Abstract
A prototype algorithm for hemispheric scale detection of autumn soil freezing using space-borne L-band passive microwave observations is presented. The methodology is based on earlier empirical and theoretical studies of L-band emission properties of freezing and thawing soils. We expand a method originally developed for soil freeze–thaw (F/T) state detection from L-band tower based observations to satellite scale, applying observations from the European Space Agency's Soil Moisture and Ocean Salinity (SMOS) mission. The developed algorithm is based on first establishing spatially variable thresholds for L-band brightness temperatures representing frozen and thawed states of soil, and comparing these to current values of different indicators of soil freezing, calculated based on observed brightness temperature at different polarizations and incidence angles. An exponential relation between the freezing indicators and the depth of soil frost is developed based on a large amount of manual soil frost tube observations across Finland. An additional processing filter based on observed physical temperature and snow cover information is used to flag obvious F/T detection errors. The estimated soil F/T-states provided in this study are limited to the autumn freezing period, as melting snow in spring effectively prevents acquisition of information from the soil surface using microwaves for large areas in Northern latitudes. The F/T estimate is produced as daily information and provided in the equal-area scalable Earth (EASE) grid. Soil F/T-state is categorized into three discrete levels: ‘frozen’, ‘partially frozen’, and ‘thawed’, and accompanied with a quality data matrix estimating the data reliability for each freezing season separately. Comparisons to in situ data were conducted at 10 different locations in Finland, Northern America and Siberia. These comparison results indicate that the onset of autumn soil freezing can be estimated from SMOS observations to within 1 to 14 days, depending on the freezing indicator applied and the in situ data used in comparison. Although the initial results are encouraging, more comprehensive assessment of SMOS based soil F/T estimates still requires further comparison to other reference sites, particularly to sites with measurements available for all locally representative land cover types, as well as other satellite-based soil freezing products.
Highlights
The European Space Agency's SMOS (Soil Moisture and Ocean Salinity) satellite mission (Kerr et al, 2010) provides regular low-frequency (L-band, 1–2 GHz) passive microwave observations from space
Two soil F/T-state estimates are generated on a daily basis for the Northern Hemisphere equal-area scalable Earth (EASE) grid 1.0 data format: one categorizing the soil F/T-state using brightness temperatures at vertical polarization only (FFrel,X,orb with X = V corresponding to definition (1)) and the one based on normalized polarization ratio (FFrel,X,orb with X = NPR corresponding to definition (2))
The date of soil freezing estimated from SMOS (DoFX, X = V, NPR) was compared to the soil freezing date assessed from both liquid soil moisture VWCG and soil temperature TG measured in situ at 5 cm depth (DoFin situ, in situ = TG, VWCG)
Summary
The European Space Agency's SMOS (Soil Moisture and Ocean Salinity) satellite mission (Kerr et al, 2010) provides regular low-frequency (L-band, 1–2 GHz) passive microwave observations (brightness temperatures) from space. Recent theoretical studies (Schwank et al, 2014, 2015), corroborated by experimental investigations (Lemmetyinen et al, under review), have shown that passive L-band observations TBp are sensitive to dry snow cover, in particular at horizontal polarization. Before this recognition L-band brightness temperatures TBp were assumed to be insensitive with respect to dry snow due to its acknowledged low extinction at L-band. The investigated study period is from February 2010 until December 2014, covering four autumn freezing seasons across the northern hemisphere
Published Version (Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have