Ivan S. Ashcraft

Comparison of methods for melt detection over Greenland using active and passive microwave measurements

Microwave measurements have been used in various studies to detect melt based on their sensitivity to liquid water present in snow. To contrast different melt detection methods used with different sensors, six different melt detection method/sensor combinations are compared using data for the summer of 2000. The sensors include the Special Spectral Microwave Imager (SSM/I), SeaWinds on QuikSCAT (QSCAT), and the European Remote Sensing (ERS) Advanced Microwave Instrument (AMI) in scatterometer mode. Existing melt detection methods are compared with melt detection based on a simple physical model. The model relates the moisture content and depth of a surface melt layer of wet snow to a single channel melt detection threshold. The model can be applied to both active and passive sensors and improves the consistency between brightness temperature (Tb ) and normalized radar backscatter (σ°) based detection of melt. Model‐based melt estimates from different sensors are highly correlated and do not exhibit the unnatural phenomenon observed with previous methods. Relative merits and limitations of the various methods are discussed.

Observation and characterization of radar backscatter over Greenland

Characterization of the microwave signature of the Greenland snow surface enables delineation of the different snow facies and is a tool for tracking the effects of climate change. A new empirical observation model is introduced that uses a limited number of parameters to characterize the snow surface based on the dependence of radar backscatter on incidence angle, azimuth angle, spatial gradient, and temporal rate of change. The individual model parameters are discussed in depth with examples using data from the NASA Scatterometer (NSCAT) and from the C-band European Remote Sensing (ERS) satellite Advanced Microwave Instrument in scatterometer mode. The contribution of each model term to the overall accuracy of the model is evaluated. The relative contributions of the modeled dependencies vary by region. Two studies illustrating applications of the model are included. First, interannual changes over the Greenland ice sheet are investigated using nine years of ERS data. Surface changes are observed as anomalies in the /spl sigma//spl deg/ model parameters. Second, intraannual variations of the surface are investigated. These changes are observed in the average backscatter normalized to a given observation geometry. The results indicate that the model can be used to obtain a more complete understanding of multiyear change and to obtain low-variance high temporal resolution observations of intraannual changes. The model may be applied for increased accuracy in scatterometer, synthetic aperture radar (SAR), and wide-angle SAR studies.

Differentiation between melt and freeze stages of the melt cycle using SSM/I channel ratios

Microwave remote sensing detection of snow melt and ablation generally focuses on the detection of liquid moisture in the snow-pack. For ablation estimation, it is important to determine if wet snow is in the process of melting or freezing. The different stages of the melt cycle are observed in the diurnal variation of T/sub b/ measurements from the Special Sensor Microwave Imager (SSM/I) over Greenland. SSM/I channel ratios exhibit patterns indicating that they are sensitive to melt and freeze stages of the daily melt cycle. The horizontal to vertical polarization ratio is sensitive to surface wetness associated with melting. The 19-37-GHz frequency ratio is sensitive to a frozen surface layer over wet snow which is associated with the freeze stage of the melt cycle. These observations are supported by conceptual models presented here and in in situ measurements from other investigators.

SeaWinds views Greenland

Data from the Ku-band SeaWinds scatterometer is used to investigate the extent of the ice facies in Greenland. The duration of the summer melt over the Greenland ice sheet is calculated using the temporal signature of the SeaWinds data. Daily ascending and descending enhanced resolution radar backscatter images of Greenland are produced to investigate diurnal variations in backscatter measurements. These are compared with diurnal variations in the brightness temperature measured by SSM/I. Multi-annual changes in the Greenland ice sheet are studied by comparing 1999 SeaWinds data to 1996 NSCAT data.

Radar scatterometer observations of sastrugi on the great ice sheets

The SeaWinds instrument on the QuikSCAT satellite was designed to measure near surface winds over the ocean; however, this remarkable remote sensing instrument has proven very useful in polar ice studies. Unlike previous radar scatterometers which were limited to 2 or 3 azimuth angles, the Ku-band SeaWinds instrument uses a circular scanning pencil beam, allowing it to make radar backscatter measurements from all azimuth angles. This geometry makes it an ideal candidate for studies of azimuth modulation of the normalized radar cross section of natural surfaces. Previous studies have observed a second order azimuth modulation of radar backscatter on the Antartic ice sheet, which has been related to wind-generated sastrugi (snow dunes) on the surface. In this paper we use SeaWinds data to make more detailed studies of the azimuth modulation in both Antarctica and Greenland where little has been done. Using the higher azimuth resolution possible with SeaWinds, we find that the azimuth variation of the backscatter is better described using a fourth order model in areas with the highest modulation. The orientation of these fourth order terms appears to be highly correlated to the katabatic wind direction. Azimuth modulation is as observed over Greenland, but it is much smaller than over Antarctica. Comparing SeaWinds and ERS-1/2 satterometer mode data we examine the frequency dependence, finding the modulation larger at C-band than Ku-band. The largest azimuth modulation in Greenland is observed in the transition region between dry snow and percolation zones.

Azimuth variation in microwave backscatter over the Greenland Ice Sheet

Scatterometer backscatter measurements are becoming an important tool for monitoring the dynamic behavior of the Greenland Ice S heet. However, most Greenland studies assume constant backscatter for varying azimuth angles. Detailed analysis of scatterometer data sets show non-negligible (1-2 dB) azimuth modulation. The magnitude and orientation of the azimuth modulation observed by SeaWinds, NSCAT, and ERS are documented herein. The dominate factor is second order azimuth modulation, which is largest for C-band (ERS). For SeaWinds, the h-pol azimuth modulation is larger than for v-pol.

σ retrieval from SeaWinds on QuikScat

NASA has developed a new scatterometer, SeaWinds, which is scheduled to be launched on two missions, one in 1999 and the other in 2000. SeaWinds will measure the speed and direction of near-surface winds over the ocean. A fundamental part of the processing of SeaWinds data is the computation of (sigma) degree which includes solving the radar equation for every pulse sent and received by the scatterometer. However, for SeaWinds, this calculation is too computationally intensive to perform in real tie. Instead, a method of tabularizing most of the calculation has been developed. This method includes corrections for attitude and orbit perturbations and accounts for the elevation of the local topography. It also includes a table which will be used to assist ground data processing in determining the locations of the measurements. Tests comparing the tabulated result with the actual numerical calculations have shown that using this table is accurate to within +/- 0.1 dB and gives locations that are accurate to within 160 meters. We provide a description of this algorithm. The innovations developed as part of this algorithm may be of interest in processing data from other remote sensing systems.

Relating micro ave backscatter azimuth modulation to surface properties of the greenland ice sheet

Azimuth modulation of the normalized radar cross- section in satellite data sets over Greenland is investigated. Data sets from the NASA Scatterometer (NSCAT) and from the European Remote Sensing Advanced Microwave Instrument (ERS) are employed. Azimuth dependence is clearly observed. The largest azimuth dependence occurs in the C-band ERS data with peak-to-peak azimuth modulations up to 3.0 dB. The Ku- band NSCAT data exhibits slightly smaller modulations of upto 2.0 dB. Azimuth modulation is largest in the lower dry snow zone for ERS and in the dry to percolation transition zone for NSCAT. The incidence angle dependence of the azimuth modulation is parameterized over the ice sheet. In general, the azimuth modulation is found to either decrease with increasing incidence angles, or be relatively independent of incidence angle. Regions of large incidence angle dependence for the azimuth modulation include the western dry snow zone for ERS and the northeast dry snow to percolation transition zone for NSCAT. The second order azimuth modulation orientation is highly correlated with wind direction. A new simple surface model is introduced to relate azimuth modulation to surface properties. Using this model, the size and orientation of surface sastrugi are estimated. I. INTRODUCTION The Greenland ice sheet is a critical area of study in esti- mating effects of global climate change. With only a limited number of in situ measurements due to the considerable effort associated with on site studies, remote sensing is an essential tool for studying the dynamics of this region. While satellite based normalized radar cross-section (σ o ) data have been used in a variety of successful studies over Greenland, past studies have generally ignored microwave measurement dependency on azimuth angle. This work is an extension of previous work found in (1). Data from the C-band European Remote Sensing Advanced Microwave Instrument in wind scatterometer mode (hereafter ERS) and the Ku-band NASA Scatterometer (NSCAT) are employed. Both sensors cover a large range of incidence angles. For NSCAT, only the vertical polarization (v-pol) measurements have sufficient azimuth coverage to be used in the study. The ERS measurements are also v-pol. The data shown herein is from the time interval Julian Day (JD) 330 to 360, 1996. This is during the winter when the Greenland surface is relatively constant. Although scarcely studied in Greenland, azimuth modula- tion is common in microwave remote sensing studies over other areas of the Earth. The primary application of azimuth modulation is in using scatterometers to measure vector wind speeds over the ocean surface. The azimuth modulation over Greenland is attributed to a complex snow surface structure created by wind driven snow deposition, aeolian transport, and the formation of wind slabs and hoar layers. The scale of the surface roughness varies from dunes on the km scale to meter scale erosional features known as sastrugi. First, the location and properties of the azimuth modulation are discussed. In particular, the incidence angle dependence of the azimuth modulation is addressed and the locality of the maximum modulation with respect to the different ice facies is addressed. Then the orientation of the azimuth modulation is presented to compare with Greenland wind models. Finally, a simple surface model is presented which relates the azimuth modulation to the geophysical properties of the snow surface.

Microwave backscatter over Greenland: changing with time

Changes in the Greenland ice sheet are considered important indicators of global climate change. These changes can be monitored using space-borne scatterometers which provide frequent coverage of the entire ice sheet. This paper provides a general overview of backscatter measurements over Greenland and the distinguishing attributes of the data sets over the different snow facies including temporal signatures. Seasat-A scatterometer (1978), NSCAT (1996-1997), SeaWinds (1999-present), and ERS AMI (1992-2000) scatterometer data are analyzed to evaluate the long term changes in the ice sheet. An increase in backscatter is observed in the dry snow zone near the dry snow zone/percolation zone boundary. A simple algorithm is applied to determine the length and extent of the melt for the summer of 1999 as observed by SeaWinds and ERS. A comparison between the two sensors shows similar results with the apparent differences attributed to the higher temporal resolution of SeaWinds and the difference in frequencies between the two instruments.

SeaWinds applications for land and ice studies

While originally designed for making wind vector measurements over the ocean, radar scatterometers are also effective for large-scale monitoring of the Earths land and ice surfaces. The recently launched SeaWinds scatterometer is the next generation of Ku-band scatterometers and offers many advantages over previous scatterometers including a very wide swath and constant incidence angle measurements. The wide swath of SeaWinds enables much more frequent coverage of the Earths surface than has previously been possible and enhances the potential of the data in land and ice applications. SeaWinds data has been particularly effective in polar ice applications and SeaWinds data is currently being used for operational sea-ice extent monitoring and large iceberg tracking. Additional applications include monitoring glacial and sea-ice melting and studying the firn structure in Greenland and Antarctica. New applications of SeaWinds are being developed, including high resolution wind retrieval. Several such applications are briefly considered.