4. MEASUREMENTS
4.1.
SATELLITE OBSERVING SYSTEMS
4.1.1.
Passive Microwave Sensors
4.1.2.
Active Microwave Sensors
4.1.3. Optical Sensors
4.2.
AIRCRAFT OBSERVING SYSTEMS
4.2.1.
NASA-JPL Airborne Synthetic Aperture Radar (AIRSAR)
4.2.2.
Polarimetric Scatterometer (POLSCAT)
4.2.3.
Airborne Passive Microwave Radiometry (PSR-A, AESMIR)
4.2.4.
National Weather Service GAMMA Snow System (GAMMA)
4.2.5.
Mission Plan for Aircraft Observing Systems
4.3.
GROUND-BASED MICROWAVE REMOTE SENSING
4.3.1.
University of Michigan Systems
4.3.2.
Japanese Passive Microwave AMSR Simulator
4.3.3.
Ground-Based FMCW Radar Remote Sensing
4.4. GROUND
DATA COLLECTION
4.4.1.
ISA Data Collection Strategy
4.4.2.
ISA Data Collection Plan
4.4.3.
Soil and Snow Measurements at the North Park MSA
4.4.4.
Ground Measurements at the Local-scale Observation Site
4.4.5.
Snow Measurements
4.4.6.
Soil Measurements
4.4.7.
Micrometeorological Measurements
4.4.8.
Regional Snow Measurement Networks
4. MEASUREMENTS
4.1. SATELLITE OBSERVING SYSTEMS
Data from several spaceborne microwave and optical sensors
will be available to support the objectives of the field experiment. Many
of these data are available routinely from various data centers, at little
or no cost to the experiment. Data from some sensors are available only
through solicited participation with the managing agency. In these cases,
data requests have been made and approved. The following satellite remote
sensing data sets will be collected for the CLPX study areas during the
2002-2003 period. Where possible, data sets will be collected for the full
Large-Regional Study Area. Some high-resolution image data are provided
in smaller granules. These data sets will only be collected for the Small-Regional
Study Area or for the 25-km x 25-km Study Areas. All satellite data sets
collected will be made centrally available as part of the CLPX data set.
4.1.1. Passive Microwave Sensors
Data from two spaceborne passive microwave sensors will be available to support CLPX objectives. These data will be routinely available through existing data centers, including the National Snow and Ice Data Center (NSIDC).
AMSR-E. The Advanced Microwave Scanning Radiometer-EOS (AMSR-E) is a 12 channel, 6 frequency passive microwave radiometer providing horizontal and vertically polarized measurements at 7-, 11-, 19-, 24-, 37-, and 89-GHz. It will be flown on the AQUA platform, which is scheduled for launch by March, 2002. AMSR-E data will become available 90-days following launch. Gridded snow data products will be produced at 25-km resolution. Validation of the at-launch AMSR snow retrieval algorithms [Chang et al., 1999] and data products is a significant objective of the CLPX.
SSM/I. The Special Sensor Microwave Imager (SSM/I)
is a passive microwave radiometer flown aboard Defense Meteorological Satellite
Program (DMSP) satellites. The SSM/I is a seven-channel, four-frequency,
linearly-polarized, passive microwave radiometric system which measures
atmospheric, ocean and terrain microwave brightness temperatures at 19.35-,
22.235-, 37.0-, and 85.5-GHz.
4.1.2. Active Microwave Sensors
Radar data will be available from four spaceborne sensors to support experiment objectives.
ASAR. The Advanced Synthetic Aperture Radar (ASAR) is a C-band, HH or VV radar with 30-m resolution in imaging mode. It will be flown aboard the ENVISAT-1 platform, which is scheduled for launch in January, 2002. A request for ASAR data for the CLPX study area has been approved by the European Space Agency.
RADARSAT-2. This Canadian Space Agency satellite is scheduled for launch in 2002, and is planned to provide spatial resolution from 100 m to 3 m in swaths ranging from 500 km to 20 km. It will be a multi-polarization C-band SAR. RADARSAT-2 data sets for this experiment will be collected by the NASA Jet Propulsion Laboratory.
QuikSCAT. The SeaWinds instrument is a Ku-band scatterometer flown aboard the the QuickSCAT satellite, launched in June 1999. SeaWinds uses a rotating dish antenna with two spot beams that sweep in a circular pattern, producing imagery with a resolution of roughly 25-km. QuickSCAT data will be provided by the NASA Jet Propulsion Laboratory.
PALSAR. The Phased
Array type L-band Synthetic Aperture Radar (PALSAR) will be flown aboard
the NASDA Advanced Land Observation Satellite (ALOS). ALOS is scheduled
for launch in 2002. It's normal high-resolution operating mode will provide
10-m (2-look) resolution data in HH or VV / HH+HV or VV+VH modes. ALOS
data have been requested and approved for the CLPX study area for three
years (2003-2005) to support experiment objectives.
Data from several optical sensors will be collected to support the objectives of the experiment, including site characterization, snow cover mapping, and other science interests.
ASTER. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), flown aboard the TERRA platform, provides high spatial resolution (15 - 90 m) imagery in 14 visible, near-infrared, and thermal infrared channels. With a swath width of 60 km, ASTER data will be useful for evaluating the retrieval of the areal extent of snow cover from microwave sensors, particularly within Intensive Study Areas of the field experiment. An ASTER snow mapping algorithm has been developed and compared to high-resolution (2-3 m) aerial photography. The field experiment will in turn be useful for further evaluating the accuracy of the ASTER snow mapping algorithm under a variety of viewing and illumination geometries, land cover, atmospheric and terrain conditions.
HYPERION. The Hyperion, flown aboard EO-1, is a high resolution hyperspectral imager capable of resolving 220 spectral bands (from 0.4 to 2.5 µm) with a 30 meter resolution. The instrument can image a 7.5 km by 100 km land area per image and provide detailed spectral mapping across all 220 channels with high radiometric accuracy. At least three images (one for each MSA) will be acquired during IOP-1 and IOP-2. The hyperspectral data will be useful for examining the optical albedo of the snow pack, snow grain size properties, and snow/vegetation interactions.
MISR. The Multi-Angle Imaging Spectroradiometer (MISR) provides simultaneous imaging of the Earth's surface at nine different angles. One camera points toward nadir, and the others provide forward and aftward view angles, at the Earth's surface, of 26.1°, 45.6°, 60.0°, and 70.5°. As the instrument flies overhead, each region of the Earth's surface is successively imaged by all nine cameras in each of four wavelengths (blue, green, red, and near-infrared). The MISR instrument data can be used for investigating bidirectional reflectance distribution functions (BRDFs) of snow during the field experiment, which may in turn provide insight regarding surface roughness characteristics that affect microwave signals.
MODIS. The Moderate Resolution Imaging Spectroradiometer (MODIS), flown aboard the TERRA platform, provides moderate resolution (250-1000 m) imagery in 36 spectral bands. With a 2330 km swath width, MODIS data will be useful for evaluating the retrieval of the areal extent of snow cover from microwave sensors at substantially larger scales than ASTER.
MERIS. The Medium Resolution Imaging Spectrometer (MERIS) will be a MODIS-like instrument operating with 15 spectral bands in the visible and near-IR, with 300 m ground resolution. It will be flown aboard the ENVISAT-1 platform, which is scheduled for launch in October, 2001. A request for MERIS data has been approved by the European Space Agency.
AVNIR-2. The Advanced Visible and Near Infrared Radiometer type 2 (AVNIR-2) will be flown aboard the NASDA Advanced Land Observation Satellite (ALOS). ALOS is scheduled for launch in 2002. AVNIR-2 will have 4 spectral bands (0.42-0.50, 0.52-0.60, 0.61-0.69, and 0.76-0.89 microns), with a nadir ground resolution of 10-m. ALOS data have been requested and approved for the CLPX study area for three years (2003-2005) to support experiment objectives.
GOES. The Geostationary Operational Environmental Satellites (GOES) Imager is a five-channel optical instrument designed to sense radiant and solar-reflected energy at 0.65-, 3.9-, 6.7-, 11-, and 12-microns. The instrument can produce full-Earth disc images, sector images that contain the edges of the Earth, and various sizes of area scans completely enclosed within the Earth scene.
AVHRR. The Advanced
Very High Resolution Radiometer (AVHRR) is a broad-band, four or five channel
(depending on the model) scanner, sensing in the visible, near-infrared,
and thermal infrared portions of the electromagnetic spectrum. This sensor
is carried on NOAA's Polar Orbiting Environmental Satellites (POES).
4.2. AIRCRAFT OBSERVING SYSTEMS
Aircraft remote sensing will
be focused on the three 25-km x 25-km MSAs. Remote sensing data sets will
be collected from five instruments during the experiment. The focus of
airborne data collection is on coincident active and passive microwave
measurements. These measurements will provide complete coverage of the
three study areas. Active microwave data will be collected using the NASA-JPL
Airborne Synthetic Aperture Radar (AIRSAR) instrument aboard the NASA DC-8
aircraft, and from the NASA-JPL Polarimetric Scatterometer (POLSCAT) instrument,
also flown aboard the NASA DC-8. Passive microwave data will be collected
using the NOAA Polarimetric Scanning Radiometer (PSR-A) instrument aboard
the NASA DC-8 aircraft in 2002. In 2003, either the PSR-A or the NASA AESMIR
instrument, similar to the PSR-A, will be flown on the NASA P3-B aircraft.
Terrestrial gamma radiation measurements of snow water equivalent will
be collected using the GAMMA instrument aboard the NOAA AC690A aircraft.
4.2.1. NASA-JPL Airborne Synthetic Aperture Radar (AIRSAR)
The AIRSAR instrument, flown aboard the NASA DC-8 ( Figure 38 ), will provide polarimetric and cross-track interferometry radar data at P-, L, and C- bands. Complete information about the AIRSAR system is available at http://airsar.jpl.nasa.gov .
Instrument Characteristics. The AIRSAR system operates in the fully polarimetric mode (POLSAR) at P-, L-, and C-band simultaneously or in the cross-track interferometry mode (XTI, or TOPSAR) in both L- and C-band simultaneously. Thus, collection of both POLSAR and TOPSAR data requires two overpasses. In AIRSAR, transmit polarization diversity is achieved by alternately transmitting the signals using horizontal or vertical polarizations. Receive polarization diversity is accomplished by measuring six channels of raw data simultaneously, both H and V polarizations at all three frequencies. The video data are digitized using 8-bit ADCs, providing a dynamic range in excess of 40 dB and, together with navigation data, stored on tape using high-density digital recorders. The AIRSAR system also includes a real-time processor capable of processing any one of the 12 radar channels into a scrolling image. In addition to checking the health of the radar, the scrolling display is also used to ensure that the correct data have been imaged. A summary of AIRSAR system characteristics is given in Table 13 [Lou, Kim, and van Zyl, 2001].
AIRSAR Data Processing and Calibration. A variety of processors and processing techniques are used to process AIRSAR data to imagery. A real-time correlator is used to produce low resolution (approximately 25-m) two-look survey imagery. The same on-board equipment is used to generate a slightly higher resolution (15-m), 16-look image of a smaller area (12-km x 7-km) within 10 minutes of acquisition using the quick-look processor. These on-board processors are useful for assessing the general health of the radar and the success of data taking in real time.
Final processing of selected portions of the data to high-quality, fully calibrated image products happens in the weeks and months following a flight campaign. Final products are expected to be available approximately 8-months following data acquisition.
AIRSAR Flight Plan. The AIRSAR is scheduled to be deployed in IOP-1, IOP-2, and IOP-4. Preliminary flight line planning for the three MSAs has been completed for both POLSAR and TOPSAR coverage. The primary mapping lines for each mode are planned as three east-west lines with one perpendicular crossing line over the more rugged part of the ISA ( Figure 39 ). The crossing line will provide data to fill in holes due to radar shadow in the data and also to tie down the primary flight lines during the mosaicking process. The location of the crossing line can be moved. The nominal altitude for the AIRSAR instrument is 8000-m above the mean terrain height, thus for the North Park, Rabbit Ears, and Fraser ISAs, the nominal DC-8 altitude will be at 10,500-m (34,450'), 10,725-m (35,188'), and 11,100-m (36,419') respectively, well below its service ceiling of 12,500-m (41,000').
The POLSAR lines are planned at 40 MHz bandwidth for C-band and L-band and 20 MHz bandwidth for P-band. Due to FAA restrictions, P-band data cannot be transmitted at 40 MHz. This will result in a P-band data swath that is wider than those shown in the simulated images, and with greater overlap between the P-band data swaths. Also, due to the difference in bandwidth, the P-band data will not be co-registered with the C-band and L-band data in the final processed product. The overlap between the primary mapping lines is 3.5 km.
The TOPSAR lines are planned in XTI2P mode which yields DEM data at C-band and L-band with CVV and LVV data. As with the POLSAR data, the TOPSAR data are planned at 40 MHz for C-band and L-band and at 20 MHz for P-band. The P-band data will be fully polarimetric. The overlap between the primary mapping lines for TOPSAR is also 3.5 km.
AIRSAR Data Products.
For POLSAR data transmitted at 40 MHz bandwidth (L- and C-band), the azimuth
pixel spacing is 9.26-m and the range pixel spacing is 3.3-m (18 looks
taken during processing). At 20 MHz bandwidth (P-band), the azimuth pixel
spacing is 9.26-m and the range pixel spacing is 6.6-m. The POLSAR data
products are in slant range projection. For TOPSAR data, the azimuth and
range pixel spacing is 5-m. Data products in the TOPSAR mode are in the
ground range projection. Polarimetric data collected in TOPSAR mode are
co-registered to the DEM and are also in the ground range projection.
4.2.2. Polarimetric Scatterometer (POLSCAT)
The NASA POLSCAT instrument will be flown aboard the NASA DC-8 during IOP-1 and IOP-2 in conjunction with AIRSAR. POLSCAT is a Ku-band polarimetric scatterometer built and operated by NASA JPL. The instrument has a demonstrated calibration stability of better than 0.1 dB over several hours and a polarization isolation of better than 35 dB.
The POLSCAT operates at Ku-band (13.95 GHz), similar to the NASA SeaWinds and NSCAT scatterometer frequencies. It uses a conical corrugated antenna horn for transmit and receive. An Orth-Mode Transducer (OMT), connected to the antenna horn, separates vertical and horizontal polarization channels. The ferrite switch bank, consisting of several Ferrite switches and isolators, enables the selection of transmit polarization and provides the receiver protect during transmit. The radar transmitter is a Ku-band solid state power amplifier (SSPA) with 10 W output. The timing generator provides all the switch control signals and the SSPA-enable signal. The frequency synthesis assembly derives all the mixing frequencies from a 10-MHz local oscillator, which are provided to the SSPA and the receivers.
The transmitting polarization is alternated between V and H from pulse to pulse. The V- and H-polarized radar echoes are detected simultaneously by two receivers. Each receiver down-converts the frequency of the backscatter signals and provides in-phase (I) and quadrature-phase (Q) components to a personal computer (PC) for data acquisition. In the PC is a National Instrument data acquisition card (NI DAQ PCI-6110) with 4 12-bit analog to digital (A/D) converters, which sample the I/Q signals from both receivers at 4 MHz rate. The PC processes the A/D samples acquired over every set of 1600 transmit pulses with 800 of them for each transmit polarization. The samples are squared and averaged to produce the power of each polarization. The complex multiplication required to obtain the polarimetric correlation coefficients between the signals from different polarization channels is also performed by cross-multiplying the I and Q samples from two polarization channels. Note that pVVHV and pHHVH are computed using the samples from the same transmit pulses, while pVVHH , pVVVH , pHHHV and pHVVH are computed using samples from adjacent pulses. The power and correlation products are averaged over samples acquired within every 0.56 second and saved to the PC hard disk. The aircraft navigation data from C-130, including latitude, longitude, pitch, yaw and roll angles, are also recorded at the rate of 1 Hz by the PC.
The in-flight calibration is performed using a calibration loop, consisting of one waveguide attenuator and two directional waveguide couplers in the radar front-end, which leaks a small transmit signal into the receivers. For about every 10 seconds, the receiver range gate is moved to the transmit period to sample this leakage signal, which is a product of the SSPA transmit power, receiver gain, and the loss of calibration loop. The loss of calibration loops, about 80 dB, was measured in the laboratory. Tests conducted in the laboratory and aircraft flight data suggest that the calibration loop measurements are very stable with a drift of less than 0.1 dB over several hours.
The POLSCAT instrument will be deployed on the NASA DC-8 during IOP-1 and IOP-2. The POLSCAT antenna will be mounted on the DC-8 62-degree port, and will point approximately in the center of the JPL AIRSAR swath. Flown at the AIRSAR altitude of 8000-m above mean-terrain-height, the POLSCAT target will be approximately 1-km wide, for the length of each AIRSAR flight line (approximately 30-km). Consequently, the POLSCAT data will cover only a fraction of each MSA. To ensure that POLSCAT data are acquired over each 1-km x 1-km ISA, two additional flight lines will be flown over each MSA at the same altitude.
POLSCAT Data Processing
and Products. Processing of the POLSCAT data will occur immediately
following the completion of IOP-2, when the instrument is removed from
the aircraft. A report describing the POLSCAT/CLPX02 flight experimental
data will be released in three months after the completion of IOP-2. Radiometric-calibrated,
geo-located POLSCAT data on CD-ROM will also be delivered to the CLPX data
management facility within three months of the completion of the flight.
4.2.3. Airborne Passive Microwave Radiometry (PSR-A, AESMIR)
The NOAA PSR-A instrument, flown aboard either the NASA DC-8 or P3-B ( Figure 40 ), will provide polarimetric passive microwave data at 10.7-, 18.7-, 21.4-, 37.0-, and 89.0-GHz frequencies. Complete information about the PSR-A system is available at http://www1.etl.noaa.gov/radiom/psr.html. The PSR-A will be flown on the DC-8 during IOP-1 and IOP-2. The similar NASA AESMIR instrument is scheduled to fly on the NASA P3-B during IOP-3 and IOP-4. Its design is nearly identical to the PSR-A instrument, but includes 1.5 and 6.9 GHz frequencies (L- and C-bands), which would be useful to the objectives of the experiment. The AESMIR instrument is currently under development; if it is not ready to fly or fails to meet specifications, the PSR-A will be flown in 2003 instead.
The NOAA PSR-A instrument consists of a set of five polarimetric radiometers housed within a gimbal-mounted scanhead drum ( Figure 41, Figure 42 ). The scanhead drum is rotatable by the gimbal positioner so that the radiometers can view any angle within 70E elevation of nadir and any azimuthal angle, as well as external hot and ambient calibration targets. The configuration thus supports conical, cross-track, along-track, fixed-angle stare, and spotlight scan modes. The conical scan mode allows the entire modified Stokes' vector to be observed without polarization mixing.
The precise radiometric bands measured by the PSR-A are X (10.6-10.8 GHZ), Ku (18.6-18.8 GHz), K (18.6-21.7 GHZ), Ka (36-38 GHz), and W (86-92 GHz). Calibration of all the PSR-A radiometers is performed in-flight using standard (unpolarized) hot and cold blackbody targets. Each of the radiometers also incorporate sub-interval calibration hardware to supplement the standard hot- and ambient-view calibration. The absolute accuracy of the PSR-A is approximately 1-K.
The AESMIR instrument is an airborne AMSR simulator covering all AMSR frequencies with a single imaging package: 6.9-, 10.7-, 18.7-, 23.8-, 36.5-, and 89-GHz frequencies with 5-10E beamwidths. An L-band option is under development, with a likely completion date in 2003-2004 may be available during the experiment. All frequencies will have V and H polarizations. Additionally, the 6.9- and 10.7-GHz radiometers will be fully polarimetric (4-Stokes). The other radiometers (except 23.8-GHz) will also become fully polarimetric in the 2003-2004 time frame.
The AESMIR design is unique in that it will perform imaging at all AMSR frequency bands using only one sensor head/scanner package. Scanning modes include conical, cross-track, fixed-beam and arbitrary geometries. AESMIR is compatible with both the DC-8 and the P-3B aircraft to be used in the experiment.
The microwave radiometers
incorporate state of the art receivers, with particular attention given
to instrument calibration for the best possible accuracy and sensitivity.
Continuous in-flight calibration will be accomplished using hot and cold
external blackbody targets as well as internal hardware, and an accuracy
of 1K is expected.
PSR-A/AESMIR Flight Plan. The PSR-A has previously been integrated into the NASA DC-8 aircraft (nadir-7 port) as well as the NASA P-3B (aft end of the bomb bay). The PSR-A will be deployed on the DC-8 during IOP-1 and IOP-2. The AESMIR instrument will be deployed on the NASA P-3B aircraft during IOP-3 and IOP-4 (with PSR-A as a back-up). Final decisions regarding sensor selection and platform are subject to coordination with other field experiments and aircraft schedules (discussed in a later section), as well as the progress of the AESMIR instrument development.
The flight planning objectives for either instrument are the same. There is a trade-off between two goals: 1) maximize the image resolution (minimize footprint size), and 2) minimize total time required to image a single 25-km x 25-km study area. The first goal is to collect data at the maximum resolution possible to a) minimize mixed pixel effects, b) improve comparability with AIRSAR data, and c) ensure that entire footprints are completely sampled within the ground sampling framework for the 1-km2 ISAs. The second goal is to minimize the effects of changing conditions within a 25-km x 25-km study area during the image acquisition period.
Preliminary flight planning for the three 25-km x 25-km MSAs has been completed for the PSR-A instrument flown aboard the NASA DC-8 aircraft ( Table 14; Figure 43, Figure 44, and Figure 45 ). Following AIRSAR data collection, the aircraft will be flown at a nominal altitude of 1525-m (5000') above mean terrain height. Thus for North Park, Rabbit Ears, and Fraser MSAs the DC-8 altitude will be 4024-m (13,203'), 4250-m (13,944'), and 4591-m (15,063') respectively. These altitudes are 1298-m (4259'), 970-m (3183'), and 629-m (2064') above the maximum terrain height in each area, respectively.
Coverage of each MSA will consist of 12 flight lines spaced 2.5-km apart, providing 4.35-km wide swaths with 1.85-km overlap. The geometric spot size in this configuration will be 0.49-km for the 10.7-21.5 GHz frequencies, and 0.14-km for the 37 GHz and 89 GHz frequencies. Each MSA will be completely imaged in approximately 1-hour using the DC-8 platform, and in 1.7 hours on the P3-B platform.
The PSR-A/AESMIR flights will require fair weather conditions without significant turbulence, particularly for the Rabbit Ears and Fraser MSAs.
The PSR-A scanner will be
set to 55o, coincident with that of AMSR-E. Calibration maneuvers
(three 60o left-bank rolls) will be conducted during each flight
to provide cold-space views. The use of cold-space views is expected to
permit absolute calibration of the PSR-A channels of better than 1K.
PSR-A Data Processing
and Products. Quick-look images (printouts and graphical displays of
uncalibrated PSR-A data) will be available during and immediately after
flight for lines of particular interest. A PSR-A Data log describing the
PSR-A flight lines and imaging characteristics will be available within
~10 days to facilitate planning for scientific analysis. Level 2.1 imagery
(calibrated, non-registered) will be available for archival at the data
management facility for 90% of MSA flight lines within 3 months of completion
of the flight campaign (May 31, 2002). Level 2.3 imagery (calibrated, georegistered)
will be available for archival within 6 months of completion of the flight
campaign (August 31, 2002). A low-resolution digital elevation map will
be incorporated into the PSR data stream for Level 2.3 processing to improve
georegistration in rugged terrain. The NOAA PSR-A post flight processing
will be completed by October 1, 2002.
4.2.4. National Weather Service GAMMA Snow System (GAMMA)
The main purpose of airborne gamma data collection in this experiment is to provide an accurate estimate of the average snow water equivalent within each 25-km x 25-km MSA. This will serve as "ground truth" for snow water equivalent retrieval algorithms for spaceborne passive microwave sensors and for snow water equivalent estimated by meso-scale models. This is a reasonable approach because the error characteristics for airborne gamma snow water equivalent measurements are well known.
The GAMMA instrument (
Figure 46
) is flown aboard the NOAA AC690A Turbo Commander aircraft (
Figure 47
). It has been used operationally by the National Weather Service
since 1979 to measure the mean areal snow water equivalent and soil moisture
along flight lines [Peck and Carroll, 1980; Jones and Carroll,
1982; Carroll, 1987]. The instrument and aircraft are operated by
the National Operational Hydrologic Remote Sensing Center (NOHRSC). Attenuation
of natural terrestrial gamma radiation by water (any phase) in the snow
and soil is the basis of the snow water equivalent and soil moisture measurements.
Airborne Snow Water Equivalent Measurement. The ability to make reliable, airborne gamma radiation snow water equivalent measurements is based on the fact that natural terrestrial gamma radiation is emitted from the potassium, uranium, and thorium radioisotopes in the upper 20-cm of soil. The principal contributors to terrestrial gamma rays are the 40K, 238U and 232Th series. The radiation is sensed from a low-flying aircraft flying 150-m above the ground. The terrestrial gamma rays are attenuated by the atomic cross-section of the intervening mass -- water, vegetation, and air -- between the ground and the aircraft.
The technique first requires a measurement of the background terrestrial gamma radiation along a flight line. This is done with no snow cover present. An estimate of the mean areal soil moisture along the flight line at the time of the background measurement is necessary. This is used to calibrate the background radiation measurement to account for the attenuation effects of the existing soil moisture. Collecting the background radiation data during dry soil conditions facilitates estimation of the background soil moisture conditions. The absorption and re-radiation of gamma radiation by intervening vegetation mass is accounted for in the calibration. Background radiation and soil moisture values are collected only once under no-snowcover conditions and used to calibrate flight lines.
Subsequent measurements of
terrestrial gamma radiation over calibrated flight lines determine the
attenuation of the radiation signal due solely to the intervening water
mass in the snow and soil. Consequently, the difference between airborne
radiation measurements made over snow-free ground and snow-covered ground
can be used to calculate a mean areal snow water equivalent value. To do
this, the mean areal subnivean soil moisture along the flight line must
be measured or estimated. This enables partitioning of the measured attenuation
into the part due to snow water equivalent and the part due to soil moisture.
The technique is not sensitive to the phase of the water mass in either
the snow or soil. Attenuation caused by the air mass (for both background
and over-snow flights) is accounted using independent air mass measurements
taken from the aircraft. Airborne snow water equivalent measurements are
made using the following relationship:
where:
C and C0 = Uncollided
terrestrial gamma count rates over snow and bare ground,
M and M0 = Percent
soil moisture for snow-covered and bare ground,
A = Radiation attenuation
coefficient in water (cm2/g).
The inverse radiation attenuation coefficients used in Equation (1) for the potassium, thorium, and total count windows are 14.38, 18.85, and 17.73, respectively. An independent snow water equivalent value is calculated for each of the three radioisotope peaks. A weighted snow water equivalent is calculated by multiplying each of the three independent snow water equivalent values by a weighting coefficient (which sums to unity) and summing the results. The potassium, thorium, and total count weighting factors are 0.346, 0.518, and 0.136, respectively.
Estimation of the background and subnivean soil moisture values (M0 and M) is a source of uncertainty in the gamma snow water equivalent measurement. The coefficient of variation of soil moisture along a flight line is often large. Consequently, a large number of soil moisture samples are required to accurately estimate the mean areal soil moisture along a flight line. In most cases such intensive sampling is prohibitively expensive. Instead, the values for M0 and M are usually estimated approximately using other methods.
Percent soil moisture by weight is calculated as the weight of water divided by the weight of dry soil multiplied by 100. By this formulation, it is possible to have a percent soil moisture value greater than100%. Field holding capacity is largely a function of land use and soil type. For a typical loam soil, field holding capacity is about 35% soil moisture. Under frozen soil conditions, it is possible to accumulate interstitial ice which can raise the percent soil moisture values for the upper 20-cm to typical values of 50-70%. Soil moisture can be generally characterized by the relationship given in Table 15. The attenuation rates of gamma radiation by water are very well known. Attenuation increases nearly linearly with increasing water mass. Consequently, calculation of the effects of errors in soil moisture estimates and of appropriate snow water equivalent correction terms is straightforward (Table 16). A 5% error in either the background or subnivean soil moisture terms results in a 0.5-cm to 0.9-cm absolute error in measured snow water equivalent. Therefore, for deep snow areas the relative error associated with an approximate soil moisture estimate is minor. Background and subnivean soil moisture can be estimated using fewer samples without significant relative error. For shallow snow areas, however, such as the North Park MSA, accurate estimates of background and subnivean soil moisture are necessary.
The other significant source of uncertainty in the airborne snow water equivalent measurement is the count rate statistics for the uncollided terrestrial gamma count rates over snow and bare ground (C and C0). In general, larger total count rates for each term result in reduced uncertainty. Large attenuation (e.g. by large snow water equivalents or dense biomass [Glynn et al., 1988; Carroll and Carroll, 1989]) reduces total count rates and increases uncertainty. Since gamma counts are integrated over the length of the flight line, longer flight lines result in larger total counts and reduce uncertainty.
Thus, the major sources of error in airborne snow water equivalent measurements are:
Gamma Data Collection. The main objective of GAMMA data collection in this experiment is to provide an accurate estimate of the average snow water equivalent within each study area, to be used as "ground truth" for satellite microwave retrieval algorithms and meso-scale models. This requires a dense sample of measurements from as many flight lines as possible. This goal is constrained by the need for sufficiently long flight lines to increase integration time and reduce count statistics errors (8-10 km), and by flight safety considerations for low-level flight in complex terrain. These constraints considerably reduce the number of possible flight line locations within Rabbit Ears and Fraser study areas, but are not significant limitations in the North Park study area.
Flight plans for each of the MSAs have been developed and were flight-tested in April, 2001. The purpose of the flight tests was to: 1) confirm that flight lines planned using maps and digital elevation data sets could be flown safely and within aircraft performance limits, 2) collect GPS way-points for each flight line to permit more efficient navigation during the experiment, 3) determine the time requirements to complete the flight line network in each area, and 4) determine whether the magnitude of snow water equivalent along any of the lines was sufficient to cause full attenuation of terrestrial gamma radiation (which would make measurement of snow water equivalent impossible by this method). The snow water equivalent throughout this region of Colorado was very close to the long-term normal for April 1, so the flight tests provide reasonable assurance that measurements will be successful during the experiment. However, they do not rule out the possibility of full-attenuation problems along some lines if unusually heavy snow conditions are encountered during the experiment.
The North Park GAMMA flight plan consists of 28 parallel series of three 9-km flight lines, with each series spaced 1-km apart ( Figure 49 ). This sampling scheme provides 84 individual SWE samples. With a flight line field-of-view of 300-m and a line spacing of 1000-m, this sampling scheme will measure snow water equivalent over approximately 30% of the land area in the MSA. In all three MSAs, flight lines extend a short distance beyond the MSA boundaries to reduce edge effects in subsequent analysis of snow water equivalent samples. The shallow snow packs in North Park mean that gamma count statistics errors should be low for these 9-km lines, and there is essentially no risk of full-attenuation problems in this area. However, the shallow snow packs also mean that in order to minimize the relative uncertainty of the snow water equivalent measurements, background and subnivean soil moisture must be estimated especially carefully in this study area. Measurement of soil moisture for this and other purposes is discussed in a later section.
The Rabbit Ears GAMMA flight plan consists of 22 individual flight lines distributed throughout the study area ( Figure 50 ). These lines are planned to be as long as possible to reduce gamma count statistics uncertainty in areas of deeper snow cover, but are generally limited to about 10-12 km due to terrain constraints ( Table 17 ). The lines generally follow valleys (downhill) or elevation contours to provide safe exits for the aircraft in case of emergency. Results from the early April flight test at the Rabbit Ears study area indicated low gamma count statistics on seven of the flight lines (EX302, EX307, EX314, EX315, EX316, EX321, and EX322), all located in the upper plateau portion of the MSA between Rabbit Ears and Buffalo Passes. One line (EX316, the shortest line) was fully attenuated. Deep snow packs in this area will increase the uncertainty of airborne gamma snow water equivalent measurements. This will be a larger factor for the late-March campaigns than for the mid-February campaigns.
The Fraser GAMMA flight plan consists of 23 individual lines distributed throughout the study area in a similar manner as for the Rabbit Ears MSA ( Figure 51 ). These lines are generally a little shorter than in Rabbit Ears due to the more complex terrain in this study area ( Table 18 ). All of the Fraser lines had strong gamma count statistics in the early April flight test. No problems with full-attenuation were indicated.
Gamma Flight Operations and Data Processing. Weather will be a much greater factor for airborne gamma data collection than for airborne microwave data collection. Fair weather (no rainfall or snowfall) is required for airborne gamma data collection because attenuation of gamma radiation by precipitation is not accounted for in the measurement. Also, because of the risks of low-level flight in complex terrain, winds and turbulence aloft must be relatively low. In fair weather, measurements of any of the three MSAs can be completed in a single day. The primary objective will be to collect airborne gamma data coincidently with airborne microwave data collection efforts in each study area. If this is not possible due to weather, airborne gamma data will be collected as close in time to the other measurements as possible. The aircraft will remain deployed in the area for the duration of each field campaign.
All gamma data will be processed
by the NOHRSC, using operational algorithms, immediately following data
collection (same day). The processed data will be reported in Standard
Hydrologic Exchange Format (SHEF), and made available on the NOHRSC web
site. The operational data report for each flight line will include the
flight line ID, date, snow water equivalent, the background and subnivean
soil moisture values (M0 and M) used in the SWE calculation,
and pilot remarks. The data will be reprocessed following the field campaigns,
when improved estimates of subnivean soil moisture are available. The raw
gamma spectral data for all background and over-snow flights will be available
from the NOHRSC.
4.2.5. Mission Plan for Aircraft Observing Systems
Aircraft have been scheduled for each year. Scheduling and coordination of the NOAA AC690 aircraft is straightforward. It is available for use in all three years of the experiment. The two NASA aircraft have much more complex schedules, and a requirement of this experiment plan is to provide flexibility to adjust to changing aircraft schedules.
NOAA AC690A (Turbo Commander). This aircraft will be used in the Fall of 2001 to collect background gamma radiation measurements over each of the three MSAs. It will be used during both IOPs in 2002 and 2003 to measure snow water equivalent over each of the three MSAs. The aircraft will be ferried to Grand Junction, Colorado prior to the first IOP, where it will remain based until the completion of the second IOP.
The target data acquisition schedule for each MSA will be the same day as ground data acquisition for that MSA (Day 1, Day 3, and Day 5; see Section 4.5 for details). Flight tests have confirmed that all lines in any of the MSAs can be completed in a single day in fair weather. This schedule will result in three data takes for each MSA during each IOP. Given the realities of winter weather, AC690A flight operations will take advantage of fair weather at any time during the IOP or shortly thereafter to complete the data collection of all three MSAs, even if that means preceding ground data collection.
NASA DC-8. This aircraft will be flown in 2002 (IOP-1 and IOP-2) and again in 2003 (IOP-4). The SAGE experiment has priority status for the DC-8 in 2003, and will be using the aircraft during IOP-3.
For the first IOP in 2002 the aircraft will be ferried from Dryden Flight Research Center in California. It will be based at Colorado Springs, CO until the completion of IOP-1 data collection. For the second IOP in 2002, the DC-8 will be in use for the Great Lakes Ice Experiment. Currently, the DC-8 is scheduled to return to Dryden from Madison, WI on the first day of IOP-2. Since the complete CLPX instrument suite will still be aboard the DC-8 at that time, there will be an opportunity to collect limited data over the Colorado study area during that transit flight. During IOP-4, the aircraft will again be based at Colorado Springs for the duration of IOP-4 data collection.
In the event that a slip in the CLPX schedule is necessary, the CLPX has the highest priority schedule for the DC-8 in 2004.
The target data acquisition schedule for each MSA will be the same day as ground acquisition for that MSA (Day 1, Day 3, and Day 5; see Section 4.5 for details). The AIRSAR data acquisition characteristics (radar plus a high platform altitude) make weather less of a concern than for other airborne data collection. Nonetheless, precipitation conditions and excessive turbulence will both be avoided. As with the other aircraft, data collection will work around weather to the extent necessary. The time of data acquisition is also a factor. For the winter IOP (IOP-1), AIRSAR data collection will be conducted in the morning to maximize the potential for dry conditions. For the spring IOP (IOP-2 and IOP-4), AIRSAR data collection will be conducted in the afternoon to maximize the potential for wet conditions.
For each of three missions during IOP-1 (i.e. departure from base for data collection and return to base), the aircraft will fly each MSA three times: once for TOPSAR mode, again for POLSAR mode with two additional lines for POLSCAT, and a third time at lower altitude for the PSR-A. The total time for data collection for each MSA will be about 2.5 hours, for a total of 7.5 hours required per data collection. The round-trip ferry time between Colorado Springs and the study area is approximately 1-hour, for a total of 8.5 flight hours per mission. This schedule will result in three data takes for each MSA during each IOP (nominally two-days apart, depending on weather), with TOPSAR, POLSAR, POLSCAT, and PSR-A data acquisitions. Two CLPX investigators will fly on board the DC-8 during each mission to assist with quality assurance for the data collection.
IOP-2 data collection will be much more limited, since it will be conducted during a cross-country transit flight from Madison, WI to Dryden. Since there is a limit to the maximum number of flight hours that can be flown on a non-stop trip, the specific data collection during IOP-4 will depend on circumstances surrounding the transit flight, such as time of departure from Madison, weather conditions en route, etc. The CLPX budget includes ten flight hours for data collection during IOP-4. The priority for IOP-4 data collection will be AIRSAR and POLSCAT data collection (for observing wet snow from close to the transit altitude) over at least one MSA.
During both IOP-2 and IOP-2, the DC-8 will have to coordinate with the AC690 aircraft, which will be passing through the lower DC-8 altitude (13,000' - 15,000' ASL for PSR data collection) in route to its flight lines.
The IOP-4 DC-8 campaign will be similar to IOP-1, with three missions based out of Colorado Springs. POLSCAT and PSR are not scheduled to be deployed during IOP-4, so only AIRSAR flight lines will be flown in each mission. For each of three missions during IOP-4, the aircraft will fly each MSA twice, once for TOPSAR mode, and again for POLSAR mode. The total time for data collection for each MSA will be about 70 minutes, for a total of 3.5hr required for data collection in each mission. With the 1-hr round-trip ferry time, each mission will last 4.5 hours. This schedule will result in three data takes for each mission (nominally two days apart, depending on weather), with TOPSAR and POLSAR data acquisitions. Again, two CLPX investigators will fly on board the DC-8 during each mission to assist with quality assurance for the data collection.
NASA P3-B. This aircraft will be flown in 2003 (IOP-3 and IOP-4) for microwave data collection using either the PSR-A or AESMIR instruments. In early 2003 there are two other AQUA validation experiments planned that will be using the an airborne passive microwave instrument aboard the P-3B (precipitation in Wakasa Bay, Japan, and sea ice in the Arctic). The CLPX plan for 2003 is subject to close coordination with these other experiments. The precipitation experiment will conclude shortly before the first 2003 IOP. The aircraft will be ferried from Japan back to the United States, most likely via Dryden. Following a crew change, the aircraft will be ferried to Colorado. It will remain based at Colorado Springs for the duration of the first IOP. Following the IOP-3, it will be ferried to the Arctic for the sea ice experiment. It will return to Colorado following that experiment, to collect data for IOP-4. From there it will most likely return to its base at Wallops.
The target data acquisition schedule for each MSA will be the same as for AIRSAR, with morning flights for the winter IOP-3 and afternoon flights for the spring IOP-4. The low altitude planned for the aircraft will again make weather a significant factor, so data collection will work around weather to the extent necessary.
For each mission, the aircraft will fly each MSA a single time at a nominal altitude of 5000' above mean terrain height (see Sec. 4.2.2 for details). The total time for data collection for each MSA will be about 1.75 hours, for a total of 5.25 hours required for data collection. The round-trip ferry time from Colorado Springs is 1.25 hours, resulting in a total of 6.5 flight hours per mission. This schedule will result in three data takes for each MSA during each IOP. A CLPX investigator will fly on board the P-3B during each mission to assist with quality assurance for the data collection.
During IOP-4, the P3-B will
be flying simultaneously with the DC-8 and the AC690, however the DC-8
will remain at higher altitude for all components of its data collection.
The P3-B will have to coordinate with the NOAA AC690, which will pass through
the nominal P3-B altitude en route to its data collection.
4.3. GROUND-BASED MICROWAVE REMOTE SENSING
The objective of ground-based microwave remote sensing in this experiment is to collect time series of active and passive microwave spectral signatures over snow, soil, and forest, coincident with intensive physical characterization of these features. Whereas the airborne remote sensing and MSA data collection focus more on the spatial domain, the ground-based remote sensing will focus more on the temporal domain. The resulting data sets will be used to evaluate and improve microwave radiative transfer models for snow, soil, and trees. This is necessary to develop forward estimation methods for predicting microwave emission and backscattering using land surface models, a first-step towards direct assimilation of microwave remote sensing data in these models. The data set will also be important for investigating how microwave response in the time domain can be exploited more effectively in retrieval algorithms for snow characteristics and freeze/thaw status. Intensive in situ measurement of snow, soil, and vegetation will be conducted at this site to support the ground-based remote sensing. These measurements are described in Section 4.5.
Three ground-based remote
sensing systems are planned for the experiment: 1) active and passive microwave
sensors from the University of Michigan's Radiation Laboratory, 2) the
Japanese (NASDA) AMSR simulator, and 3) a FMCW radar system from the U.S.
Army Corps of Engineers Cold Regions Research and Engineering Laboratory
(CRREL). All three systems will be deployed at the Local-scale Observation
Site near the Fraser Experimental Forest Headquarters facility.
4.3.1. University of Michigan Systems
The University of Michigan (UM) L/C/X/Ku-band scatterometer is a calibrated radar system capable of measuring the amplitude and phase of the backscattered signal [Tassoudji, et al, 1989]. In the case of distributed targets, such as vegetation canopies, the calibrated Mueller matrix of the target can be measured accurately [Sarabandi, et al, 1992]. The system, which operates at L-band (1.1-1.4 GHZ), C-band (5.2-5.7 GHZ), X-band (9-10 GHZ), and Ku-band (15-17 GHZ) is mounted on a truck with the antennas and the RF equipment mounted on a platform at the top of a boom, and the control and processing equipment housed in a control room located at the bed of the truck. An HP 8753 vector network analyzer is used as the base transmit and receive unit and different up-converters and down-converters are used to shift the base frequency to the desired scatterometer frequency. The scatterometers operate in high PRF chirped pulse mode [Liepa, et al, 1989]. The PRF pulsing allows the system to reject short-range returns, including those from nearby targets, and range gating is accomplished by the time domain capability of the network analyzer.
The radar systems truck is a Ford F-800 model with a 7-liter V-8 gasoline engine (213 HP). The crane on the truck is an articulating one with a lift capacity of 990-lb and height of 56' (from the bed of the truck) for a fully extended boom at 75E elevation angle. The truck is also equipped with two 6-kW gasoline generators, one for use by the heating and cooling system and one for use by the van body interior lights and receptacles.
The UM truck-mounted radiometer system operates at L-band (1.4 GHZ) [Fischman, 2001]. This system measures radiometric brightness in V and H polarization. The L-band system will be fully polarimetric by the first IOP in February, 2002, enabling it to measure the full Stokes vector of upwelling brightness. The system has a 22E beamwidth. The receiver noise temperature is 59K, yielding a 1 second noise equivalent radiobrightness uncertainty (NEDT) of 0.2 K. Radiometers at 19.35-, 37-, and 85.5 GHz (the SSM/I frequencies) also exist. The 19- and 37-GHz units are V and H polarized, and the 85-GHz unit can be oriented for either V or H observations. These radiometers were specifically designed for long time-series data collection and have accumulated 500 days in the field, including a year on the north slope of Alaska.
The system will be mounted on a diesel-powered Norstar truck with a telescoping boom to 10-m in height. An elevation positioner at the end of the boom allows incidence angle variation from nadir to zenith. While the radiometer systems are configured to operate for long periods of time unattended, the truck requires weekly downtime of a few hours for maintenance. This will produce a gap of at least 4 hours (2 hours of actual maintenance, and 2 hours of warming the instruments up) per week in any continuous time series. Deployment of the19-85 GHz systems will require some engineering to integrate the instruments with the truck data system.
The two trucks housing the UM radars and radiometers will be stationed at the Local-Scale Observation Site (LSOS) near the Fraser Experimental Forest Headquarters during the 2003 IOPs (IOP-3 and IOP-4). The truck-mounted radars will be positioned in between an open snow field and a line of short trees (< 10-m height) while the radiometer truck will be positioned on the other side of the snow field opposite of the radar truck. In this configuration, the radars can be used to collect the polarimetric backscatter response of both the trees and the snow-covered surfaces without interfering with the radiometer data collection. Both trucks will remain stationed at LSOS for the entire 2003 experiment period (i.e. February through early April), and will most likely be moved into position during the Fall of 2002. The two trucks are equipped with AC generators, however, an external electrical power supply (connection to the electric grid at LSOS) ensures an uninterrupted system operation and less fuel consumption. Daily supply of gasoline and diesel fuel to the radar and radiometer trucks, respectively, will be needed for a round-the-clock data acquisition.
It is expected that during IOP-3 in February 2003, only dry, frozen snow conditions will be present. Instead of performing round-the-clock measurements of snow at a fixed incidence angle, both the radars and the radiometers will be used to measure the snow-covered surfaces at 3 incidence angles (20E, 35E, and 50E). For each truck system, the open snow field will be divided in azimuth into equal angular sectors. At each sector, the system will scan in elevation and collect data at the three incidence angles before moving to the next sector. At least 100 independent samples will be collected for each incidence angle. The time needed to complete a full azimuthal scan is 1 hour. The radiometer system is expected to repeat these measurements several times during the day.
Once the radar system completes its azimuthal scan over the snow field, it will turn around and perform another scan over the short trees area. The trees will be measured at two incidence angles (20E and 50E). Since the tree area is relatively small compared to the open snow field, it is expected that the radar truck would be repositioned away from the trees when the 50E incidence angle is measured. In this case, the radar system will scan in azimuth and collect data at 20E only, and at the end of the scan, will be repositioned for a second scan at 50E. This process will slow the data acquisition rate for the radar system. The radar system will alternate data collection between the open snow field and the short trees area several times a day.
A different data collection procedure will be adopted during IOP-4 in late March, 2003 to capture the impact of snow melting and tree-sap thawing on both radar and radiometer responses. In this procedure both radiometers and radars will be operated at a fixed incidence angle of 50E while scanning in azimuth. In addition, the radars will alternate between scanning the snow field and scanning the short trees area. The proposed diurnal measurements will be conducted on a continuous basis, 24 hours a day, over the duration of the second 2003 IOP.
Both the radar and radiometer systems will be calibrated several times a day (each calibration results in at most 30 minutes down time) to insure high quality data.
Although the radar truck system has four different frequency channels, only L-band (1.1-1.4 GHZ) and Ku-band (15-17 GHZ) will be used in this experiment. The trade-off here is the time required to scan using the other frequencies. Scanning additional frequencies is possible, but would reduce the total number of scans each day. For the radiometer truck system, only the L-band channel (1.4 GHZ) is truck-mounted and readily available for this experiment.
Deliverables for the UM radar component are the Average Mueller matrices (AM) of calibrated data. All relevant polarimetric information of the target can be derived from the AM matrix, including the backscattering coefficients, the phase difference statistics, etc. Each AM will be stored in a separate file whose header will include information about time, date, clutter type, incidence angle, center frequency, bandwidth, number of spatial samples, operator name, calibration files used, etc. In addition, several photos depicting the measurement site, the system, and measurement procedure will be provided. Final data sets will be provided within four months of the completion of IOP-4 (August, 2003).
Deliverables for the UM radiometer
component are Radio brightness in degrees Kelvin for both V and H polarizations
at a given incidence angle and frequency as a function of time. Stokes
brightness temperatures for 1.4 GHz and V- and H-pol brightness temperatures
for 6.9-, 19-, and 37 GHz will be provided at 30-minute intervals (55o
incidence angle) during both IOPs. The same brightness temperature data
will be measured at 7:00 am, noon, and 5:00 pm for 0-180o nadir
angle scans. Daily quick-look data will be made available during each IOP.
Calibrated data will be available on the UM server and at the data management
facility within 2 months following the end of IOP-4 (June, 2003).
4.3.2. Japanese Passive Microwave AMSR Simulator (GBMR-7)
A ground-based passive microwave
instrument (GBMR-7) will be provided by a team from Nagaoka University
of Technology in Japan at no cost to the experiment. The GBMR-7 has channels
at 18.7-GHz V/H, 23.8-GHz V, 36.5-GHz V/H, and 89.0-GHz V/H. The instrument
will be deployed at the LSOS site during IOP-1 and IOP-2. Details of the
data collection, processing, and products are still under negotiation with
Nagaoka University staff.
4.3.3. Ground-Based FMCW Radar Remote Sensing
Frequency modulated continuous wave (FMCW) radar has been used by CRREL investigators for a wide range of snow cover studies in Alaska, Greenland, Antarctica and Northern Vermont. These include radar backscatter cross-section measurements, snow depth measurements [e.g. Holmgren et al., 1998], detection of hoar layers in polar firn, and detection of melt channels in seasonal snow cover. These applications demonstrate the versatility of FMCW radars, as well as their ability to operate under harsh environmental conditions. Of particular importance to the objectives of this experiment is the fact that FMCW radars are able to observe the "electromagnetic stratigraphy" of snow and soil, i.e. the vertical structure of electromagnetic discontinuities that affect microwave scattering. The FMCW radar dataset will be important for relating snow and soil stratigraphic characteristics observed in snow pits during the experiment to microwave response.
FMCW Radar Theory. The salient feature of FMCW radar for this experiment is that it measures the "electromagnetic distance" to a target (i.e. electromagnetic discontinuities in a snow cover) from the instantaneous frequency difference between a target signal and a reference signal. The target signal refers to the frequency modulated signal (frequency that varies linearly with time) generated by a microwave source that is routed to a target and back via a transmitting and receiving antennas. A portion of the source signal that is routed via directional coupler to be mixed with the incoming target signal is referred to as the reference signal. The instantaneous frequency difference between the target and reference signal is obtained via a fast Fourier transformation of the mixed signal.
The equation governing the operation of FMCW radar is given by:
From (1) and (2), the following relationship between Fd and Tr is seen:
A graph depicting the relationship between these two parameters is presented in Figure 52 .
FMCW Radar at CRREL. The FMCW radar capability at CRREL covers a wide range of bandwidth starting from 0.5 GHZ and ending at 110 GHZ. These FMCW radars were designed and constructed at CRREL based on extensive field experiences in winter conditions (the radars were designed for robustness and easy maintenance). The modular radar system consists of a PC-based radar data acquisition system (a laptop-based data acquisition system is currently under development), a radar control unit, a microwave source, and a front end consisting of transmit and receive antennas ( Figure 53 ). For this experiment, the microwave source will consist of an HP8353B sweep oscillator with various RF plug-ins to cover radar frequencies from 1-40 GHZ. A backup data acquisition system and a microwave source are available to ensure success in the field experiment.
Bandwidth Diversity. A key feature of the FMCW radar is the wide bandwidth capability. This bandwidth diversity will allows us to easily duplicate the radar bandwidth of space-borne sensors ( Figure 54 ). The figure shows radar backscatter responses from a snow cover obtained using narrow (9.5-10.5 GHZ) and broad (8-12 GHZ) bandwidth FMCW configurations. The figure illustrates the ability of a broadband system to resolve several scattering layers in the snow (this is the typical operation mode of FMCW radar for snow cover studies; Figure 55 ). Figure 54 also shows that the FMCW radar can be operated at narrow bandwidth to mimic the operation of space-borne sensors. A calibrated, range integrated backscatter cross-section can easily be determined using calibrated targets in the field of view of the radar.
Polarization Diversity. Previous CRREL radar work did not require polarization-sensitive measurements. Therefore, our systems were limited to HH and VV measurements. However, to fully support the objectives of this experiment, the existing FMCW radars will be modified to also accommodate HV and VH capability.
FMCW Radar Data Collection. CRREL will measure radar backscatter cross-sections at L-, and Ku-bands (C- and X-bands are also possible) during both campaigns in 2002 (IOP-1 and IOP-2) and again in 2003 (IOP-3 and IOP-4). The FMCW radars will be mounted on a motorized pan/tilt head. The pan/tilt head will allow us to vary the depression angle of the radar, and then sweep azimuthally across a snow covered field. At each depression angle, we will measure HH, VV, VH and HV backscatter. Corner reflectors will be used to provide calibrated radar backscatter cross-section.
CRREL will also conduct a
broadband FMCW radar survey of the snow site to determine the location
of electromagnetic discontinuities in the snow cover.
The collection of in situ
measurements of snow and soil characteristics is a major and critical component
of the field experiment. The objective of the ground data collection is
to provide high-quality, spatially intensive data sets to address experiment
objectives for validation, testing and development of microwave remote
sensing algorithms and land surface models. The ground data collection
component is primarily focused on quantifying the spatial mean, variance,
and distribution of snow and soil properties at scales up to 1-km2
(i.e. the Intensive Study Areas, or ISA). At the local-scale observation
site, the focus is on monitoring temporal changes.
4.4.1. ISA Data Collection Strategy
The sampling strategy to be used for each of the 9 intensive study areas has been developed to meet four measurement objectives regarding key snow and soil properties:
A major goal of the ground data collection is to ensure that ground measurements are representative of the conditions at the time of the microwave remote sensing overflights. Risks of significant changes in ground conditions increases with each passing day, therefore the ground data collection strategy is to complete all measurements in a given MSA within a two-day period. Within any given two-day period, risks of significant change are not equal for different variables. First, snow surface wetness and roughness at the time of the overflight can strongly influence the remote sensing signal, and can also be very dynamic. Second, many internal snow pack and soil characteristics are much less dynamic than surface characteristics (depending on conditions). Next-day observations of these variables pose lesser risk. Third, recent changes in snow pack characteristics (e.g. new snow accumulation) can often be identified easily from snow pit information, but are more difficult to identify from surface observations. Therefore, the ground data collection strategy also prioritizes measurements to be made on Day 1 (the target day for airborne data collection in the study area), and measurements to be made on Day 2:
Day 1: Surface observations:
snow depth probing, with surface wetness and roughness observations.
Day 2: Snow pit and soil
observations.
This strategy requires six
days to complete the intensive observations in all three MSAs. Poor weather
is likely to be a factor at some time during each IOP, so seven days are
allotted to complete the ground surveys. Ground data collection can proceed
in overcast conditions that may include moderate precipitation (with consideration
for synchronization of airborne data collection). If weather conditions
on any given day prevent scheduled airborne data collection, the ground
data collection plan will adjust accordingly. Contingency plans will allow
for two additional days each IOP (for a total of nine days) to complete
data collection.
4.4.2. ISA Data Collection Plan
The baseline ground data collection plan for ISAs is based on 12 teams of 2 persons each (24 persons total). All three ISAs within a given MSA will be sampled simultaneously within a two-day period, using four teams per ISA. Each team will be responsible for a 500-m x 500-m quadrant of the ISA. Sampling on Day 1 (same day as airborne data collection) will consist of intensive spatial measurement of snow depth, surface wetness, and surface roughness. Sampling on Day 2 will consist of snow pit and soil measurements of several variables, described below.
Sampling Scheme. A stratified random sampling framework has been established for the collection of all snow and soil samples in the ISAs. A three-tiered approach is used to achieve objectives 1-3 listed above ( Figure 56 ):
Tier 1 (Snow Depth, Surface Wetness and Roughness)
1. The ISA is stratified using a 10 x 10 (100-m interval) grid.
2. X and Y coordinates of a starting point within each 100-m grid cell
are selected at random.
3. A transect interval of 5, 10, 15, 20, or 25-m is selected at random
for each grid cell.
4. Two orthogonal transects of two points each are defined using the starting
point and transect interval.
5. The direction of each transect is determined by the location of the
starting point within the grid cell.
Transects are oriented to ensure that the transect remains within the grid
cell. Thus, if the starting point is in the:
- lower left quadrant, the transects run north and east;
- upper left quadrant, the transects run south and east;
- upper right quadrant, the transects run south and west;
- lower right quadrant, the transects run north and west.
Tier 2 (Snow Depth, Surface Wetness and Roughness) - IOP-3 and IOP-4 only
Preliminary tests of this scheme suggest that it will satisfy the measurement objectives identified above. The scheme was tested using a Monte Carlo approach, using various arrays of 1000 x 1000 grid cells intended to represent different 1-m resolution snow distributions. Each array was sampled 10,000 times, and the results compared to the population mean, standard deviation, and coefficient of variation of the sampled array. The average error of the estimated mean was less than 0.5% for all cases, and the maximum error in any iteration for any case was less than 2%. Average errors for the estimated standard deviations and coefficients of variation were also less than 0.5%, but the maximum errors for these were as large as 16%. These results suggest that for snow depth, and snow surface wetness and roughness, the sampling scheme will provide a very reliable estimate of the mean, and a reasonably accurate estimate of the variance.
The 16 snow and soil pit locations (Tier 3) are expected to provide a reasonably accurate measure of the spatial distribution of snow density. This information will be used with the intensive snow depth measurements to determine the distribution of snow water equivalent throughout each ISA (past studies have shown that the spatial variability of snow density tends to be low and relatively insensitive to snow depth). For other snow pack and soil variables with greater spatial variability, 16 sample locations may not be sufficient for some objectives. However, realistically this is a large number of snow pits for each 1km2 area, and the data set will provide critical insight into snow and soil conditions.
To evaluate how well the sampling scheme will support quantification of the spatial structure of snow depth, surface wetness and surface roughness, distances between all possible point pairs (spatial lags) were calculated ( Figure 57 ). Of particular importance is that short lag distances are well represented. These are necessary to help extend understanding of physical processes at point scales upward to larger scales. A sufficient distribution of sample lag distances is obtained by this approach. An observed semivariogram, sampled from one of the 1000 x 1000 arrays noted above, demonstrates this ( Figure 58 ). This figure also reveals a limitation of the sampling scheme (decreasing semivariance beyond 700-800-m lag distance). It is standard practice in geostatistics to truncate semivariograms at approximately half the maximum lag distance. This is because sample pairs at distances greater than this must necessarily come from opposite edges and eventually opposite corners of the sampled domain. This artificially increases correlation between point pairs at these distances. The maximum lag distance of a 1000-m x 1000-m ISA is 1410 m, so semivariograms would normally be truncated at roughly 700-m lag distance. Ideally, the spatial analysis could be carried through to at least 1000-m, but larger ISAs do not appear to be feasible for many reasons.
To evaluate the fourth objective, the sampling scheme was field tested in three ISAs (St. Louis Creek, Fool Creek, and Walton Creek) in April, 2001. The tests established confidence that the scheme is reasonable, and that each ISA can be completed in two days by 12 field personnel (weather permitting). Poor weather will hamper data collection efforts. To reduce the risks of weather delays, all sample locations will be pre-located and marked with flags (in forested areas) or poles (in open areas). Also, field books will be prepared in advance for each team, with locations to be sampled and required measurements identified. These measures will allow data collection to proceed more efficiently, especially in foul weather.
Field Methods for ISA Sampling. At the Fraser and Rabbit Ears ISAs, teams will consist of two persons. Snow measurements on Day 1 in each ISA will be focused on collecting snow depth, snow surface wetness, snow surface roughness. Each team of two will be responsible for a 500-m x 500-m quadrant of the ISA, where they will have 125 measurements to collect over 25 100-m x 100-m grid cells. At each cell, teams will begin by measuring snow depth and surface wetness and roughness at the first (nearest) transect point. (See sections 4.4.3 - 4.4.5 below for details on methods).
At the North Park ISAs, teams will consist of three persons. Day 1 data collection will be identical here to the methods planned for Fraser and Rabbit Ears ISAs, with the addition of snow/soil interface temperature observations from the Tier 1 and Tier 2 locations. Once snow depth is determined at the first point, one team member will remain at the point and collect three measurements of the snow/soil interface temperature using a thermistor probe. Each measurement requires 1-2 minutes for the probe to equilibrate. This would require too much time to collect these measurements at every point in the transect. Meanwhile, the other two team members will proceed to the remaining transect points and collect the required measurements. When the temperature probing is complete, that team member will proceed to the first transect point of the next grid cell. When the transects of the first cell are completed, the two team members will also proceed to the first transect point of the next grid cell, where the process will continue. In this way, the three team members will remain in close proximity to one another throughout the survey, but will also be able to collect both fast and slow measurements simultaneously and efficiently.
Snow measurements on Day
2 in each ISA will be focused on snow pit and soil measurements from each
of the Tier 3 locations. Teams will consist of a minimum of two persons,
with most teams having three persons. Each team will be responsible for
collecting observations from four snow pits in their quadrant. This will
require a full day in Rabbit Ears and Fraser MSA, but should be completed
much more quickly in the North Park MSA.
4.4.3. Soil and Snow Measurements at the North Park MSA
During the two days of North Park ISA data collection, additional snow and soil measurements will be collected throughout the full MSA. The MSA will be divided into four quadrants, and four teams of two will conduct measurements in each quadrant. Ideally, this expanded data collection would occur in all three MSAs, but this isn't feasible due to rugged terrain and poor access. In contrast, most parts of North Park are accessible by 4WD vehicle in winter, allowing enhanced data collection in this MSA. North Park is the principal AMSR validation site in the experiment, making enhanced MSA data collection particularly important.
Each team will sample approximately
25 sites within their quadrant over a two-day period. They will measure
snow depth and density, snow temperature and snow/soil interface temperature,
and will collect two soil samples for gravimetric soil moisture analysis.
4.4.4. Ground Measurements at the Local-scale Observation Site
Measurements collected at
the local-scale observation (LSOS) site will focus on producing a comprehensive
assessment of the snow, soil, and vegetation characteristics viewed by
the ground-based remote sensing instruments. The viewed area itself will
not be disturbed. Unlike the other components of the experiment, which
focus on spatial distributions at relatively brief "snapshots" in time,
measurements at the local-scale site will focus on the temporal domain.
This site has been dubbed the "super-site" because it will contain the
most comprehensive array of instrumentation to measure snow, soil, and
vegetation properties in the field experiment. The 100-m x 100-m area will
be fenced to restrict access and preserve natural snow conditions in the
field-of-view of the ground-based remote sensing instruments. Measurements
will include daily in situ sampling of snow, soil, and vegetation characteristics
at several locations within the fenced area. Snow and soil measurements
will include snow and soil pits, snow depth probing, and snow/soil interface
temperature probing. The stem and canopy temperature, and xylem flux of
several trees within the area will be monitored. Two micrometeorological
towers, one located in the open snow area and the other in the forested
area, will monitor ambient conditions and provide forcing data sets for
1-D snow/soil models. These measurements, together with the ground-based
remote sensing, will provide the framework for evaluating and improving
microwave radiative transfer models and coupling them to land surface models.
Sampling locations will be designated on-site after the ground-based remote
sensing vehicles are situated. Details of these measurements are provided
in the next section.
Snow Depth Measurement (Probing). Snow depth will be measured at each Tier 1, Tier 2, Tier 3, and LSOS sample location. It will be measured to the nearest 0.01 m by probing vertically into the snow pack using a collapsible, graduated snow probe.
Snow Surface Wetness Observation. The approximate liquid water content of the upper 3-cm of the snow cover will be observed at each Tier 1, Tier 2, Tier 3, and LSOS sample location. The general classification scheme of the International Association of Scientific Hydrology (IAHS) [Colbeck et al., 1990]) will be used (Table 19). The percentage of liquid water by volume is classified as dry (0%), moist (<3%), wet (3-8%), very wet (8-15%), and slush (>15%). The determination is based on observation of the behavior of the snow when pressed together. Instrumental methods of measuring liquid water content are either too time-consuming for intensive measurements, or have substantial uncertainty associated with them.
Snow Surface Roughness Observation. The snow surface roughness will be observed at each Tier 1, Tier 2, Tier 3, and LSOS sample location. The average depth of surface irregularities will be estimated in the field and recorded. The type of roughness (smooth, wavy, concave furrows, convex furrows, or random furrows), the wavelength, and the orientation will also be noted. The snow surface roughness will be recorded for subsequent photogrammetric analysis by taking a digital photograph of a black backdrop inserted into the snow pack.
Vertical Snow Profile
Measurements. Measurements of the vertical profile of several snow
properties will be collected from each Tier 3 location and from designated
LSOS locations. Snow pit measurements will be taken from the vertical wall
of each pit. Significant layer boundaries will be observed, using a paint
or drywall brush as necessary to distinguish layers. From each significant
layer, three replicates of grain size (long-axis) measurements will be
made using a standard crystal card and loupe-style hand lens. Snow density
will be measured at 10-cm intervals using a 1000-cm3 snow density
cutter and electronic scale. Snow temperature will be measured at 10-cm
intervals using a calibrated dial-stem thermometer. Snow wetness will be
observed on the surface in every case, and at 10-cm intervals if the snow
temperature at that interval is greater than or equal to 0EC.
Gravimetric soil moisture will be collected at approximately 100 sites within each of the three MSA during BG-1 to support calibration of background gamma data collection. Soil bulk density will be measured at each Tier 3 site in the Fall of 2002 to support micrometeorological measurements and 1-D physical modeling. The bulk density will be measured at five intervals to a total depth of 40-cm. Soil samples from each of the 192 Tier 3 sites will be analyzed for general characterization and major hydraulic properties. Also in the fall of 2002, a thermistor string will also be installed near each site, for future temperature profile sampling during the IOPs. At selected sites, column wet biomass and height of the surface litter or understory vegetation will be determined by weighing samples from 1-m2 areas representing the major observed types of litter/understory. Corresponding dry biomass will be obtained by drying the wet samples at 70EC until equilibrium is reached.
Winter soil measurements will be simpler, and will focus on determining whether the soil is frozen or thawed, and measuring the water content. The temperature profile of the soil will be measured using the pre-installed thermistor string. Three replicate 20-cm soil cores will be collected at the base of each snow pit. One of these will be split on-site and inspected for evidence of ice. The other two will be segmented on-site, bagged, weighed, and removed from the site for later analysis. Where possible, a set of second-level (20-40 cm) samples will also be collected in the same method.
Soil moisture and temperature
will also be monitored at five sites within each ISA, including the micrometeorological
site (discussed in the following section).
4.4.7. Micrometeorological Measurements
A single complete instrument station will be deployed near a Tier 3 site close to the center of each of the nine ISA in the fall of 2002. These stations are designed to provide fundamental micrometeorological observations that will be used to support analysis and modeling investigations at the ISA scale. These stations will measure:
Two comparable instrument stations will be deployed at the LSOS site. One tower is located in a wooded area, and is operated by personnel at the Fraser Experimental Forest station. Data from this station will be provided at no cost to the experiment. The second will be deployed in an open snow field, and is operated by the University of Michigan as part of their ground-based radiometer system.
A total of 11 instrument stations will be deployed. These stations will be operated continuously from October, 2002 through July, 2003 to monitor the snow accumulation and ablation seasons. Prior to their initial field installation in September, and following their removal in June, all instruments will undergo a two-week cross-calibration at the LSOS site. The temperature, humidity, and wind speed sensors will be recalibrated periodically during their field deployment.
Mini-Stations. In addition to the complete micrometeorological stations, four mini stations will be deployed in each ISA. These are designed to support spatial analysis of air temperature and relative humidity, and soil moisture and temperature within each ISA, augmenting the single site measurements provided by the micromet stations. The stations will be located at the four corners of each ISA, creating a five-point pattern of observations within the ISA. The stations will measure:
1. Air Temperature and Relative
Humidity (shielded HOBO HO8-003-02)
2. Soil Moisture and Temperature
(2 levels) (Vitel Hydra)
These instruments will be
powered by a battery and a small solar panel. These instruments will be
operated continuously from October, 2002 through July, 2003. These data
will be logged using a HOBO data logger. The data will be retrieved from
the logger in July, 2003.
4.4.8. Regional Snow Measurement Networks
Three large-scale snow measurement and monitoring networks currently exist within the Small-Regional Study Area: 1) SNOTELs, 2) Snow Courses, and 3) Airborne Gamma surveys.
SNOTEL. The National Resource Conservation Service (NRCS) operates over 50 automated SNOw TELemetry sites within the area. These sites monitor snow water equivalent on either an hourly or daily basis using a pressure-sensing pillow. Other hydrometeorological variables, including temperature and precipitation, are also collected. Data are telemetered using meteor-burst technology to a central processing facility in Portland, Oregon. Data are generally available a few hours after they are collected. Data that have undergone improved quality control measures are available within 24 hours. The SNOTEL locations are selected for statistical runoff and water supply forecasting procedures, rather than for providing a spatially representative sample of SWE. Locations tend to be in small forest clearings at similar elevations within a given region.
Snow Courses. Manual snow courses are conducted principally by the National Resources Conservation Service, and to a lesser extent by other federal, state, and local agencies. There are 60-70 manual snow courses within the area. These are transects of varying lengths that are generally measured once or twice a month during the winter and spring. Not all of the NRCS snow course sites may be active in any given year. Many are co-located with SNOTEL sites. In many cases snow course records go back several decades.
Gamma Snow Surveys.
The NOHRSC maintains about 40 operational airborne gamma survey flight
lines within the study area, in addition to the flight lines created for
this experiment. Not all lines are flown each year. Lines are selected
based on operational demand.