Session B1: Atmospheric Effects, GNSS Remote Sensing, and Scientific Applications

Preliminary Results of Nearshore Ice and Water Level Monitoring in Arctic using Single-Antenna Ground-based GNSS-Reflectometry
Althaf Azeez and Jihye Park, Oregon State University
Location: Beacon B

Introduction
-----------------
Our research objective is to develop an efficient and reliable GNSS-Reflectometry (GNSS-R) based algorithms to detect the presence of sea ice, monitor water level variations, and estimate the horizontal and vertical motion of sea ice.
In this study, we investigate a system to continuously monitor the Arctic coastal marine environment by measuring the sea level variation and the formation of landfast ice throughout the year using ground-based GNSS-Reflectometry (GNSS-R) technique. The proposed system, with a single antenna, is advantageous due to it being a passive sensor and thus cost effective, and also having the flexibility to simultaneously monitor both the water level and sea ice.
This study presents the GNSS-R setup deployed in the field, along with the algorithms and initial results from our campaign.
Site, Sensors and Data Campaign
---------------------------------------------
Our test setup is deployed in Nome, Alaska. While the GNSS and tide gauge are
continuously operating (GNSS and tide gauge), in-situ measurements are also collected during the field campaign of 3-7 April 2024.
1) GNSS
We deployed a Septentrio PolaRx GNSS receiver with a zenith-facing antenna in Nome. It is continuously operating from October 2023 recording data from all major GNSS constellations. The Fresnel zone in the sea, from the GNSS satellites to this antenna, lies between the azimuth angles, 245 - 350 deg, and elevation angles, 3 to 15 deg. The signal-to-noise observable from GNSS satellites, whose line-of-sight falls within this azimuth and elevation angles ranges, will be processed to estimate water level/sea ice. The Fresnel zone in the land covers the azimuth angles (55-75 deg and 135-170 deg) and elevation angles (19-30 deg). The reflection zone in the land helps to determine the buildup of snow, which is explained later in the snow estimation section.
2) Laser level
In-situ measurements using laser, stakes and survey rods have been collected to measure water level, ice depth, and snow depth from the shore to about 100m toward the ocean during the field campaign in April 2024.
3) Tide gauge
NOME has a tide gauge which will be used as reference for water level.
A tide gauge and laser level in-situ measurements are used to validate the results
estimated using GNSS-R technique, which are described next.
State of the reflector: Confidence Level of Retrieval (CLR)
-----------------------------------------------------------------------------
A zenith-facing GNSS antenna receives both the direct and reflected signals thus forming an interference pattern which is clearly visible in the SNR observable ([1], [2]). Since the properties of the reflecting surface is captured by reflections coming from low elevation satellites, the trend of direct signals can be removed and the dominant peak frequency in power spectral density of the de-trended reflected signal is used to antenna-reflecting surface height, assuming the height remain constant for a certain duration. Using the dominant power frequency, the reflector height, in this case the water level, is estimated.
[2] introduced a numerical measure, Confidence Level of Retrieval (CLR), using the
dominant peak to ascertain the sea state. A higher value of CLR means a calm sea state as the reflections would be coherent and specular reflection whereas a low value of CLR, usually < 4, means a turbulent sea state. We have used CLR to identify sea states as well as investigate the impact of sea ice/snow.We have implemented an algorithm using de-trended signal’s dominant peak, and the preliminary results are demonstrated in the next section.
Preliminary results
-------------------------
1) Water level estimation
Water level (WL) is estimated using NOME’s GNSS-R from 1 Nov in 2023 to 14 Apr in 2024, and compared with the water level time series recorded at a nearby tide gauge. The vertical offset coming from the difference between the geodetic datum for GNSS and tide datum for a tide gauge is maintained as a constant, derived from processing a week’s data in October 2023. This was done to see the effect of snow/sea ice on the GNSS-R estimates that may increases over time during winter whereas the tide gauge estimates are immune to the buildup of snow/ice.
The deviation between tide gauge and GNSS-R is small (5 cm of less) for most of the time in Fall 2023 except for a few days, November 12,13, 20 and 28, when it is maximal. One reason for these deviations in water level estimates using GNSS-R are the severe weather conditions which might cause outliers due to weak reflected waves.
2) Snow/Sea Ice estimation
During winter, especially on Jan 12, there is a significant jump in water level estimated using GNSS-R compared to the tide gauge; and, as the winter progressed, this difference grew from 15 cm in January 2024 to 35 cm in Apr 2024. The positive jump in GNSS-R measurements compared to the tide gauge is due to the snow/sea ice buildup.
To verify the buildup of snow as well as to cross validate water level estimates with the tide gauge, multiple sessions of in-situ leveling from the shoreline to 100 m toward the ocean on the landfast ice was carried out using a laser level on 3-6 April. Water level estimated using the laser level matched well with the tide gauge, and the buildup of snow determined using the laser level is around 23 cm. The difference between GNSS-R estimate and tide gauge is ~ 30 cm, which is the thickness of sea ice and snow. Subtracting the snow depth of 23 cm from it gives the part of sea ice (~ 7 cm) which floats above the water.
3) Estimation of snow depth by looking inland
GNSS-R dominant peak technique shows that the reflector height is changing during
winter, but it does not tell if this is due to snow or sea ice. We could resolve this ambiguity by using the same GNSS-R sensor, but by looking inland, to measure the thickness of the snow. The snow depth thus calculated can be removed from the GNSS-R estimates from the ocean, thus providing sea ice level. To measure the in-land snow depth, first, we processed SNR measurements from land surfaces nearby by modifying the azimuth and elevation angles as explained in GNSS-R sensor section. In addition, the reflected data points from the land were fitted to a straight-line, instead of a second-order polynomial for water level measurements.
4) Sea State Inference using CLR
CLR plots from November to April 2024 showed significant decrease ( < 4) in November (12-15, 19-22, 28-30), December (6-8, 12-15), January (12-16) and March (19-21), that depicts a signature of turbulent sea state. For the dates in November and December, we have verified that there may be higher waves due to wind speeds exceeding 20 kts. The effect of Snow/sea ice on the magnitude of CLR needs to be investigated.
Conclusions
----------------
We present the techniques of estimating water level, snow and sea ice from NOME, Alaska using GNSS-R from our continuously operated setup as well as a dedicated field campaign. Before the buildup of snow/sea ice, the water level estimates conform well with the tide gauge estimates except for a few days when sea state was severe causing high errors in GNSS-R WL estimates. These severe events are detected using CLR, thus ensuring integrity of the GNSS-R estimates. During winters, GNSS-R estimates is a measure of water level, sea ice and snow. A technique is provided to estimate snow buildup by looking inland, and thus remove the thickness of the snow from GNSS-R output. To detect sea ice, as well as to measure its thickness, we are investigating various techniques: de-trended SNR amplitude,and indirectly on CLR; phase lag/lead of interference signal power spectral density; Delay Doppler Map, etc.
References
----------------
[1] Larson, K.M.; Small, E.E.; Gutmann, E.; Bilich, A.; Axelrad, P.; Braun, J. Using GPS
multipath to measure soil moisture fluctuations:Initial results. GPS Solut. 2008, 12, 173–177.
[2] Kim, S.-K.; Lee, E.; Park, J.; Shin, S. Feasibility Analysis of GNSS-Reflectometry for
Monitoring Coastal Hazards. Remote Sens. 2021, 13, 976.
https://doi.org/10.3390/rs13050976