Freeze and Thaw States Detection in High Latitude Inundated Areas Using High Resolution ALOS PALSAR Observations

Inundated surfaces in Northern latitudes experience freeze and thaw (FT) cycles seasonally. These surfaces are among the important sources of positive carbon and methane (CH4) feedback to the atmosphere as well as their crucial role in biogeochemical transitions, hydrology and prediction of boreal-arctic ecosystem. Wetlands, in particular, are the regions that contribute mostly as a CH4 source. In the past, remote sensing observations from satellites have shown a great potential capability in detecting freeze and thaw states of the surfaces especially in remote areas. Active and passive microwave observations are shown to be more sensitive to the change of surface state and are more promissing than other observations because they are less affected by the atmosphere. Active microwave measurements such as the Advanced Land Observing Satellite Phased Array L-Band SAR (ALOS PALSAR) can provide a viable higher resolution estimates of the inundated surfaces and their states than those from passive microwave brightness temperatures with coarser and higher temporal observations. Therefore, the link between active and passive estimates may potentially enhance our understanding with the advantages of higher spatial and temporal predictions. In this study, we utilize PALSAR ScanSAR mode data with more frequent temporal coverage of up to 40 days along with the static map dervied from Fine Beam Data to study the timing of the inundation for wetland classes as well as their FT states using data from year 2007 to 2010 period. A pixel-based and object oriented-based classification methods to derive freeze/thaw maps is applied. The dynamic inundation maps then are developed at 100 m resolution. JERS and PALSAR Fine Beam mode based static wetlands map and Landsat Based land cover data (NLCD) are used to train and assess the classification at high resolution along with other ancillary data sets. The developed thresholds are employed for the FT detection. Comparison of the results with ground observations with a wide ranges of surfaces, elevation, and vegetations classes show reliable agreements. Finally, the results are compared and evaluated against passive microwave estimates of FT states. The preliminery results suggests that PALSAR estimates can be used to downscale passive-based estimates to deliver high frequent estimates.