Combining active and passive microwave remote sensing measurements to detect soil thawing events

Variations of soil freeze/thaw (F/T) state greatly affect the energy balance, ecosystem, and hydrological cycles. However, the detection of soil F/T state using L-band passive microwave sensors is significantly restricted by the melting snow. In this study, we combine SMAP L-band brightness temperature (Tb) with ASACT C-band backscatter coefficient (σ⁰) to detect soil thawing events, aiming to reduce spurious detections caused by snowmelt. First, we calculate normalized polarization ratios (NPR) based on SMAP Tb. Then, we detect a decrease of σ⁰ together with an increase of NPR as a snowmelt event. Validating using soil and meteorological measurements at in situ stations across Alaska, we find that the snowmelt-induced σ⁰ decreasing and NPR increasing occurred at 3 out of 17 stations while diminished at other stations, which are possibly because the thawing of vegetation increases the backscatter. Furthermore, we use the diurnal difference of SMAP Tb, i.e., the difference between AM and PM overpasses, as the L-band signal is less influenced by vegetation than C-band, to differentiate snowmelt from soil thawing events. During the spring thaw, snow typically melts during the day and refreezes at night, resulting in Tb during AM overpasses show higher value compared to PM overpasses. This diurnal variation is more significant at the horizontal polarization (H-pol) than the vertical polarization (V-pol). We find that significant Tb diurnal differences at 12 out of 17 stations during the spring thaw. In dense vegetation covered regions (forest coverage > 30%), the Tb differences were 2.50 K at the V-pol and 5.47 K at the H-pol. In sparse vegetation covered regions (forest coverage < 30%), the mean Tb differences were 5.40 K for V-pol and 10.83 K for H-pol. Thus, the snowmelt can be determined based on the larger Tb diurnal variation at H-pol.