Under this NASA EOCAP project, we have made contact with a number of water quality experts and have begun involving them in various aspects of our project. Nearby organizations include the Monroe County Environmental Health Labs, New York State Department of Environmental Conservation, and the Genesee / Finger Lakes Regional Planning Council (GFLRPC). Additional individuals, representing national and international organizations are also being contacted.

A major thrust of this effort is the development and testing on improved algorithms for measuring and mapping the major constituents in optically-complex waters when viewed through a complex atmosphere. We have defined the end-to-end model for this effort and where possible acquired components from various other modelers.

This report and subsequent ones will include discussions of the common and unique HSI requirements relative to our task, applications and markets, and technology gaps as each are identified.

Work in Phase I consists of two main thrusts: 1) modeling of the radiometric non-linear mixing that occurs in remotely-sensed water; and 2) identifying applications and associated technology gaps.

We have identified two possible AVIRIS collections (one of the West Coast and one of Mon Lake) for initial algorithm testing. We have also requested an AVIRIS collection over our primary site on Lake Ontario. In addition, starting in April, RIT will be flying a series of collections with its Modular Imaging Spectrometer Instrument (MISI) over Lake Ontario and these data will be available for analysis on this effort.

RIT has implemented a baseline atmospheric correction model that uses the 760 nm oxygen absorption feature to determine pressure depth, ht 940 nm H₂0 feature to determine water vapor concentration, and a broad band visible fit to estimate aerosols. All these tools use a non-linear least squares fit (similar to that of Green et al. 1993) of MODTRAN-modeled data to observed data. Our initial approach will be to use a correction over land for presssure depth and water vapor and over water for aerosols. Assuming a spatial homogenous atmosphere over the near-shore region, the atmospheric correction derived in this fashion will be compared to the baseline SEAWiFs approach.

A radiative transer model that links the MODTRAN atmospheric transport code, an RIT-developed air-water interface model and a hydrolite subsurface radiative transfer model is under development. The MODTRAN 4 code is presently running at RIT as is the hydrolite model in a form modified to ingest data generated by the air-water interface model. Ongoing work is focused on completion of the air-water interface model. We have also received the Canadian National Water Resource Institute data, including the optical properties of Lake Ontario waters which will constitue the baseline dats set for initiation of all of the modeling.

One of the significant challenges in using HSI for water quality monitoring is that most of the parameters measured on site today are not directly observable from hyperspectral imagery. However, there is reasonable hope that many of these parameters will produce secondary effects that can be observed (e.g., increased phosphates cause increased growth of algae, which can be observed). The identified HSI observables that apply to water quality monitoring are (1) water color; (2) dissolved organic carbon; (3) suspended particulate matter; (4) macrophytes (such as algae); and (5) temperature.

Table 1 shows the parameters that our regional user group indicated are currently monitored. The potential HSI observables are shown to the right.

Table 2 lists the applications identified to date along with the technology gaps that currently hinder the satisfaction of those applications.

For purposes of obtaining feedback on water quality monitoring issues, we have contacted a number of organizations. Our contacts fall logically into two groups--(1) regional and (2) the wider national/international community. The most active regional participants, which attended a meeting at RIT on Dec. 10, are listed in Table 3.

The following relevant information was gleaned from our meeting with users on Dec. 10:

1. The New York State DEC has a data collection program, but historical data is not well organized. A geographical information systems (GIS) approach was suggested to organizing data, particularly new data as it is acquired.

2. The Monroe County Health Dept. also has data, perhaps with better organization than the DECís.

3. Lots of data are available for smaller water bodies. The Great Lakes Monitoring Program presently monitors nine stations down the center of Lake Ontario. Current sampling is done in April/May and August. More frequent sampling (both temporally and spatially) is desired. There is a lag time of two years from when the data is collected to when it is used.

4. Spatial sampling of around one meter resolution is needed near the shore while the resolution can be much coarser for monitoring farther out in the Great Lakes.

5. HSI could probably best be used to extrapolate sparse in situ sampling over larger areas.

6. If HSI can separate out silt from clay, this would be signifcant in determining the source of the material (i.e., clay is associated with agricultural run-off whereas silt is associated with bank erosion).

7. High bacterial levels are associated with die-off of algae which turns from green to yellow. We should determine whether or not HSI can see this.

8. Near-shore observations are very event oriented. For example, a shift in wind can dramatically change results. For our collection program, we will try to get near-shore samples within 1/2 hour of overflights, and within two days for deep-lake sampling.

9. Cloud cover is important in modeling hydrodynamics.

Table 4 shows our national/international users group. These individuals were selected through a variety of methods including an internet search, personnal contacts, and referals. Each has expressed an interest in being involved is some way. Several people have contributed literature on the subject. A few selected visits will be made to individuals in this group. We expect to add others to this list as time goes on.

To help in communicating with our user groups, we have established a web site (www.cis.rit.edu/research/dirs) where we will report the progress of our work. The site presently contains a general description of the program and a sample survey form.

To help understand the common and the unique HSI requirements of our project and other potential applications, we are developing the following table showing required spatial resolution and spectral regions of importance. Each box contains letter designators for applications that can be satisfied with that particular combination of spatial resolution and spectral content. The applications include various water categories as well as other HSI applications. More detail will be added to this table as we converse with individuals who are studying other potential HSI applications.

Application Key: O: Open Oceans S: Streams

La: Large Lakes M: Minerals

Li: Littoral V: Vegetation

P: Ponds and Small Lakes D: Dirt/Soils

R: Large Rivers G: Gas and Oil Exploration

We can currently make the following observations relative to the water quality application. First, the blue spectral band is more significant than in most other applications. Unfortunately, due to the low reflectance of water and the high atmospheric scatter in the blue, signal-to-noise is a greater issue with our application than with many others. One way to increase the signal-to-noise is by cooling the detector arrays, a complicating factor with added expense.

We believe that the use of HSI for water quality monitoring will work best in areas where some information is already known about the water body.

Light-scattering interactions in optically-complex water present a non-linear mixing problem not encountered in applications such a mineral exploration and precision farming. Oceans tend to have less complex optics, suggesting that they should be easier to monitor with HSI than lakes.

Green, R. O., Roberts, D. A., Conel, J. E., "Estimation of aerosol optical depth, pressure elevation, water vapor and calculation of apparent surface reflectance from radiance measured by the airbourne visible/infrared imaging spectrometer using a radiative transfer code," SPIE Vol. 1937, Imaging Spectrometry of the Terrestrial Environment, pp. 2-11, (1993).