Coastal Services Center

---

Coastal Water Quality Remote Sensing Demonstration

---

Background | Technology | Results | Challenges | Lessons Learned

Background

The NOAA Coastal Services Center compared four different remote-sensing technologies to measure water quality in the coastal waters of Patuxent River, Maryland, on August 18-19 and October 12, 2004. The purpose of the study was to determine the current capabilities of remote sensing to operationally measure water quality parameters. Four vendors used different remote-sensing technologies to collect water quality data in downstream reaches of the River — an area of approximately 120 square kilometers.

Parameters of interest included chlorophyll concentration, temperature, and total suspended solids. Independent field measurements were collected by the Maryland Department of Natural Resources (DNR), the Maryland Department of the Environment (MDE), and the NOAA Coastal Services Center.

Map of Patuxent River, Maryland, study area and field station locations

Return to Top

Technology

Remote-sensing measurements of the Patuxent River were collected using two airborne hyperspectral imagers (Airborne Imaging Spectroradiometer for Applications, or AISA, and Compact Airborne Spectrographic Imager-2, or CASI-2), an airborne multispectral point sensor (SeaWiFS Aircraft Simulator III, or SAS III), and a high-resolution satellite imager (IKONOS). Independent ground control data were collected by the Maryland Department of Natural Resources (DNR), the Maryland Department of the Environment (MDE), and the NOAA Coastal Services Center during survey period.

For a comparison of the sensors used and their characteristics please see the table below:

Team Members	Instrument	Parameters	Resolution (meters)	Area Covered	Strengths
Spectrum Mapping and ENSR International	AISA	Chlorophyll-a and total suspended solids	2.5	Partial coverage	Rapid deployment Rapid access to preliminary information (24 hrs – 7 days)
EarthData International and ITRES Research Limited	CASI-2	Chlorophyll and total suspended solids	4	Partial coverage	High spectral resolution (32 bands) Relative chlorophyll concentration trends identified well
Applied Analysis Incorporated and Space Imaging	IKONOS	Chlorophyll, CDOC, suspended minerals, and secchi depth	4	Full coverage – satellite image	Large spatial coverage Rapid turnaround (36 hours) Concentrations generally within 1 standard deviation from ground truth data
Horn Point Lab at the University of Maryland Center for Environmental Science	SAS III	Chlorophyll-a and sea surface temperature	point	Point coverage along area of interest	Useful for long-term monitoring as opposed to acute events

Return to Top

Results

The parameters measured were chlorophyll concentration, temperature, and total suspended solids. As few as two and as many as four different parameters were examined and extracted from the remote sensing data by the different vendors.

AISA

Spectrum Mapping and ENSR International used the Airborne Imaging Spectroradiometer for Applications (AISA) hyperspectral sensor to collect remote sensing data for chlorophyll and total suspended solids in the lower Patuxent River on August 18, 2004. The overall trend of the AISA chlorophyll data was similar to the ground control data; however, the range of values was about half the magnitude of the ground control data range. In general the AISA total suspended solids data were positively correlated with the ground control turbidity data. Because of cloud cover during the time of collection, the AISA data did not cover the full extent of the Patuxent River study area.

CASI-2

EarthData and ITRES Research Limited measured chlorophyll and total suspended solids from remote sensing data collected on August 19, 2004. While the spatial distribution of the two parameters was successfully portrayed, quantitative comparisons between the CASI-2 and ground control data were difficult to determine. Because of minimal cloud coverage, the spatial extent of the CASI-2 imagery was fairly high.

IKONOS

The IKONOS data provided the highest spatial coverage of the area of interest but were very noisy. Turbidity was used as a proxy for comparing to IKONOS suspended minerals data. Quantitative comparisons were not feasible, but qualitative comparisons revealed a general agreement between the two data sets. The discrepancy between the two data sets is likely because of different collection times for the ground control data and satellite data. Overall, the IKONOS sensor was able to capture the range and spatial variability of suspended minerals well.

SAS III

The SAS III chlorophyll-a data differed significantly from the ground control data collected a day earlier. The difference in the times of data acquisition is the most likely reason for the observed differences between the two data sets. Tidal influence, variances at different depths, instrument sensitivity, and calibration errors can affect the integrity of the measurements. The SAS III instrument seemed to be less sensitive to extremely high chlorophyll concentrations, since it did not detect the high values extant in certain areas of the lower Patuxent River.

IKONOS (collected 10/12/2004)	SAS III (collected 8/19/2004)
AISA (collected 8/18/2004)	CASI-2 (collected 8/19/2004)
Chlorophyll concentrations for each of the four sensor types. Field data for individual stations are represented as squares.

Return to Top

Challenges

Assessing water quality using remote sensing is generally difficult in the nearshore environment because of the spatial and temporal environmental variability and the optical complexity of inland waters. Some of the specific challenges of monitoring water quality with high-resolution remote sensing technology include the following:

• Partial coverage of the area of interest due to clouds, swath width, etc.
• Problems created by mosaicking adjacent images (edge-matching artifacts)
• Image features such as sea foam, wave trains, and cloud reflections, etc., that affect data quality
• Obtaining imagery coincident with the location and timing of field validation data collection given tides, currents, and winds

Return to Top

Lessons Learned

The demonstration of the four different remote-sensing techniques showed that the current technologies did a good job of conveying spatial variability of the parameters in question. Determining absolute accuracy was difficult because of the complexity of the experimental site. The demonstration illustrated a number of items that should be considered when thinking about the utility of remotely sensed water quality measurements for management applications.

Some of the lessons learned:

• During collection of ground-truth data to validate high-resolution remote sensing imagery, the exact boat location, time of data collection, drift while on station, and small variations in sampling depth can all introduce variability in the water quality data that are collected. Because of the high spatial variability, it was insufficient to visit a set of Global Positioning System (GPS) coordinates and start taking measurements. Boat drift and vertical mixing created by the boat itself must be monitored while taking samples.
• The end user of these data should be familiar with the ecosystem in order to be aware of any dynamics that might produce artifacts in the data sets. In this case, extremely high dinoflagellate concentrations at depths of one meter or more affected chlorophyll concentrations in the water samples but were not visible in the remote-sensing imagery because of the high chlorophyll concentrations in the overlying waters.
• No one sensor or technique was ideal. The different sensors and approaches had diverse strengths and weaknesses. Managers should consider the specific application they are attempting to address and identify the right technique for that application.

Return to Top