Figure 4. (Top) ALOS2 interferograms during the five periods, (a—e). (Bottom) Sentinel-1 stacked interferograms during the five periods, (f)—(j), derived so that the temporal coverage could nearly match those from (a) to (e); all the interferograms are overlaid on shaded relief maps. Warm and cold colors indicate LOS changes away from and toward the satellite, respectively.
We compared the ALOS2 and stacked Sentinel-1 interferograms for five periods (Figure 4) and assessed their differences (Figure 5). Despite differences in look directions both ALOS2 and Sentinel-1 indicated extensions in the LOS during periods (a, f) from the middle of June to the end of July and (b, g) from the end of July to the early October. Also, their deformation areas and amplitude were mostly consistent, suggesting that LOS changes were largely due to summer subsidence (see section 5.1 below for details). In terms of the spatial distribution of deformation signals, we noticed that the LOS changes over higher-elevation areas such as ridge and peak were insignificant, whereas the boundaries between the burned and unburned areas were clear. The north-western area, however, showed few LOS changes (see section 5.2 for the relationship between LOS changes and burn severity). During the period (c, h) from early October to early December both ALOS2 and Sentinel-1 indicated shortening in the LOS by an approximate 5 cm maximum, and the deformation areas and amplitude were quite similar. This observation presumably indicated frost heave in the early freezing period. In view of the previous three periods, both subsiding and uplifting areas were nearly the same. The following period (d, i) from early December to the middle of March also included the winter season with much colder air temperatures, but we did not observe any significant deformation signals, indicating that frost-heave virtually stopped in early December. While the good interferometric coherence during mid-winter was an unexpected result, we speculate that it could have been due to drier, lower amounts of snowfall.
In the periods (e, j) from the middle of March to early June, both ALOS2 and Sentinel-1 suffered from decorrelation, and we could not identify clear deformation signals. However, in light of Figure 6 below, each of the Sentinel-1 interferograms had overall good coherence with the exception of the data acquired in the middle of May. These observations suggested that the decorrelation may be attributable to the rapid changes on the ground surface during the initiation of the thawing season when the air temperature rises above the freezing point and the active layer begins to thaw.