Study of Satellite and GroundUV Index: Climatology andEffects of GeophysicalParameters

Abstract

This thesis presents UV index (UVI) comparison between the AURA-OMI satellite and ground measurements using Norwegian and Nepalese UV Network data for the coincident measurements (Paper I and III). Influencing parameters that can modify the OMI UVI are also analyzed qualitatively and quantitatively using ANOVA and multiple regressions (Paper I and II). The bias found in the OMI data are empirically corrected and used for a climatological study (Paper III and IV). The UVI data from Norway are categorized into four groups: cloud-free/snow-free, cloud-free/snowy, cloudy/snowfree, and cloudy/snowy using the proxies Lambert equivalent reflectivity (LER) at 360 nm and surface albedo. Nepal data are categorized into two groups: cloud-free/snowfree and cloudy/snow-free using the proxies Cloud Transmission Factor (CTF) and surface albedo. Descriptive statistics are provided for the coincident measurement pairs of OMI and ground data for each station and condition. The influencing factors are further analyzed using Analysis of Variance (ANOVA) and multiple regressions.

The results show OMI overestimates the ground measurements, and the bias is below 20% for Norwegian data (Paper I), and the bias is above 40% (Paper III) for Nepal’s data under cloud-free and snow-free conditions. The presence of cloud increases the bias and the bias reaches to 29-73% and more than 100% for Norway and Nepal data respectively. Under cloud-free but snowy conditions (Norwegian case), OMI again shows overestimation above 20%. Although there is some variation in the mean bias among the stations, such variations are statistically insignificant for this case as well as for the cloud-free and snow-free (Paper I)

Under cloudy/snow-free conditions, OMI overestimates the ground measurements by 29%-73% for Norwegian data and under cloudy/snowy conditions, OMI both overestimates and underestimates the ground UVI depending on the station’s conditions. At those stations (Ny-Ålesund and Finse) where snow cover is extensive, the bias is negative, while for the others the bias is positive. The ANOVA’s result infers that while adding cloud or cloud/snow, some stations have significantly different biases while for the others the bias is statistically insignificant at 95% confidence level (Paper I). For example, under cloudy/snowy conditions, Ny-Ålesund and Finse both show a negative bias on OMI UVI and the difference of the biases are statistically insignificant, while for cloudy but no-snow conditions, the positive biases of these stations are statistically significant (Paper I).

Thus, the ANOVA test confirms that various geophysical parameters (surface albedo, reflectivity) play a role to make such a difference in OMI. In the next phase of the research, a linear multiple regression tool is used to analyse the eight different geophysical parameters including the two identified from the ANOVA using Norwegian data to explain the bias by considering all sky data. Based on the standardized β coefficients, the parameters or factors are ranked according to the importance for that station up to the significance level (Paper II).

UVI climatology for six sites of the Nepal Himalayas using the OMI data is studied after applying a correction factor. Diurnal variability using ground measurements shows that UVI as high as 16 is observed in one of the high altitude stations of Nepal (Paper IV). The monthly average UVI shows that UVI reaches extreme values in the high altitude stations during the summer months. Columnar ozone (TOC) climatology is also presented using TOMS and OMI data from November 1978 to March 2012 (Paper IV). The TOC has reached at minimum level in December (250-260 DU) and at maximum level in April or May (285-295 DU). The regression line plotted in the time series of monthly average data shows a negative trend in the columnar ozone during that period.