3. VEGETATION AND LAND COVER
Vegetation data is needed in deriving soil moisture from the microwave observations at a 1 km scale. The basic strategy that will be used involves three components. First, vegetation characteristics will be measured at various locations on the ground that represent an appropriate range of conditions. Then, satellite observations will be used to perform a land classification. On location surveys will be used as part of this supervised approach. This classification will be linked to the ground observations of parameters and used to predict values for each pixel. Classification would be done using TM data resulting in a 30 m data base. However, because the availability of TM data is never certain a plan will be developed that uses both TM and AVHRR.
3.1. Vegetation Sampling
3.1.1. Sampling Plan
Possible stratified random sampling allocation of resources (historic Landsat TM images will be used to locate the sites). The intended allocation is shown in Table 6.
Table 6. Proposed Distribution of Vegetation Sampling Sites | |||||
|---|---|---|---|---|---|
Location | Prairie (30%) | Pasture (50%) | Wheat (10%) | Crops (10%) | Total |
Little Washita (1/3) | 6 | 10 | 2 | 2 | 20 |
El Reno (1/6) | 3 | 5 | 1 | 1 | 10 |
ARM/CART (1/6) | 3 | 5 | 1 | 1 | 10 |
Other (1/3) | 6 | 10 | 2 | 2 | 20 |
Total Sites | 18 | 30 | 6 | 6 | 60 |
For logistics, all sites used for gravimetric and profile soil moisture sampling will be used for vegetation measurements. However, additional sites will be required. Some considerations in selection include:
1. Minimum field size: 300 m by 300 m (approx 20 acres).
2. Three individual samples per site (separated by at least 100 m and at least 100 m from edge)
3. Each sample from 0.5 m2 area (0.71 m by 0.71 m)
3.1.2. Resource Requirements
For each of the 60 sites there will be 3 replications (180 samples). An attempt will be made to complete sampling cycle within 2 weeks in order to revisit and resample sites with actively growing vegetation. Teams will consist of one experienced researcher and one assistant each. Each team can collect data and samples from 3 sites/day or 9 samples/day. Three teams will be involved.
Each team will a GPS receiver (1-5 meter accuracy), Plant Canopy Analyzer LAI-2000 (for indirect LAI), Accu-PAR (for fraction absorbed PAR), 35-mm camera and film (for % cover and documentation), electronic balance (for fresh weights in the field), clippers, sampling frame, bags, markers, measuring tapes, labels, etc.
In addition drying facilities for the samples are required. Large capacity dryers for plant samples are available at Chickasha and El Reno. Plant samples in paper bags require 5- 7 days to dry at 70-80oC. If samples average 1 ft3 each, then 180 ft3 dryer space will be required.
3.2. Land Cover Classification
Plans call for the generation of a land cover map for the entire study region. The planned base for this is TM data acquired during the experiment time frame. Classification will be enhanced using supervised techniques based upon field surveys of vegetation/cover conditions conducted prior to and during the experiment.
To date TM scenes have been acquired to assist in various aspects of the planning process and to develop information for the classifier and field survey program.
TM scenes (Path/Row) 28/34, 28/35 and 28/36
Four dates for each scene (April 4, 1991; July 9, 1991; August 26, 1991;
September 11, 1991
Classification will focus on the aircraft mapping area. It is anticipated that TM scenes for 1997 study would be from the same path/rows. Up to three dates, three scenes could be acquired. Scheduled overpasses for scene 28/34 (Path/Row) are;
April 21
May 6, 22
June 7, 23
July 9, 25
August 7, 23
NDVI images derived from the July 9, 1991 TM data are presented in Figure 12 (Little Washita
area), Figure 13, El Reno Area, and Figure 14, (Central Facility)
3.3. CASI Aircraft Based Multispectral Data Collection
The CASI imaging spectrometer, a commercial sensor manufactured by ITRES Research Ltd., has undergone extensive evaluation in remote sensing projects around the world. The instrument that will be used in this experiment was acquired under NSERC support, is a commercial unit with some custom features to enhance its utility for research purposes
In the CASI optical design (Anger et al. 1990) a reflection grating provides spectral dispersion of the incoming optical signal over a spectral range of 403 NM to 947 NM (for CASI302) with a spatial resolution of 512 pixels across the 37.9 degree field of view (FOV). Ground resolution ranges from one to ten meters depending on the aircraft altitude. The spectral resolution is nominally 2.5 NM FWHM (full width, half maximum), with 288 spectral channels centered at 1.8 NM intervals. The CCD sensor is read out and digitized to 12 bits by a programmable electronics system which is controlled by an internal single-board computer. Data are recorded on dual built-in digital 8500 Exabyte tape recorder which uses 8 mm cassettes. This low cost, standardized, data storage medium greatly facilitates post processing of the data. Each tape can store up to 2.5 gigabytes of data or depending on the frame rate, up to one hour of imagery. Representative values for the frame rate under typical conditions is 60 frames (lines) /sec for six spectral bands and 37 frames /sec for 16 spectral bands.
Because of the high data rate of the CASI sensor under normal operating conditions various user-selectable operating modes are employed in the CASI system. Each mode maximizes the information content while keeping the data rate at a manageable level.
In the Spatial mode, imagery is obtained at full spatial resolution of 512 spatial pixels across the full swath. Band centre wavelength and bandwidth are operator programmable for up to 18 bands.
In the Spectral mode, imagery is generated at a full spectral resolution of 288 channels for normally up to 39 look directions across the full swath. Look direction spacing and centre location are user specified to sample the array. This sampling normally produces an image rake or comb. A single channel full spatial scene recovery channel can be selected to aid in scene orientation when viewing the imagery.
In HyperSpectral mode imagery is generated by decimating the 288 channels by any integral value that is evenly divisible into 288 i.e. (2, 3, 4, 6, 9 ...). The number of look directions is increased. If a value of 4 is selected, for example, 72 spectral channels with nominal bandwidth of 8 nm are generated in 405 looks. Contact, Lawrence Gray (gray@isl.ists.ca) for further information regarding this or any other aspect of CASI operations.
The CASI data tapes are calibrated to radiance at ISTS. After recovery of the data from the tapes standard processes are applied which compensate for electronic offset and scattered light and frame shift smear within the system. A dark offset correction is then applied. Radiometric calibration of the imagery is undertaken based on calibrations undertaken at ISL at ISTS using software written by EOL staff at ISTS.
The ISTS CASI is also equipped with a roll and pitch correction system. A vertical gyro provides real time pitch & roll aircraft attitude data which is integrated within the CASI data stream written to tape. GPS data from a Novatel receiver is also integrated within the data stream written to tape. A GPS base station is also operated to provide differential correction of the airborne data. This information is used in the postprocessing of the data to produce geo-referenced images.
As a custom feature, the CASI sensor is equipped with a dual optical fibre input fixed to the entrance slit of the spectrograph providing the ability to sample as part of the recorded data stream the spectral content the illumination field. Two cosine receptors, one on the aircraft roof and another on the aircraft belly, provide a measure of up welling and down welling irradiance. A zenith sky radiance probe, also mounted in the roof of the aircraft, is usually multiplexed between the up welling irradiance probe with a switch box.
A program of design, evaluation, calibration and improvement of the diffuser performance is underway to assess the ability of our CASI to provide direct measurements of at-sensor reflectance and estimated surface reflectance.