Application of a multi-satellite data fusion-disaggregation algorithm to obtain daily cloud-free snow cover and snow water equivalent maps (Case study: Karun watershed, Ahwaz, Iran)

Snow cover, as an important component of land cover, is one of the most active natural and plays an important role in hydrological processes and climate. Variability in snow covered area has a significant influence on water and energy cycles, as well as socioeconomic and environmental repercussions. Frequent and long-term snow observation, accurate snow cover (SC) mapping and snow water equivalent (SWE) estimation are crucial for operational flood control, water delay planning, and resource management in snowmelt-dominated basins. Today, satellite-derived snow products obtained from visible and infrared imagery, as well as passive microwaves are available on the Internet, with few recently availability in near-real time . Optical sensors (e.g., Landsat, Advanced Very High Resolution Radiometer-AVHRR, Moderate Resolution Imaging Spectroradiometer-MODIS, Systeme Probatoire d’Observation de la Tarre-SPOT) have been well developed to provide snow information with good temporal and/or spatial resolution. But, cloud cover is a major factor in optical remote sensing that limits our capability to map the Earth’s surface. It is often not an easy task to collect a time-series cloud-free images for a particular area of interest, using optical remote sensing, which limits their wider applications for SC monitoring. Space-borne passive-microwave radiometers (e.g., Scanning Multichannel Microwave Radiometer-SMMR, Special Sensor Microwave/Imager-SSM/I, Advanced Microwave Scanning Radiometer-Earth Observing System-AMSR-E), can penetrate cloud to detect microwave energy emitted by snow and ice and provide information on SC and SWE. These passive microwave data are well suited to snow cover monitoring because of the characteristics such as all-weather imaging, large swath width with frequent overpass times. But, the coarse spatial resolution (e.g., 25 km of AMSR-E daily SWE product) hinders their applications in operational hydrological modeling. It seems that the combination of MODIS and AMSR-E can take advantage of both high spatial resolution of optical data and cloud transparency of passive microwave data. In this study, daily cloud-free SC and SWE maps at the 500-m resolution were produced for Karun watershed, Ahwaz, Iran (January 25–32, 2004). The daily MODIS-Terra, MODIS-Aqua, and AMSR-E snow data products were used via a fusion-disaggregation algorithm. The developed SC and SWE maps were evaluated, using total accuracy of snow mapping in clear-sky (Os) and all-sky (Oa) conditions, underestimation (UEc) and Overestimation (OEc) of snow covered area in clear-sky condition, snow accuracy in clear-sky (Sc) or in all-sky (Sa) conditions, and no snow accuracy in clear-sky (NSc) or in all-sky (NSa) conditions. The results of this study showed that the combination of MODIS-Terra and MODIS-Aqua considerably reduced the cloud coverage in such high resolution optical data. Although MAC-SC and MAC-SWE products have been developed to have 500 m spatial resolution, the massive and continuous cloud cover (larger than 25 km in size) in the MODIS-Terra and MODIS-Aqua and, hence, TAC products were simply replaced by the coarse AMSR-E pixels. In this case, although those cloud coverages were removed in the MAC-SC and MAC-SWE products, the actual resolution of the snow or no-snow pixels kept 25 km and such pixels had the false spatial resolution of 500 m. The SWE redistribution of AMSR-E based on MAC-SC products enhanced (to some extent) the spatial resolution of the AMSR-E SWE products. However, there was no measured data to evaluate the accuracy of the enhanced SWE products. It can be concluded that for pixels with scattered cloud cover (less than 25 km in size) in the TAC products, the MAC-SC and MAC-SWE products indeed improve the spatial resolution of those pixels to 500 m, while for massive cloud cover (larger than 25 km in size), the actual resolution of those pixels in the MAC-SC and MAC-SWE products are 25 km, even in 500 m pixel size. Despite of these limitations, the MAC-SC and MAC-SWE maps are suitable for hydrological, meteorological modeling on a daily basis in the study area.