2002 Progress Report: Application of Remotely-sensed Data To Regional Analysis and Assessment of Stream Temperature in the Pacific NorthwestEPA Grant Number: R827675
Title: Application of Remotely-sensed Data To Regional Analysis and Assessment of Stream Temperature in the Pacific Northwest
Investigators: Burges, Stephen J. , Booth, Derek B. , Gillespie, Alan R.
Institution: University of Washington - Seattle
EPA Project Officer: Packard, Benjamin H
Project Period: April 1, 2000 through March 31, 2003 (Extended to March 31, 2004)
Project Period Covered by this Report: April 1, 2002 through March 31, 2003
Project Amount: $998,395
RFA: Regional Scale Analysis and Assessment (1999) RFA Text | Recipients Lists
Research Category: Ecosystems , Ecological Indicators/Assessment/Restoration
The objectives of this research project are to: (1) develop efficient methods for regional assessments of stream temperature, and (2) demonstrate how these methods can be applied to assess effects of land use on stream temperature.
In Year 3 of the project, we have accomplished the activities described below.
Field Campaigns. In Year 3 of the project, we continued the collection of temperature data from the ground-based stream temperature monitoring network, as well as a few intensive field campaigns to collect ground truth for satellite-based remote sensing images.
Nearly 3 years of data now are available for most of Soos Creek and its tributaries. More than 1 year of data now are available for the Green, Cedar, and Yakima Rivers. Loggers data lost to theft and natural causes create some holes in the record, but analysis of the time series, as well as the spatial distribution of temperatures, is now possible.
Three intensive field campaigns were conducted during August and September of 2002. These campaigns involved from 5 to 12 volunteers who made detailed temperature surveys of the study rivers. Volunteers were provided with global positioning system units, temperature sensors, and field notebooks. The first campaign observed temperatures along Soos and Covington Creeks to identify suspected controls on local water temperatures. The second campaign involved floating down the Yakima River to collect information on how water temperatures changed with time and location. This also served as ground truth for Thermal Infrared (TIR) Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and Landsat images. The third ground campaign collected temperature observations in the Hanford Reach of the Columbia River to obtain ground truth on fully resolved river locations. The Green, Cedar, and Yakima Rivers are not sufficiently wide to guarantee fully resolved water pixels from the ASTER imager (90 m pixels), and the lakes used in the summer of 2001 were not well mixed enough to minimize differences between bulk and surface temperatures.
Analysis of along-stream temperatures to identify controls on local stream temperatures is underway. By plotting observed temperatures along the stream channel, they can be compared with other sources of data, including maps of vegetation cover and geology (see Figure 1). We hope that this analysis leads to the development of better tools for the identification of important monitoring sites along the stream. Using readily available regional information, such as vegetation cover and geology, should improve initial site selection.
Figure 1. Linearly Interpolated Stream Temperatures for the Green River on August 25, 2001, based on (a) In Situ Data Loggers; and (b) 5 m Resolution Moderate Resolution Imaging Spectroradiometer (MODIS)/ASTER (MASTER) Brightness Temperatures. The top figures show along-stream temperature. The bottom figures show along-stream elevation (white line), with along-stream land use above and along-stream geology below the elevation line.
Remotely Sensed Imagery. Efforts during this reporting period focused on remote sensing image accuracy and the ability to determine actual stream temperatures for fully resolved pixels. Kay (2002) conducted extensive tests to identify the effects of correcting MASTER and ASTER images for the effects of emissivity and atmospheric aerosols (primarily water vapor). A 1 percent change in the emissivity value can result in a 0.6°C change in observed brightness temperatures. Even atmospheric gases (primarily water vapor) can cause large changes in observed brightness temperatures, as shown in Figure 2.
Figure 2 also shows brightness temperatures for Lake Youngs, in the Green River watershed, for three MASTER passes. All images were taken on August 25, 2001, at different times of the day and at different altitudes (and resulting pixel resolution) of the National Aeronautics and Space Administration (NASA) MASTER aircraft. The image taken from an altitude of 6 km clearly shows temperatures 1 to 1.5°C cooler than the images taken from an altitude of 2 km, despite being taken later in the day. This difference is attributed largely to increased absorption of the TIR signal caused by the increase in water vapor between the airplane and the ground surface as it flies at higher altitudes. The cooling will be even worse for the satellite imagery, which records earth surface TIR images where the signal path length is the entire atmosphere.
To correct for the effects of the atmosphere, the temperature and water vapor profiles between the imaging device and the ground surface must be known. During the ground campaigns on August 25-29, 2001, we had access to comprehensive total column water (Microtops) and limited profile (radiosonde) measurements of atmospheric water vapor. Despite these efforts to constrain atmospheric water vapor, Kay (2002) found that the best corrections were actually based on total column water vapor predictions from MM5, a regional weather prediction model. This is a promising development, as MM5 predictions are easier to obtain than in-field observations, especially for images already taken.
Figure 2. MASTER Observed Brightness Temperatures of Lake Sawyer on August 25, 2001. Three flight lines are shown, all displayed with the same color scale. Images on the left were taken from an altitude of 2 km, resulting in 5 m pixels. The image on the right was taken from an altitude of 6 km producing 15 m pixels. Times indicate when the aircraft was over the visible path of the flight line.
To analyze the absolute accuracy of stream temperatures derived from TIR images, comparisons were made to ground truth observations. The remotely sensed TIR temperatures used for the comparisons had been corrected for the effects of emissivity and atmospheric aerosols, and were compared to kinetic (bulk) water temperatures taken using hand-held digital thermometers or in-stream temperature loggers. Handheld TIR devices originally were supposed to be used for the ground truth comparisons, but they proved unreliable under field conditions. After correction for water emissivity and atmospheric absorption and emission, 90 percent of remotely sensed lake temperatures are within 1°C of persistent and concurrent radiant and surface kinetic ground-truth temperatures. However, recovered stream temperatures deviate from gage measurements by up to 2.3°C. By propagating errors through the entire process of deriving TIR stream temperatures, Kay (2002) found that for fully resolved stream pixels, the resulting temperatures were accurate to ± 2°C.
Perhaps even more important than the absolute accuracy of obtaining stream temperatures from TIR images is the ability to resolve the actual stream. Figure 3 shows ASTER 90-meter resolution TIR images for four stream reaches. The figure clearly shows that the ASTER imager is guaranteed to view 100 percent water pixels only on the Columbia River. Fully resolved water pixels also may exist for the Yakima River, but most pixels will be mixed with the bank or riparian vegetation. The higher resolution MASTER images can resolve most of the Yakima and Green Rivers, but because of thick riparian vegetation, Covington Creek remains impossible to differentiate from shade.
Figure 3. 90 m TIR Images From ASTER for: (a) the Columbia River (300 to 500 m wide), (b) the Yakima River (50 to 100 m wide), (c) the Green River (10 to 50 m wide), and (d) Covington Creek (< 10 m wide). Red lines indicate the stream banks as observed from the 15-meter resolution visible ASTER bands.
Future activities include continuing field work that largely will involve the maintenance of the installed in-stream logger network and the meteorological station. This includes downloading data, replacing lost or damaged equipment, and merging new data into the existing archives.
The bulk of the continuing work on this project will involve the processing and analysis of existing data sets, including the publication of project results. Some of the research underway includes:
· Continued analysis of how well stream temperature dynamics are monitored by the in-stream temperature logging network.
· Analysis of the observed spatial variability in high resolution TIR images and what information about the stream it might contain.
· Analysis of the effect of spatial scaling on the ability to resolve streamtemperatures, including the effects of the near-bank environment and mixed pixels.
· Analysis of the image sharpening affects image accuracy, and whether it can be used to obtain useful information about stream temperatures from images with spatial resolutions too coarse for direct observation of the stream channel.
· Analysis of the Yakima River float data to identify the effects of local river conditions on stream temperatures.
· Analysis of the Hanford Reach data to further illustrate the issue of accuracy from TIR images.