Premium
A statistical estimate of daily mean temperature derived from a limited number of daily observations
Author(s) -
Zaiki Masumi,
Kimura Keiji,
Mikami Takehiko
Publication year - 2002
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1029/2001gl014378
Subject(s) - statistics , linear regression , environmental science , mean radiant temperature , estimation , statistical analysis , mean value , mathematics , meteorology , climatology , atmospheric sciences , geography , climate change , geology , oceanography , management , economics
When a study is made of past climatic variations, it is preferable to obtain as long an instrumental record as possible. In Japan, the earliest official meteorological observations by JMA (Japan Meteorological Agency) date back to Hakodate in 1872. For about 130 years, the frequency and time of daily observation has varied. This is one of the major factors in creating data inhomogeneity. Accordingly, statistical estimation has been attempted to bridge the gap in daily mean temperatures caused by the limited number of daily observations. A Multiple Linear Regression Analysis was practically applied in the estimation of the daily mean temperature. The true value obtained from 24 hourly observational values (real daily mean temperature) was compared to a derived result using only three to four plus maximum and minimum daily observed temperatures. A high correlation between real values and estimated values was proven to exist.