Evaluation of MODIS 16-Days-Albedo Products Using ASTER Albedo Products for Homogeneous Semi-Arid Surfaces

Nowadays, evaluation and monitoring of change in environment and climate at the regional to global scale is an important issue for nowcasting and forecasting purposes. Surface Albedo is a key parameter in climate studies. Land surface Albedo is a product of Moderate Resolution Imaging Spector radiometer (MODIS) in global scale but in low resolution where the access to these products is free. However, Climate model outputs could be affected by errors in MODIS Albedo products. In this paper, the accuracy of Short wave Broadband Albedo estimated from MODIS global Albedo products (MCD43A3) is evaluated. The field used for this purpose is a plain homogeneous, semi-desert and non-vegetated region near the Qom-Tehran highway. Since MODIS and ASTER sensors are both on board of the same platform and due to the higher spatial and radiometric resolution of ASTER sensor, ASTER Albedo products are used as ground truth for evaluating MODIS Albedo Products.
The data used in this work consist of MODIS Albedo products, ASTER images, reflectance data, MODIS optical depth data, climatological data and land cover maps. To extract albedo from ASTER image, the geometric, atmospheric and radiometric corrections were implemented first. Then using the surface reflectance of ASTER, the shortwave broadband albedo was calculated. To calculate the MODIS actual albedo, the sun zenith angle image was co-registered with respect to ASTERs. Then after calculating fraction of diffused skylight, the MODIS actual albedo was calculated. Then to compare albedo images of ASTER and MODIS, the ASTER albedo products were resampled to 500m pixels. This was done by averaging all ASTER albedo values locating in a MODIS pixel.
In this work, using ASTER shortwave albedo, MODIS shortwave albedo was evaluated. Comparison between these two products showed that the ASTER albedo was more sensitive to the daily atmospheric conditions as well as surface cover than the MODIS albedo. However it is seen that the 16 days average of MODIS albedo reduces these effects. Also it is seen that except on Oct23, 2004, the MODIS shortwave albedos were always smaller than those of ASTERs. However, precipitation and increase of soil moisture reduces albedo values where this for ASTER albedos were more pronounced. Also it was seen that the RMSD between ASTER and MODIS shortwave albedo values on Sep14, 2003 and Oct23, 2004 were greater. The reason for this may come from the fact that on Oct23, 2004 we had raining and on Sep14, 2003 we had strong wind from nearby salt lake.
On the other hand, on Sep29, 2004 and Jul28, 2001, the atmospheric condition for ASTER acquisition day and for 16 days MODIS passage was the same where the difference between these two products minimized. Based on this, if the weather condition for the ASTER acquisition date and 16 days period of MODIS were the same, a correction of 4% increase to MODIS albedo values is suggested otherwise based on the severity of the weather conditions, this difference could be enormous and the MODIS albedo values are less reliable.
In this work, the actual MODIS shortwave albedo products were compared to those of ASTER shortwave albedos. The result of this investigation in years 2001 to 2004 showed that whenever the weather conditions in the ASTER acquisition date and 16 days period of MODIS were the same, a correction of 4% increase to MODIS albedo values is needed. Otherwise if the weather conditions of ASTER acquisition day were different from MODIS 16 day’s period, the difference between these two albedo products may reach to 9%, this difference could be enormous and the MODIS albedo values are less reliable. Due to the importance of surface albedo values in meteorological and climatological models, the error in estimation of albedo may cause erroneous conclusion in the environmental studies.