Estimation Of Land Surface Erythemal UVB Irradiance With MODIS Data

Accurate estimation of UVB radiation plays an important role in monitoring harmful effects on plant growth and human health.Most of the existing satellite products of UVB radiation are generated from Total Ozone Mapping Spectrometer(TOMS),Ozone Monitoring Instrument(OMI),Global Ozone Monitoring Experiment-2(GOME-2),with either spatial resolution or temporal resolution being too coarse to satisfy the need for high spatio-temporal monitoring of UVB.In this study,we proposed a method for estimating high-resolution all-sky erythemal UVB irradiance(UVER)from the Moderate Resolution Imaging Spectroradiometer(MODIS)and other ancillary information.We prototyped the approach with ground measurements at the Surface Radiation Budget Network(SURFRAD)stations and built the relationship between UVER and solar zenith angle,surface downward shortwave radiation(DSR),and total ozone column.We extended the approach and applied it to MODIS data,by estimating the instantaneous and 3-hourly DSR using a lookup table method and then estimating the instantaneous,3-hourly,and daily UVB radiation.Validation of the MODIS-based UVER estimation at the SURFRAD stations from 2001 to 2008 showed overall mean bias errors(MBEs)of 0.63 m W/m~2(0.62%)and 0.6 m W/m~2(0.93%)for the instantaneous and 3-hourly estimations,respectively.Additional validation covering a much wider spatial distribution was carried out at 28 UV-B Monitoring and Research Program(UVMRP)stations,8 Network for the Detection of Atmospheric Composition Change(NDACC)stations and 6Australian Radiation Protection and Nuclear Safety Agency(ARPANSA)stations.The results at UVMRP stations showed that the 3-hourly UVER estimations at 22 stations had MBE within±5%,and the instantaneous UVER estimations at 20 stations had MBE within±5%.We observed～30%and～20%relative root-mean-squared-errors(RMSEs)for the 3-hourly and instantaneous UVER estimation,respectively,among all the SURFRAD and UVMRP stations.Further verification at NDACC and ARPANSA stations indicated that our method was effective at other places around the world.The erythemal daily dose(EDD)derived from the3-hourly UVER estimation closely matched the ground measurements with small RMSEs(mostly between 10%and 20%).Comparisons with the EDD from OMI Level 3 product showed a strong spatial consistency.The main difference of the two EDD data on the four selected days in 2008 existed in regions with low latitude,which might be caused by the frequent occurrence of cloud or lack of enough data with low solar zenith angle when building our model.With high spatial and temporal resolutions as well as a potential for a long temporal coverage,our UVER estimation will provide accurate and timely information for UV monitoring and long-time climate applications.