**2.5. Estimating chlorophyll-***a* **concentrations in the coastal St. Jones River Watershed using AISA images**

The St. Jones River Watershed (**Figure 24**) is located in central Kent County, Delaware, and drains a portion of the coastal plain. The total area of the St. Jones River Watershed is 23,300 ha, including most of the capital city of Dover and a portion of Dover Air Force Base [157]. The upper St. Jones River is impounded by a dam 17 km upstream from the bay to form Silver Lake, a municipal recreation area. Much of the eastern portion of the watershed consists of wetlands, agricultural lands, and forests, including lands and waters managed for waterfowl, wild turkey, deer, and other wildlife.

Relationship between Land Use and Water Quality and its Assessment Using Hyperspectral Remote... http://dx.doi.org/10.5772/66620 205

**Figure 24.** Location of the study watershed within the state of Delaware [158].

that would not be seen by conventional monitoring program which probably only involve several sampling stations in the same area. However, the retrieval models in this study were derived from the dataset that is specific to the study area, which do not necessarily represent the coastal waters in other areas. The accuracy of such algorithms is always subject to the loca-

**2.5. Estimating chlorophyll-***a* **concentrations in the coastal St. Jones River Watershed** 

The St. Jones River Watershed (**Figure 24**) is located in central Kent County, Delaware, and drains a portion of the coastal plain. The total area of the St. Jones River Watershed is 23,300 ha, including most of the capital city of Dover and a portion of Dover Air Force Base [157]. The upper St. Jones River is impounded by a dam 17 km upstream from the bay to form Silver Lake, a municipal recreation area. Much of the eastern portion of the watershed consists of wetlands, agricultural lands, and forests, including lands and waters managed for waterfowl,

**Figure 23.** Maps of Chl-*a*, TSS concentrations, and CDOM absorption at the mouth for the Patuxent River, summer 2005

tion of ground-truthing dataset.

204 Water Quality

**using AISA images**

[150].

wild turkey, deer, and other wildlife.

The percentage of various land uses in the St. Jones River Watershed in 1992 and 2001 was calculated based on the USGS land use/cover map of Delaware in 1992 and 2001, and there were significant changes in land use in the St. Jones River Watershed during this period (**Table 1**). While more lands were converted to urban and agricultural land, there was significant loss of wetland and open water in the St. Jones River Watershed. However, agriculture is always dominant land use in the St. Jones River Watershed at both time periods. Thus, it is important to examine chlorophyll-*a*, an indicator of eutrophication induced by agriculture runoff, in water bodies.


**Table 1.** Percentage of land use in the St. Jones River Watershed in 1992 and 2001 [160].

In this watershed, eight water segments were in the 303 list for nutrients, dissolved oxygen (DO), and bacteria that were contributed from non-point sources (NPS) [19]. Many stations experience low DO levels less than the state minimum WQS of 4 mg/L with elevated chlorophyll-*a* levels. Also, nutrient concentrations at most of stations were high enough to support algal growth [19]. Therefore, it is essential to monitor chlorophyll-*a* concentrations in water bodies in the St. Jones Watershed.

The Airborne Imaging Spectrometer for Applications (AISA) is a commercially produced hyperspectral push-broom-type imaging spectrometer system capable of collecting data within a spectral range of 430–900 nm. The AISA airborne system, manufactured by Spectral Imaging (Finland), offers a high degree of positional accuracy provided by the combination of a differential global positional system (DGPS) with an integrated navigation system (INS) and real-time Kalman filter.

Aligned with the second goal in this chapter, this study determined if hyperspectral data could be used to measure water quality indicators such as chlorophyll-*a*, TSS, and CDOM in Delaware coastal waters. The correlation between a spectral index and in situ chlorophyll-a concentration was examined. Spectral indices were compared to determine the best index used for retrieval of chlorophyll-*a* from Delaware coastal waters with high confidence.

From July 20 to July 23, 2004, Delaware State University and the NOAA-sponsored Environmental Cooperative Science Center (ECSC) conducted a hyperspectral imaging flyover of the St. Jones River Watershed, Delaware. AISA data were acquired by the Center for Advanced Land Management Information Technologies (CALMIT) at the University of Nebraska using the AISA-Classic as the primary on-board sensor flown in a specially modified Piper Saratoga aircraft. The AISA sensor was configured for 35 spectral bands from 441 to 873 nm, and bandwidths varied for different portions of the spectrum. During the flyover, 25 flight lines (strips) of hyperspectral imagery were obtained between 11 AM and 2 PM covering the major parts of the St. Jones River Watershed. A ground resolution of 3.0 m was achieved by flying at an average altitude of 1500 m. One of the goals for this flyover was to determine if hyperspectral data could be used to measure water quality indicators such as chlorophyll-a, TSS, and CDOM in Delaware coastal waters.

Ground-truthing measurements included water sampling for laboratory analysis (chlorophyll-a, nutrients, and total suspended solids) and in situ measurement of water quality parameters including salinity, conductivity, dissolved oxygen, and temperature. Water samples were collected as close to the aircraft overpass as possible, with the exact coordinates of the sampling stations determined using a Garman GPS unit. All samples were taken from the 0- to 0.4-m surface layer. After acquisition, the AISA images were radiometrically and geometrically corrected by CALMIT using calibration files fed to AISA CaliGeo® Software and converted to Environment for Visualizing Images (ENVI) file format.

ENVI software (RSI, Version 4.2) was used to process preprocessed AISA images. AISA images for the study area and one shape file of sampling sites (GPS points) were loaded into ENVI. To reduce noise in the data, one 3 × 3 kernel window was used at each sampling site to extract spectral profiles, which include values for each band. Then, all spectral profiles were exported and input into Excel for further processing.

In Excel, NIR/Red "index" values were calculated based on different combinations of NIR and red bands, using one near infrared band between 670 and 780 nm and one red band between 620 and 670 nm. Fifty different NIR/red ratios were computed in total. Correlations between spectral indices and in situ chlorophyll-a data were examined and the significance of the correlations tested. If there was a significant relationship between an index and measured chlorophyll-*a*, the index was examined again to see if it had any significant relationship with measured TSS. The residual difference between estimated chlorophyll-*a* and measured chlorophyll-*a* was also tested to determine if there was a significant relationship between the residuals and measured TSS for each initially identified spectral index. This procedure was used to determine if TSS was potentially responsible for any estimate bias in the spectral indices. Based on all three procedures, indices that were a good fit for retrieving chlorophyll-*a* in Delaware coastal waters were finally identified.

In this watershed, eight water segments were in the 303 list for nutrients, dissolved oxygen (DO), and bacteria that were contributed from non-point sources (NPS) [19]. Many stations experience low DO levels less than the state minimum WQS of 4 mg/L with elevated chlorophyll-*a* levels. Also, nutrient concentrations at most of stations were high enough to support algal growth [19]. Therefore, it is essential to monitor chlorophyll-*a* concentrations in water

The Airborne Imaging Spectrometer for Applications (AISA) is a commercially produced hyperspectral push-broom-type imaging spectrometer system capable of collecting data within a spectral range of 430–900 nm. The AISA airborne system, manufactured by Spectral Imaging (Finland), offers a high degree of positional accuracy provided by the combination of a differential global positional system (DGPS) with an integrated navigation system (INS)

Aligned with the second goal in this chapter, this study determined if hyperspectral data could be used to measure water quality indicators such as chlorophyll-*a*, TSS, and CDOM in Delaware coastal waters. The correlation between a spectral index and in situ chlorophyll-a concentration was examined. Spectral indices were compared to determine the best index used for retrieval of chlorophyll-*a* from Delaware coastal waters with high confidence.

From July 20 to July 23, 2004, Delaware State University and the NOAA-sponsored Environmental Cooperative Science Center (ECSC) conducted a hyperspectral imaging flyover of the St. Jones River Watershed, Delaware. AISA data were acquired by the Center for Advanced Land Management Information Technologies (CALMIT) at the University of Nebraska using the AISA-Classic as the primary on-board sensor flown in a specially modified Piper Saratoga aircraft. The AISA sensor was configured for 35 spectral bands from 441 to 873 nm, and bandwidths varied for different portions of the spectrum. During the flyover, 25 flight lines (strips) of hyperspectral imagery were obtained between 11 AM and 2 PM covering the major parts of the St. Jones River Watershed. A ground resolution of 3.0 m was achieved by flying at an average altitude of 1500 m. One of the goals for this flyover was to determine if hyperspectral data could be used to measure water quality indicators such as chlorophyll-a, TSS, and CDOM in

Ground-truthing measurements included water sampling for laboratory analysis (chlorophyll-a, nutrients, and total suspended solids) and in situ measurement of water quality parameters including salinity, conductivity, dissolved oxygen, and temperature. Water samples were collected as close to the aircraft overpass as possible, with the exact coordinates of the sampling stations determined using a Garman GPS unit. All samples were taken from the 0- to 0.4-m surface layer. After acquisition, the AISA images were radiometrically and geometrically corrected by CALMIT using calibration files fed to AISA CaliGeo® Software and

ENVI software (RSI, Version 4.2) was used to process preprocessed AISA images. AISA images for the study area and one shape file of sampling sites (GPS points) were loaded into ENVI. To reduce noise in the data, one 3 × 3 kernel window was used at each sampling site to extract spectral profiles, which include values for each band. Then, all spectral profiles were

converted to Environment for Visualizing Images (ENVI) file format.

exported and input into Excel for further processing.

bodies in the St. Jones Watershed.

206 Water Quality

and real-time Kalman filter.

Delaware coastal waters.

Results are shown in **Table 2**. It shows all ratio indices possessing a significant relationship with in situ chlorophyll-*a*. There were significant relationships between chlorophyll-a and ratios covering near infrared bands only from 692 to 705 nm. Gitelson [159, 160] stated that reflectance near 700 nm was sensitive to chlorophyll-a in inland and coastal waters, and our result confirmed his finding.


**Table 2.** Correlation coefficients between spectral indices and chlorophyll-a in the St. Jones River Watershed [160].

There was no significant relationship between the ratios and in situ TSS found. There was also no significant correlation between the residuals of estimated versus in situ chlorophyll-*a* and in situ TSS. These results demonstrated that TSS did not have significant impact on estimating chlorophyll-*a* in Waters in St. Jones River Watershed when using the indices identified from this study.

While significant, the correlation coefficients for sensitive spectral indices are not high (maximum *R* = 0.68). This may be caused by two reasons. First, there was a time gap between the time to collect AISA images and the time to obtain water quality samples. Second, no atmospheric correction has been performed on the AISA images in this study. Despite these limitations, significant correlations obtained from this study indicate that it is possible to monitor chlorophyll-*a* using AISA images in Delaware coastal water bodies.

Based on this study, some simple ratio algorithms using spectral bands in the red and rededge ranges were adequate for quantifying chlorophyll-*a* in coastal waters. These algorithms were validated in other environments such as Nebraska with some modification to the coefficients [160]. Therefore, there is a high potential that chlorophyll-*a* in coastal waters could be estimated using hyperspectral remote sensing without the interference from TSS. NIR/ red ratios covering red-edge bands from 692 to 705 nm were found to be useful for retrieving chlorophyll-*a* in Delaware coastal waters when using AISA images and they could work in similar situations in other coastal areas. This suggests the advantage of using ratio-based indices in estimating coastal chlorophyll-*a* concentrations by remote sensing.
