysis
We wanted to estimate the precipitation rate of the weather station before 1875 by regression analysis(Fig7).
(1)A simple regression analysis of "precipitation rate at the weather station" and "detail rate" yielded the equation Y=0.7781x+0.0811.
(2)A simple regression analysis of "precipitation rate at the meteorological observatory" and "precipitation rate in the diary" yielded the equation Y=1.5267x+0.0412.
9 Results of multiple regression analysis
Using multiple regression analysis of "the rate of occurrence of rain in the Oba Misa Diary during the Meiji period,"the "detail rate," and "the rate of occurrence of rain recorded at the meteorological observatory starting in 1875," with "the rate of occurrence of rain at the meteorological observatory before 1875"as the objective variable,the equation Z=0.65x+0.56y-0.024059296 is obtained.(Fig8)
When we plotted the difference between the regression line result and the actual result, we found that the multiple regression analysis (marked with a red X) gave better results with less variability(Fig9).
Therefore, we decided to use multiple regression analysis to recover the precipitation rate before 1875.(Fig10)
10 Discussion & Summary
In the Oba Misa Diary, the rate of rain in the summer and autumn of 1869 increased.
The average of “detail rate" in the “Oba Misa Diary" was 45.4%. The correlation coefficient with the rainy appearance was 0.74.
Analyzing a return analysis with “the precipitation rate", “detail rate", and “the precipitation rate of meteorological observatory", we found it was good using the heavy return analysis.
Therefore, we decided to use multiple regression analysis to recover the precipitation rate before 1875.The precipitation rate recovered by the multiple regression analysis is shown as a dark green line, with the highest rate of 56.5% and the lowest rate of 36.9% in 1871 for the period 1860-1874.The lowest occurrence of rainfall in the diary was 26.0% in 1863, which means that the analysis was more accurate.
11 Future Plans
We will create a weather database by looking for old documents written in the Chugoku and Shikoku regions between the Edo and Meiji periods.
Last year, we proved that the precipitation rate in the "Nakamura Heizaemon Diary" written in Kitakyushu during the Edo period was affected by the "detail rate" which measures the accuracy of the diary, and I decided to quantitatively reconstruct the precipitation rate in the Edo period before the observation by the meteorological observatory by using the "precipitation rate in the diary," "detail rate" and "precipitation rate at the meteorological observatory" as the objective variables in a regression analysis(Fig1). This year, we used the "Oba Misa Diary" written in Setagaya from the Edo period to the Meiji period as a reference material.
2 Motivation for research
In the past seven years, we have analyzed seven historical documents and have created a database of the weather recorded in these documents.
3 Purpose of the research
(1)The past seven years have been analyzed in seven ancient documents, and a database of weather records created.(Fig2)
(2)Verify the abnormal rainfall of 1868 in the diary based on the record of precipitation in Yokohama recorded by Hepburn, an American medical missionary.
(3)Using the "detailed rate" that measures the accuracy of the diary records, the "precipitation rate in the diary" and the "precipitation rate at the meteorological observatory," we will use regression analysis to reconstruct the "precipitation rate in the Edo period"before records at the meteorological observatory started.
4 The research
We classified the weather in the historical documents to make the classifications closer to the definitions currently used by Japan Meteorological Agency and the weather was classified as snowy, rainy, cloudy, or sunny.
5 Data processing
The data obtained was 11,478days over 45 years.
6 Accuracy of old document records
We made a graph of the amount of precipitation and number of rainy days which Hepburn recorded(Fig3). Also, in 1868, we found there to be abnormally rainfall in the summer. We examined the rate of precipitation appearance recorded in the Sekiguchi Diary(Fig4) and the Oba Misa Diary(Fig5), which our seniors had created a database for in 2015.
7 Detail rate of the Oba Misa Diary
The accuracy of the records was examined using the "detail rate" from a previous study (Sho et al., 2017).A graph of the Oba Misa Diary detail rate and precipitation rate show a correlation coefficient of 0.74 between the two sets of data, suggesting that the precipitation rate of the diary is affected by the level of detail(Fig6).
8 Simple regression analysis
We wanted to estimate the precipitation rate of the weather station before 1875 by regression analysis(Fig7).
(1)A simple regression analysis of "precipitation rate at the weather station" and "detail rate" yielded the equation Y=0.7781x+0.0811.
(2)A simple regression analysis of "precipitation rate at the meteorological observatory" and "precipitation rate in the diary" yielded the equation Y=1.5267x+0.0412.
9 Results of multiple regression analysis
Using multiple regression analysis of "the rate of occurrence of rain in the Oba Misa Diary during the Meiji period,"the "detail rate," and "the rate of occurrence of rain recorded at the meteorological observatory starting in 1875," with "the rate of occurrence of rain at the meteorological observatory before 1875"as the objective variable,the equation Z=0.65x+0.56y-0.024059296 is obtained.(Fig8)
When we plotted the difference between the regression line result and the actual result, we found that the multiple regression analysis (marked with a red X) gave better results with less variability(Fig9).
Therefore, we decided to use multiple regression analysis to recover the precipitation rate before 1875.(Fig10)
10 Discussion & Summary
In the Oba Misa Diary, the rate of rain in the summer and autumn of 1869 increased.
The average of “detail rate" in the “Oba Misa Diary" was 45.4%. The correlation coefficient with the rainy appearance was 0.74.
Analyzing a return analysis with “the precipitation rate", “detail rate", and “the precipitation rate of meteorological observatory", we found it was good using the heavy return analysis.
Therefore, we decided to use multiple regression analysis to recover the precipitation rate before 1875.The precipitation rate recovered by the multiple regression analysis is shown as a dark green line, with the highest rate of 56.5% and the lowest rate of 36.9% in 1871 for the period 1860-1874.The lowest occurrence of rainfall in the diary was 26.0% in 1863, which means that the analysis was more accurate.
11 Future Plans
We will create a weather database by looking for old documents written in the Chugoku and Shikoku regions between the Edo and Meiji periods.