1. Parameterization of the middle and upper tropospheric water vapor from ATOVS observations over a tropical climate region
- Author
-
Khiruddin Abdullah, Ezekiel Kaura Makama, and Hwee San Lim
- Subjects
Atmospheric Science ,010504 meteorology & atmospheric sciences ,Mean squared error ,0208 environmental biotechnology ,Regression analysis ,02 engineering and technology ,Atmospheric sciences ,01 natural sciences ,020801 environmental engineering ,law.invention ,Latitude ,Troposphere ,Geophysics ,Space and Planetary Science ,law ,Linear regression ,Radiosonde ,Environmental science ,Satellite ,Water vapor ,0105 earth and related environmental sciences - Abstract
Precipitable water vapor (PWV) is a highly variable, but important greenhouse gas that regulates the radiation budget of the earth. Its variability in time and space makes it difficult to quantify. Knowledge of its vertical distribution, in particular, is crucial for many reasons. In this study, empirical relationships between isobaric layers of PWV over Peninsular Malaysia are examined. Analysis of variance (ANOVA) technique on Advanced Television and Infrared Observation Satellite Operational Vertical Sounder (ATOVS) observations, from 2005 to 2011, has been used to propose a relationship of the form, W=α(WL)β for the middle (MW) and upper (UW) layers PWV. W is either MW or UW with α and β as regression coefficients, which are functions of latitude. Coefficients of determination (R2) and root mean square error (RMSE) of respective values between 0.75–0.86 and 1.65–2.38 mm, across the zones, were obtained for both the MW and UW predictions, with a mean bias (MB) below ±1 mm.The predicted and observed PWV presented a better agreement northerly. Initial predictability test for each model was done on two independent data sets: ATOVS (2012–2015), and radiosonde (2010–2011) at Penang, Kuantan and Sepang stations, with very good outcomes. The results of the tests revealed remarkable performances, when compared with two previously reported models. The inclusion of variable regression coefficients, and the utilization of satellite-derived data, which provide soundings of data-void regions between radiosonde networks, proved to have optimized the results.
- Published
- 2018