Soil moisturecontent (volumetric soil water content, skin-layer―one centimeter thickness) estimates are given inunit g/cm³ by the NASA-algorithm and by the JAXA-algorithm for AMSR-E [23] [24] . The original “at launch” algorithm was a 6-channel (6.925, 10.65, 18.70 GHz in horizontal and vertical polarization) iterative forward inversion of radiation transfer functions of per-channel observed brightness temperature with

Instantaneous Field of View (IFOV), Equator Crossing Time (ECT), all channels have horizontal and vertical polarization.

dependence on surface emissivity and reflectivity to the geophysical retrieval parameter vector estimates of soil moisture (m_s), vegetation water content (w_v) and land-surface temperature (T_s, a proxy for skin temperature) in a minimizing Chi-square process at grid points (i) given by

such that [25] [26] . Early post-launch saw high Radio Frequency Interference (RFI) at the 6.925 GHz channel, which necessitated replacement by the 10.65 GHz channel and furthermore necessitated usage of ancillary pre-existing datasets (look-up tables) to estimate T_s and its sensitivity to modifying factors of polarization and soil moisture indices in an iterative correction scheme [27] - [29] . The ancillary look-up tables are global coarse resolution class-parameterizations of vegetation cover (generalized density and type), shallow (nominal 0 to 0.30 m depth) soil texture, ocean and large in-land water masks and landform classes (including frozen ground, mountainous terrain, permanent ice and perennial snow). AMSR-E Level-3 datasets are gridded in the equal-area Ease-grid projection system using a spheroid reference at 25-by-25 kilometer grid intervals [18] . We re-project these using the World Geodetic System (WGS) reference ellipsoid WGS-84. For Earth science applications the AMSR-E datasets employing the NASA-algorithm are archived at the National Snow and Ice Data Center, University of Colorado.

The JAXA-algorithmretains the iterative inversion forward radiation transfer functions for soil moisture and vegetation watercontent, emissivity and reflectivity from multi-channel brightness temperatures and polarization ratios as the current NASA-algorithm [22] [30] - [33] . Land-surface temperature from MODIS is utilized in the algorithm-retrieval instead of prescribed annual averaged global look-up table scheme. Vegetation roughness which was previously estimated by generalized vegetation type from course resolution global look-up tables, a modifier of vegetation water content that attenuates soil moisture is estimated from the 10.65 and 36.5 GHz channels by incorporating a fractional coverage parameter from the Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) 16-day composite indices and polarization index, an index of soil moisture from the channel brightness temperatures [30] . AMSR-E Level-3 reprocessed (JAXA archive) datasets are global daily grids in Geographic projection system at 0.25-by-0.25 degree (~28-by- 28 km) with WGS-84 reference ellipsoid.

For the AMSR2 sensor JAXA has carried forward the algorithm development for reprocessing AMSR-E to process AMSR2 soil moisture content retrieval [22] [30] [31] . AMSR2 Level-3 global daily grids are in geographic projection at 0.25-by-0.25 degree (~28-by-28 km) relative to WGS-84 in units of moisture volume fraction m³/m³ as a percentage. AMSR-E (reprocess) and AMSR2 datasets are archived by JAXA for the GCOM- W1 mission [14] . We use a dimensionless volume fraction vol./vol. for comparisons.

Our sites within 1 to 5 km of the Arctic Ocean are within the coastal tundra of the continuous permafrost zone at Barrow and West Dock in Alaska, and at Bykovsky and Sukharna in Russia, Figure 4(a) (Barrow and West Dock) and Figure 4(b) (Bykovsky and Sukharna), respectively. Vegetation consists of mosses and dense-grass tussock

1 to 5 cm in height in the summer season. During summer the space between the dense-grass tops and the mineral-organic soil surface can be covered with standing water (Arctic wetlands).

Daily time series illustrate the remote-sensing microwave-based NASA- and JAXA-algorithmic retrieval soil moisture content performance relative to our VHF-probe sub-surface measurements.

The VHF-probes at 0.10 m depth (Figure 4(a), Barrow), 0.08 m depth (Figure 4(a), West Dock), 0.15 m depth (Figure 4(b), Bykovsky) and 0.20 m depth (Figure 4(b), Sukharna) all show a seasonal cycle of thawing beginning in April reaching a fully thawed state in June followed by freezing in late August. Soil moisture content measurements are near 0 vol./vol. (frozen) in April rising to near 0.40 vol./vol. (thawed) at Barrow, near 0.45 vol./vol. at West Dock, near 0.27 vol./vol. at Bykovsky and near 0.31 vol./vol. at Sukharna, in June. The variation of soil moisture content per-site is due to differences soil texture and organic material. Organic material is not accounted for in our calibrations of the VHF-probes.

Daily AMSR-E NASA-algorithm retrievals show no change of freeze-thaw state, although they show a cyclical variation from soil moisture content of 0.1 to 0.3 vol./vol., on-average (Figure 4(a), Barrow and West Dock, Figure 4(b), Bykovsky and Sukharna, respectively). The peak retrieval soil moisture content occurs during late- May into early-June. This is followed by a decrease in retrieval soil moisture content on through July, giving the daily 13:30 and 01:30 ECT time-series a punctuated cyclical response. June 2002 to October 2011 NASA-algo- rithm retrieval average soil moisture content is 0.12 (13:30 ECT) and 0.11 vol./vol. (01:30 ECT) with standard deviations of 0.05 vol./vol. at Barrow, 0.13 (13:30 ECT) and 0.12 vol./vol. (01:30 ECT) with standard deviation of 0.05 and 0.04 vol./vol. at West Dock, 0.17 (13:30 ECT) and 0.15 vol./vol. (01:30 ECT) with standard deviations of 0.03 vol./vol. at Bykovsky, respectively.

Daily AMSR-E JAXA-algorithm retrievals show change of freeze-thaw state that in most cases corresponds with our VHF-probe measurements on a daily basis (Figure 4(a), Barrow and West Dock, Figure 4(b), Bykovsky and Sukharna, respectively). On a season-average basis the closest correspondence occurs at West Dock. JAXA-algorithm retrieval averages over-perform relative to the VHF-probe averages at Barrow and Bykovsky. June/July retrieval maxima are near 0.6 vol./vol. at these sites.

Daily AMSR2 JAXA-algorithm retrievals at these sites show seasonality and similar characteristics to the AMSR-E JAXA-algorithm retrievals at Barrow (Figure 4(a)). The VHF-probe ranges in soil moisture content are much the same over the period of one season a previous. At Sukharna (Figure 4(b)) with a depth of 0.20 m the seasonal range in soil moisture content is from 0 to 0.31 vol./vol. with a thaw season average of 0.24 vol./vol. and standard deviation of 0.02 vol./vol. and similar to Bykovsky. The AMSR2 JAXA-algorithm retrieval thaw season range is 0.02 to 0.60 vol./vol. with averages of 0.39 (13:30 ECT) and 0.26 (01:30 ECT) vol./vol. with standard deviations of 0.17 and 0.15 vol./vol., respectively.

Co-location and co-time comparisons Figure 5(a) (Barrow and West Dock) and Figure 5(b) (Bykovsky and Sukharna) and Table 5 show the AMSR-E JAXA-algorithm retrieval versus measurement site R² with the highest values, though significance is modest. The AMSR-E NASA-algorithm retrieval versus measurement sites R² have the lowest values and, not significant (Figure 6(a) and Figure 6(b) and Table 5). This scoring by R² does indicate that the AMSR-E JAXA-algorithm is significantly different relative to the AMSR-E NASA-algorithm (Figure 7(a) and Figure 7(b) and Table 5). The AMSR2 JAXA-algorithm retrievals versus measurement site R² scores are similar to the AMSR-E JAXA scores, and of modest significance (Figure 8(a) and Figure 8(b) and Table 5).

Our sites Franklin Bluffs, Happy Valley and Galbraith Lake cross the Arctic coastal plain and Foothills also within the continuous permafrost zone, a region of short height vegetation of lowland-wetland to up-land tundra zones (Figure 4(a), Franklin Bluffs, Happy Valley and Galbraith Lake, respectively, and Table 4). During the short-summer thaw season our VHF-probe soil moisture contents show bimodality at each site. At Franklin Bluffs (Coastal Plain, low relief except for the Bluffs about 2 km east across the Sagavanirktok River and flood- plain, Figure 4(a), Franklin Bluffs) the thaw season average soil moisture content is near 0.44 vol./vol., standard deviation of 0.03 vol./vol. at 0.11 m depth with early June and August modes up to 0.51 vol./vol. in most years. At Happy Valley (lower Foothills, moderate relief, Figure 4(a), Happy Valley) the average soil moisture content is near 0.35 vol./vol., standard deviation of 0.04 vol./vol. at 0.11 m depth in a range of 0 to 0.40 vol./vol. with the last 3-years data showing a decline relative to the remainder of the series. Galbraith Lake (upper Foothills, glacial valley north-facing from the Brooks Range, Figure 4(a), Galbraith Lake) shows thaw season average soil moisture content near 0.58 vol./vol., standard deviation of 0.02 vol./vol. at 0.14 m depth and modes up to 0.62 vol./vol. in early June and August. The relative higher soil moisture content values at Galbraith Lake are in part an orographic effect on precipitation (Arctic Ocean source) by the Brooks Range.

The daily AMSR-E NASA-algorithm retrievals at the Coastal Plain and Foothills sites do not show freeze-

thaw state change from June 2002 to October 2011 (Figure 4(a), Franklin Bluffs, Happy Valley and Galbraith Lake, respectively). The daily 13:30 and 01:30 ECT time-series show a punctuated cyclical response like the time series from Barrow, although with missing values through winter months. The retrieval soil moisture content varies from minima near 0.1 vol./vol. to maxima near 0.36 vol./vol., on average. At Franklin Bluffs the

years 2003-4 and 2009 have bimodal maxima during May and July of those years (Figure 4(a), Franklin Bluffs). At Happy Valley the retrieval maxima soil moisture content display a range from 0.2 to 0.47 vol./vol. with years 2003 and 2009 having a bimodal punctuated response (Figure 4(a), Happy Valley). At Galbraith Lake the cyclical punctuated response of the time-series ranges from 0.1 to 0.43 vol./vol. in years 2002-3 and 2009 with bimodality, to 0.15 to 0.28 vol./vol. (single mode) during the other years (Figure 4(a), Galbraith Lake). NASA- algorithm retrieval average soil moisture content is 0.14 (13:30 ECT) and 0.12 vol./vol. (01:30 ECT) with standard deviation of 0.05 and 0.04 vol./vol. at Franklin Bluffs, 0.18 (13:30 ECT) and 0.16 vol./vol. (01:30 ECT) with standard deviation of 0.05 and 0.04 vol./vol. at Happy Valley, 0.18 (13:30 ECT) and 0.17 vol./vol. (01:30 ECT) with standard deviations of 0.03 vol./vol. at Galbraith Lake, respectively.

Daily AMSR-E JAXA-algorithm retrievals show change of freeze-thaw state that have good correspondence in the early-thaw period in June but lacks correspondence with the early-freeze period August with our VHF- probe measurements on a day-of-month basis (Figure 4(a), Franklin Bluffs, Happy Valley and Galbraith Lake, respectively). The miss-match is most apparent at the Galbraith Lake site on the upper Foothills. JAXA-algo- rithm retrieval averages are 0.09 (13:30 ECT) and 0.06 vol./vol. (01:30 ECT) vol./vol. with standard deviation of 0.03 and 0.02 vol./vol. with June/July maxima near 0.44 vol./vol. at Franklin Bluffs, decrease to 0.06 (13:30 ECT) and 0.04 vol./vol. (01:30 ECT) with standard deviation of 0.06 and 0.03 vol./vol. with the June/July maxima increase to near 0.57 vol./vol. at Happy Valley then at Galbraith Lakea reduction to 0.05 (13:30 ECT) and increase to 0.29 vol./vol. (01:30 ECT) with standard deviation 0.03 and 0.01 vol./vol. and June/July maxima near 0.38 vol./vol. at Franklin Bluffs, 0.57 vol./vol. at Happy Valley and 0.40 vol./vol. at Galbraith Lake, respectively. In many respects the JAXA-algorithm retrieval averages under-perform the NASA-algorithm averages at these sites. The JAXA-algorithm retrieval ranges over-perform the NASA-algorithm retrievals and under-per- form our VHF-probe measurement ranges. The daily AMSR2 JAXA-algorithm retrieval soil moisture content display a decrease, on average, relative to those of AMSR-E at the sites and retain values near zero during winter months.

The Arctic Russia sites are within floodplains and lowlands of the Allaiha and Kolyma Rivers (Figure 4(b), Allaiha, and Cherskii, Amboliha and Pleistocene Park, respectively, and Table 4). Ground vegetation consists of

tundra tussock grasses and boreal larch forest, meadow and shrub lands. The active-layer soils can be at times near saturation as in floodplains and well drained in the forest and shrub sites. The site at Pleistocene Park, a research and ecosystem restoration experiment station 5 km south of Chersky is within the boreal steppe ecosystem of the Kolyma River lowland [48] .

At Allaiha (Figure 4(b), Allaiha) site the VHF-probe soil moisture content at 0.10 m depth has a seasonal range of 0 to 0.26 vol./vol. with a thaw season average of 0.25 vol./vol. with a standard deviation of 0.01 vol./vol. The AMSR-E NASA-algorithm retrievals show soil moisture content range of 0.2 to 0.3 vol./vol. with averages of 0.21 (13:30 ECT) and 0.2 vol./vol. (01:30 ECT) with standard deviations of 0.05 and 0.04 vol./vol., respectively, over the last days of AMSR-E operations from 9 September to 3 October 2011 corresponding with the beginning of the VHF-probe data acquisitions. The AMSR-E JAXA-algorithm retrievals show soil moisture content range of 0.11 to 0.49 vol./vol. with averages of 0.28 (13:30 ECT) and 0.18 (01:30 ECT) with standard deviations of 0.14 and 0.12 vol./vol., respectively. The AMSR2 JAXA-algorithm retrievals show soil moisture range of 0.09 to 0.59 vol./vol. with near identical averages and standard deviations at the respective ECTs.

The Cherskii site (Figure 4(b), Cherskii) is near the settlement of Chersky and the Kolyma River within tundra tussock and floodplain shrub. The VHF-probe soil moisture content at 0.12 m depth shows a thaw season range of 0 to 0.40 vol./vol. with thaw season average of 0.40 and standard deviation of 0.01 vol./vol. during four-years operation form May 2001 to July 2005. During the thaw season the daily reading have bimodal character with early June mode near 0.40 vol./vol. followed by a second mode in early July of 0.23 to 0.39 vol./vol. soil moisture content. The bimodal character may be due to spring flooding and vegetation response in addition to vegetation heterogeneity given the IFOV at this site. The AMSR-E NASA-algorithm shows a thaw season range of 0.1 to 0.2 vol./vol. from July 2002 through August 2005. The thaw season NASA-algorithm retrieval soil moisture content shows averages of 0.06 (13:30 ECT) and 0.08 vol./vol. (01:30 ECT) with standard deviations of 0.03 vol./vol. over this period. The AMSR-E JAXA-algorithm retrieval soil moisture content shows a thaw season range of 0.02 to 0.40 vol./vol. with of average 0.08 and standard deviation of 0.05 vol./vol. at both ECTs.

The Amboliha site (Figure 4(b), Amboliha) is within a sector of the Kolyma River floodplain where thaw lakes are present. The VHF-probe at 0.10 m depth shows a thaw season soil moisture content range of 0 to 0.24 vol./vol. and thaw season average of 0.24 with standard deviation of 0.01 vol./vol. for one season of data during summer 2013 to the early May 2014. AMSR2 JAXA-algorithm retrievals show the 2013 season soil moisture content range of 0.02 to 0.59 vol./vol. with averages of 0.20 (13:30 ECT) and 0.23 (01:30 ECT) and standard deviations of 0.12 and 0.16 vol./vol., respectively. We note that the AMSR2 JAXA-algorithm retrievals freeze- up timing, i.e. when the retrieval soil moisture decreases to near zero occurs almost two weeks earlier than that shown by the VHF-probe soil moisture content readings. This indicates a miss-match in the vegetation indices adapted from the MODIS 16-day NDVI product used by the JAXA-algorithm.

The Pleistocene Park site (Figure 4(b), Pleistocene Park) is within the Kolyma River floodplain south of Amboliha. The VHF-probe soil moisture content at 0.10 m depth shows a thaw season range of 0 to 0.43 vol./vol. from July 2013 through July 2014. AMSR2 JAXA-algorithm retrieval soil moisture content shows a thaw season range of 0.02 to 0.60 vol./vol., with thaw season averages of 0.18 (13:30 ECT) and 0.16 (01:30 ECT) vol./vol. with standard deviations of 0.11 vol./vol., respectively, over the two-year period.

Co-location and co-time comparisons Figure 5(a) (Franklin Bluffs, Happy Valley and Galbraith Lake) and Figure 5(b) (Allaiha, Cherskii, Amboliha, and Pleistocene Park) and Table 5 show the AMSR-E JAXA-algo- rithm retrieval versus measurement site R² with the highest values, modest significance, at Franklin Bluffs, Galbraith Lake and Cherskii. Happy Valley and Allaiha have low R² values that are not significant. The AMSR-E NASA-algorithm retrieval versus measurement site R² shows very low values and no significance (Figure 6(a) and Figure 6(b) and Table 5). At Alliaha site both the AMSR-E JAXA and NASA-retrieval versus measurement R² scores indicate no correlation. This scoring by R² does indicate that the AMSR-E JAXA-algorithm is significantly different relative to the AMSR-E NASA-algorithm (Figure 7(a) and Figure 7(b) and Table 5). The AMSR2 JAXA-algorithm retrievals versus measurement site R² scores at Franklin Bluffs, Happy Valley, Galbraith Lake, Alliaha and Cherskii (Figure 7(a) and Figure 7(b) and Table 5) indicate no correlation.

The Ivotuk and Chandalar Shelf sites (Figure 4(a), Ivotuk and Chandalar, Table 4), continuous permafrost zoneexhibit similar upland tundra vegetation cover with contrasting setting; one on a north-facing broad-flat upland flanking north on the Brooks Range and the other in a south-facing central broad-flat upland valley within the Brooks Range south of Atigun Pass. These sites show the highest VHF-probe measured soil moisture content. Thaw is rapid in late-May at Ivotuk with the June/July bimodal maxima near 0.88 vol./vol. at 0.16 m depth and a sharp drop in the last year of data, 2013 to near 0.5 without the bimodal character. After the rapid late-May thaw at Chandalar Shelf the June/July soil moisture content maxima reach near 0.96 vol./vol. at 0.07 m depth with a variation over the years from near 0.96 vol./vol. in 2002 to 0.78 vol./vol. in 2013. Over the data period at Chandalar Shelf the maxima display both a bimodal and a single modal character in VHF-probe soil moisture content measurements.

The daily AMSR-E NASA-algorithm retrievals at the Ivotuk and Chandalar Shelf sites (Figure 4(a), Ivotuk and Chandalar, Table 4) show a small range cyclical variation of soil moisture content during the short summers from June 2002 to October 2011 that does not correspond to the VHF-probe soil moisture content variation. Retrieval values during winter months are missing and summer variation of soil moisture content range from 0.2 to 0.47 vol./vol. at Ivotuk and and 0.1 to 0.34 vol./vol. at Chandler Shelf. NASA-algorithm retrieval average soil moisture content is 0.16 (13:30 ECT) and 0.16 vol./vol. (01:30 ECT) with standard deviations of 0.04 vol./vol. at Ivotuk and 0.15 (13:30 ECT) and 0.16vol./vol. (01:30 ECT) with standard deviations of 0.02vol./vol. at Chandalar Shelf. Daily AMSR-E JAXA-algorithm retrieval average soil moisture content is 0.08 (13:30 ECT) and 0.05 vol./vol. (01:30 ECT) with standard deviation 0.09 and 0.05 vol./vol. at the Ivotuk and 0.05 (13:30 ECT) and 0.04 vol./vol. (01:30 ECT) with standard deviation 0.03 and 0.02 vol./vol. at the Chandalar Shelf. At the Chandalar Shelf the NASA-algorithm retrieval soil moisture over-perform the JAXA-algorithm retrieval soil moisture by about 0.1 vol./vol. yet both algorithm retrieval soil moisture substantially under-perform relative the VHF-probe soil moisture content measurements. The AMSR2 JAXA-algorithm retrieval soil moisture content for years 2012 and 2013 show a slight, on-average increase relative to those of AMSR-E.

The JAXA-algorithm retrievals show freeze-thaw state changes such that the early-thaw period has relatively good correspondence with the VHF-probe measurements on a day-of-month basis but the early-freeze period has poor correspondence with the VHF-probes at both sites (Figure 4(a) and Table 4). NASA-algorithm retrieval seasonal maxima are up to 0.47 vol./vol. whereas the JAXA-algorithm retrieval seasonal maxima are near 0.59 vol./vol. over the period from June 2002 to October 2011. At Ivotuk June 2009 shows the upper limit of the maxima for both algorithms retrieval soil moisture content at their ECTs, 0.47 and 0.4 vol./vol. (13:30 and 01:30) NASA and 0.59 and 0.54 vol./vol. (13:30 and 01:30) JAXA. This particular feature of the retrievals does not correspond to VHF-probe soil moisture content measurements. The daily AMSR2 JAXA-algorithm retrievals from July 2012 trough December 2013 continue the AMSR-E JAXA-algorithm retrieval soil moisture content values and seasonal character.

Co-location and co-time comparisons Figure 5(a) (Ivotuk and Chandalar) and Table 5 show the AMSR-E JAXA-algorithm retrieval versus measurement site R² with the highest value and modest significance at the Chandalar site. The AMSR-E JAXA-algorithm retrieval versus measurement at Ivotuk shows low R² that is not significant (Figure 5(a) and Table 5). The AMSR-E NASA-algorithm retrieval versus measurement site R² show low values and no significance at both Ivotuk and Chandalar sites (Figure 6(a) and Table 5). The AMSR2 JAXA-algorithm retrievals versus measurement site R² scores have the lowest values and no significance (Fig- ure 8(a) and Table 5).

The Council site on the Seward Peninsula is located about 6 km northwest of Council (town) on a low-to-mod- erate reliefdense hardwood shrub, about 1.5 m height, dominated tussock tundra terrain. The site location is within the discontinuous permafrost zone. Winter VHF-probe soil moisture content measurement at 0.16 m depth has values near 0.08 vol./vol. and temperature as low as −1.3˚C. The Seward Peninsula is well known for its geothermal hot springs, Quaternary basalt flows and the Pleistocene hydromagmatic Espenberg Maars [49] . VHP-probe soil moisture content measurement show the early-thaw soil moisture rapidly rising to maxima of 0.56 to 0.62 vol./vol. then arapid drop to values near 0.4 vol./vol. that persists into November, likely a vegetation response and followed by another rapid drop in the values as the active-layer re-freezes (Figure 4(a) and Table 4). Our VHF-probe soil moisture content measurements span 2002 through 2010, due to access difficulty. To compare with the AMSR2 retrievals we compose Julian-day averages from the preceding series.

At Council site the daily AMSR-E NASA-algorithm retrievals show a cyclical response during thaw-months from near 0.11 to near 0.27 vol./vol. (13:30 ECT) and from as low as 0.03 to near 0.22 vol./vol. (01:30 ECT) with a maxima of 0.38 (13:30 ECT, May 2010) and 0.3 vol./vol. (01:30 ECT, April 2009) over the period from June 2002 to October 2011 (Figure 4(a) and Table 4). NASA-algorithm retrieval average soil moisture content is 0.14 (13:30 ECT) and 0.14 vol./vol. (01:30 ECT) with standard deviation 0.03 and 0.02 vol./vol. at this upland shrub-tussock site. The daily AMSR-E JAXA-algorithm retrievals show a soil moisture content show seasonal thaw response very similar to the VHF-probe measurements from a low near 0.02 up to near 0.56 vol./vol. (13:30 ECT) and from a low near 0.03 to up to near 0.5 vol./vol. (01:30 ECT). JAXA-algorithm retrieval average soil moisture content is 0.11 (13:30 ECT) and 0.09 vol./vol. (01:30 ECT) with standard deviation 0.12 and 0.1 vol./vol. over the period. While the thaw timing shown in the JAXA-algorithm retrievals have good correspondence with the VHF-probe measurements the timing of the freeze appears aliased, by turning the retrieval seasonal response into a spike-like function. Thus the JAXA-algorithm retrieval freeze response occurs during Augusts whereas the VHF-probe measurement response occurs during Novembers. The AMSR2 JAXA-algo- rithm retrieval soil moisture content has very similar seasonal response and value magnitudes as those of the AMSR-E JAXA-algorithm retrievals at this site.

Co-location and co-time comparisons Figure 5(a) (Council) and Table 5 show the AMSR-E JAXA-algorithm retrieval versus measurement site R² with a value and low significance. The AMSR-E NASA-algorithm retrieval versus measurement site R² shows no correlation (Figure 6(a) and Table 5). The AMSR2 JAXA-algorithm retrievals versus measurement site R² score lower relative to the AMSR-E JAXA score (Figure 8(a) and Table 5).

Southwest of the University of Alaska Fairbanks in the Tanana Flats River region of the Yukon River watershed is the Bonanza Creek Long-Term Ecological Research (LTER) site³. Our permafrost-monitoring site in the discontinuous permafrost zone is within an upland Spruce forest with moderate topographic relief and gullies on the north flank of the Tanana River and experienced wildfire burning in 1983. Thaw occurs rapidly over a few day in early May and freeze occurs rapidly over a few days in late September. 2004 was noteworthy for the early thaw in April and late freeze extending into November. VHF-probe soil moisture content measurement reach maxima of 0.40 vol./vol. at 0.19 m depth in June and some years have a reduction occur over July and August before the rapid freeze. Outages occurred in 2004 through 2005 and 2007 into October 2009 (Figure 4(b) and Table 4).

The daily AMSR-E NASA-algorithm retrievals show a soil moisture content cyclical response of 0.11 to 0.22 vol./vol. (13:30 ECT) and 0.16 to 0.17 vol./vol. (01:30 ECT) over the period from June 2002 to October 2011. Retrieval values during winter days are missing (Fig. 4b and Table 4). NASA-algorithm retrieval average soil moisture content is 0.14 (13:30 ECT) and 0.14 vol./vol. (01:30 ECT) with standard deviation 0.02 vol./vol. and 0.01 vol./vol., respectively, at this boreal forest site. NASA-algorithm retrieval maxima soil moisture content is 0.23 (13:30 ECT) and 0.29 vol./vol. (01:30 ECT), 0.26 vol./vol., on average. The daily AMSR-E JAXA-algo- rithm retrievals show a soil moisture content seasonal response during thaw of 0.03 to 0.58 vol./vol. (13:30 ECT) and 0.03 to 0.59 vol./vol. (01:30 ECT). JAXA-algorithm retrieval average soil moisture content is 0.17 (13:30 ECT) and 0.16 vol./vol. (01:30 ECT) with standard deviation 0.16 and 0.18 vol./vol., respectively. The rapid early thaw corresponds well with the timing of thaw by the VHF-probe measurements with a relative over-per- formance of maxima by about 0.20 vol./vol. over the period. The JAXA-algorithm retrieval timing of the early freeze occurs about 15 days ahead of the early freeze timing as measured by the VHF-probe measurements. The AMSR2 JAXA-algorithm retrieval soil moisture content has very similar characters and thaw-freeze responses as those of the AMSR-E retrievals at this site.

Co-location and co-time comparisons Figure 5(b) (Council) and Table 5 show the AMSR-E JAXA-algorithm retrieval versus measurement site R² a high value of modest significance. The AMSR-E NASA-algorithm retrieval versus measurement site R² shows no correlation. The AMSR2 JAXA-algorithm retrievals versus measurement site R² score is the same as AMSR-E JAXA, and of modest significance.

Our permafrost-monitoring site in the discontinuous?sporadic permafrost zone of south-central Alaska is at the High-Frequency Active Auroral Research Program (HAARP) site (Figure 4(b) and Table 4). On the east side of the HAARP facility our permafrost-monitoring site is within the spruce-hardwood forest with understory vegetation of willow and alder with litter-covered ground of lichen and moss. Our shallowest VHF-probe is at 0.35 m depth to avoid shallow roots and the surface organic mat. Winter VFH-probe soil moisture content measurements show a range from 0 to 0.1 vol./vol. and typical minima from 0.01 to 0.04 vol./vol. over the period from August 2004 to October 2011. Thaw response begins in mid-March and soil moisture content rapidly increases in value to maxima near 0.47 and as high as 0.53 vol./vol. with a quick reduction to a level near 0.41 to 0.45 vol./vol. that persists from July to September. This response is similar to our measurements at the Council site and likely a vegetation response imprinted on the soil moisture content.

The daily AMSR-E NASA-algorithm retrievals show a soil moisture content cyclical response, similar to the Council site, of 0.13to 0.26 vol./vol. (13:30 ECT) and 0.12 to 0.15 vol./vol. (01:30 ECT) over the period from June 2002 to October 2011 (Figure 4(b) and Table 4). NASA-algorithm retrieval average soil moisture content is 0.13 (13:30 ECT and 01:30 ECT) with standard deviation 0.02 and 0.01 vol./vol. at this forest site. NASA- algorithm retrieval maxima soil moisture content is 0.26 (13:30 ECT) and 0.21 vol./vol. (01:30 ECT). The daily AMSR-E JAXA-algorithm retrievals show a soil moisture content seasonal response during thaw, similar to the Council site, of 0.03 to 0.56 vol./vol. (13:30 ECT) with four of the maxima near 0.6 vol./vol., and 0.03 to 0.57 vol./vol. (01:30 ECT) over the period. JAXA-algorithm retrieval average soil moisture content is 0.12 (13:30 ECT) and 0.11 vol./vol. (01:30 ECT) with standard deviation 0.11 and 0.13 vol./vol. over the period. The rapid early-thaw corresponds well with the timing of thaw by the VHF-probe measurements with a relative over-per- formance of maxima as much as 0.20 vol./vol. relative to the VHF-probe measurements over the period. The JAXA-algorithm retrieval timing of the freeze response is similar to that found at Council site and occurs about 60 days ahead of the early freeze timing as measured by the VHF-probe measurements. The AMSR2 JAXA- algorithm retrieval soil moisture content has very similar characters and thaw-freeze responses as those of the AMSR-E retrievals at this site.

Co-location and co-time comparisons Figure 5(b) (HAARP) and Table 5 show the AMSR-E JAXA-algo- rithm retrieval versus measurement site R² with a low value and no significance. The AMSR-E NASA-algorithm retrieval versus measurement site R² has no correlation (Figure 6(b) and Table 5). The AMSR2 JAXA-algorithm retrieval versus measurement site R² score has the lowest value and no significance (Figure 8(b) and Table 5).

Least-squares Analysis of Variance of all the time-series indicates no significant trends in soil moisture content by algorithmic retrieval or ground measurement on a site-by-site basis. In addition there are no significant correlations of soil moisture content by algorithmic retrieval to ground measurement on a site-by-site basis. The ground measurements display a latitudinal variation in their maxima of soil moisture content with the low- maxima values at stations near the northern coasts of the Arctic Ocean and the high-maxima values at stations near the Brooks Range in Alaska. This is likely an effect of topography. The NASA-algorithm retrievals do not display seasonality, i.e. a change of the freeze/thaw state. Also the NASA-algorithm retrievals under-perform, i.e. are less than, relative to ground measurements at all stations in Alaska and Russia. The JAXA-algorithm retrievals by comparison do show seasonality, which does correspond to the seasonality displayed by the ground measurements. The JAXA-algorithm retrievals do have late-season issues in performance at interior Alaska boreal forest sites relative to ground measurements, and this is likely a vegetation parameter artifact of the retrieval algorithm. The JAXA-algorithm retrievals do over-perform, i.e. are greater than, ground measurements at sites in Alaska and Russia with tundra vegetation, likely a vegetation parameter artifact of the retrieval algorithm an under-perform elsewhere, in part, likely an effect of topography, mixed-vegetation types and the IFOV per channel of the AMSR-E and AMSR2 sensors.

NASA-algorithm for soil moisture content by AMSR-E under-performs, i.e. the retrieval value is less than the ground-measured value, at all locations in Alaska and Russia, Figure 4 through Figure 8. The under-perfor- mance is likely a function of, assumed T_s and vegetation parameters from look-up table values which do not correspond to actual environment conditions. The non-seasonality (freeze/thaw and vegetation response) of the NASA-algorithm and mismatch with ground measurements is evident. Missing daily retrieval soil moisture content values during winter is likely due to an arbitrary land-mask that presumes a frozen state but does not correspond to actual environmental condition. JAXA-algorithm for soil moisture content by AMSR-E and AMSR2 show improvement in performance at all locations in Alaska and Russia relative to ground-measure- ments yet over-performance at tundra sites is evident. This is likely due to incorporation of MODIS T_s as a proxy for skin temperature and MODIS NDVI vegetation parameters, which have a closer correspondence to actual environmental conditions. Use of MODIS derived parameters restores seasonality into the soil moisture content. At boreal forest sites in Alaska the JAXA-algorithm retrievals have performance issue with late-season vegetation from MODIS NDVI 16-day composite and this produces a drop-off in retrieval soil moisture content that does not correspond to ground measurements. For the northern high latitudes the MODIS NDVI 16-day composite produces aliased timing of brown-up, which is likely a miss-match presumed vegetation type from ancillary data used in the processing. An additional influence of the late-season as well as early-season miss- match derives from yearly variations in the snow cover whose microwave emissions can mask those of skin temperature, ground vegetation and soil water content [38] .

The performance of the JAXA-algorithm retrievals shows landscape-scale heterogeneity of the microwave channels IFOV (10 s of kilometers sensing-scale) versus the VHF-probeground measurements (centimeter level sensing-scale). This heterogeneity is relatively less for coastal sites nearest the Arctic Ocean in Alaska and Russia (Figure 4(a), Barrow and West Dock and Figure 4(b), Bykovsky, Cherskii and Sukharna) and increases to the south in Alaska where micrometeorology, topography and vegetation types are more varied and mixed relatively to the microwave IFOV.

The period of comparison for the AMSR-E NASA and JAXA-derived algorithm retrievals is from June 2002 to October 2011. At our Alaska and Russia sites there are no significant trends of soil moisture content. There are significant variations of soil moisture content that derive from environmental setting, particularly orographic effects north-to-south across the Brooks Range and soil composition.

Both NASA- and JAXA-algorithm derived soil moisture content shows spatially co-located and significant emergent trends for daytime and nighttime retrievals, Figure 9 and Figure 10. The trends we find have spatial correspondence to locations of civilian and governmental radar facilities and telecommunications infrastructures and population areas and oil-gas exploration activity particularly evident along the Alaska north coast. It is apparent that since launch in 2002 there has been an increase in terrestrial telecommunications infrastructure using 10.65 GHz as well as other microwave-frequencies in Alaska.

It has been recognized that shortly after the start of operations of the AMSR-E mission on NASA-Aqua contamination by terrestrial radio-frequency interference at 6.925 GHz (C-Band), which necessitated the use of the 10.65 GHz (X-Band) for the primary and soil moisture retrieval [26] . After 2-years of operation significant contamination was apparent in the 10.65 GHz channel though mostly at that time confined to parts of the British Isles, Netherlands, Italy, Southeast Asia and Japan [27] [29] .

Our results show spatial trends of nighttime-retrievals have greater magnitude than those of daytime-retrievals, Figure 10. While the sun is a weak source of radiation at 10.65 GHz the galaxy is active at 10.65 GHz [50] [51] . Therefore the nighttime increase in “soil moisture content” based on is likely an expression of galactic radio-frequency interference reflected by the land surface and received by the X-Band feed-horn of AMSR-E and AMSR2. Our investigation of the algorithm retrievals through summer 2014 shows continuity in characteristics of the AMSR-E (re-processed) at all of our sites in Alaska and Russia. Further investigation of AMSR2 retrievals for analysis of radio frequency interference is needed.

Future development microwave retrieval algorithms for near-surface soil moisture content and other important environmental parameters need to be aware of the growing effects of radio-frequency interference. As urbanization effects, e.g. the urban heat island effect contaminates near-surface temperature records, radio-frequency interference from man-made and galactic-sources is a form of contamination affecting active and passive microwave receivers and their retrieval products.