Indiana University-Purdue University Indianapolis (IUPUI) / This thesis applies several different remote sensing techniques to data collected from 2005 to 2007 on central Indiana reservoirs to determine the best performing algorithms in estimating the cyanobacterial pigments chlorophyll a and phycocyanin. This thesis is a set of three scientific papers either in press or review at the time this thesis is published. The first paper describes using a curve fitting model as a novel approach to estimating cyanobacterial pigments from field spectra. The second paper compares the previous method with additional methods, band ratio and semi-empirical algorithms, commonly used in remote sensing. The third paper describes using a partial least squares (PLS) method as a novel approach to estimate cyanobacterial pigments from field spectra. While the three papers had different methodologies and cannot be directly compared, the results from all three studies suggest that no type of algorithm greatly outperformed another in estimating chlorophyll a on central Indiana reservoirs. However, algorithms that account for increased complexity, such as the stepwise regression band ratio (also known as 3-band tuning), curve fitting, and PLS, were able to predict phycocyanin with greater confidence.

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research. / Indiana University-Purdue University Indianapolis (IUPUI)

Indiana University-Purdue University Indianapolis (IUPUI) / Resource managers need to be able to quickly and accurately map aquatic plants in freshwater lakes and ponds for regulatory purposes, to monitor the health of native species and to monitor the spread of invasive species. Site surveys and transects are expensive and time consuming, and low resolution imagery is not detailed enough to map multiple, small lakes spread out over large areas. This study evaluated methods for mapping aquatic plants using high resolution Quickbird satellite imagery obtained in 2007 and 2008. The study area included nine lakes in northern Indiana chosen because they are used for recreation, have residential development along their shorelines, support a diverse wildlife population, and are susceptible to invasive species. An unsupervised classification was used to develop two levels of classification. The Level I classification divided the vegetation into detailed classes of emergent and submerged vegetation based on plant structure. In the Level II classification, these classes were combined into more general categories. Overall accuracy of the Level I classification was 68% for the 2007 imagery and 58% for the 2008 imagery. The overall accuracy of the Level II classification was higher for both the 2007 and 2008 imagery at 75% and 74%, respectively. Classes containing bulrushes were the least accurately mapped in the Level I classification. In the Level II classification, the least accurately mapped class was submerged vegetation. Water and man-made surfaces were mapped with the highest degree of accuracy in both classification schemes. Overhanging trees and shore vegetation contributed to classification error. Overall, results of this research suggest that high resolution imagery provides useful information for natural resource managers. It is most applicable to mapping general aquatic vegetation categories, such as submerged and emergent vegetation, and providing general estimates of plant coverage in lakes. Better methods for mapping individual species, species assemblages, and submerged vegetation constitute areas for further research.