EVALUATING ALTERNATIVE FORMULATIONS OF THE SMAP LEVEL-4 CARBON MODEL FOR THE NORTH AMERICAN ARCTIC–SUBARCTIC DURING THE GROWING SEASON
Marzo 25, 2026A PRELIMINARY INVESTIGATION INTO THE POTENTIAL OF HY-4A/MICAP DATA FOR SOIL MOISTURE RETRIEVAL
Marzo 25, 2026A. Colliander1, M. Holmberg1, K. Rautiainen1, J. Lemmetyinen1, A. Kontu1, J. Pulliainen1
1Finnish Meteorological Institute
L-band (~1.4 GHz) radiometry offers unique information on the terrestrial water and carbon cycles. Because the L-band brightness temperature (TB) responds to soil permittivity (soil moisture and the soil freeze/thaw state), to vegetation optical depth (VOD), (biomass, vegetation water content, and vegetation freeze state), and to the evolution of a snow layer (wet snow, peak snow water equivalent (SWE), and snow density) it can answer questions spanning ecology, hydrology, climate science, and soil–atmosphere coupling, motivating rigorous evaluation at sites where satellite, tower, and in situ observations can be coherently integrated. The Finnish Meteorological Institute’s Arctic Space Centre (FMI-ARC) in Sodankylä, Finland, provides a natural laboratory to address this need, combining multiple stations and instruments for soil permittivity and volumetric water content; tree permittivity; biomass inventories; snow depth, density, and SWE; and air and soil temperature. Co-located tower-based radiometers deliver high-fidelity, multi-angle time series with matched reference measurements, enabling bridging between point-scale observations and satellite footprints. Since 2009, L-band radiometers have been deployed at FMI-ARC overlooking diverse ecotypes, including boreal forests and wetlands. Measurements at forested sites include observations from both above and below the forest canopy, allowing for isolating the canopy contribution on the L-band signal. Together, these assets enable us to disentangle soil, vegetation, and snow contributions to the L-band signal across seasons and hydroclimatic extremes.
We analyse a decade (2015–2025) of SMAP L-band TB over Sodankylä alongside the mission’s soil moisture (SM) and VOD retrievals. Building on the ground and tower records, we recalibrate the retrieval algorithm for boreal and arctic conditions. The workflow (i) harmonizes reference measurements and quality flags; (ii) performs precise footprint matching and polarization/angle normalization of SMAP TB; (iii) tunes key forward-model parameters governing soil permittivity, surface roughness, vegetation attenuation/emission, and snow layering; and (iv) evaluates performance across freeze/thaw transitions, wet-snow events, growing-season stages, and periods of drought or heat stress. Uncertainty is characterized via error propagation and consistency checks between independent reference streams (soil sensors, tree permittivity, biomass inventories, and snow observations) and the tower radiometers.
To place L-band in a broader microwave context, we compare SMAP-based SM and VOD with higher-frequency observations. The cross-frequency analysis diagnoses frequency-dependent sensitivities and ambiguities, highlighting conditions where L-band uniquely penetrates canopies and dry snow, versus regimes where higher frequencies add constraining power (e.g., shallow surface wetness or rapidly evolving snow microstructure). Insights from these comparisons inform the development of multi-frequency retrieval strategies.
The results demonstrate that recalibrating the SM retrieval parameters substantially reduces SMAP retrieval errors. The original product showed a mean difference (MD) of 0.086 m³/m³, a standard deviation (STD) of 0.057 m³/m³, and a Pearson correlation (R) of 0.35. After recalibration, the MD was eliminated, with an improved STD of 0.025 m³/m³ and R of 0.75. These improvements are primarily due to correcting an overestimated vegetation attenuation and the surface roughness parameter. Furthermore, retrievals of frozen-season soil permittivity and VOD indicate potential for capturing soil and vegetation states beyond the binary freeze/thaw classification. This suggests new opportunities for linking remote sensing measurements with soil biochemical and biophysical activity as well as vegetation processes. The seasonal transition from snow-covered to snow-free conditions shows a highly predictable pattern, corresponding to peak SWE and the degree of soil saturation following snowmelt. We will present these findings in detail and discuss their implications for monitoring critical environmental parameters in the Arctic-Boreal zone.
