1. Geo-Spatial Assessment of the Impact of Land Cover Dynamics and the Distribution of Land Resources on Soil and Water Quality in the Santa Fe River Watershed By Aarthy Sabesan GIS Research Lab

2. Located in north-central Florida Mixed land use watershed covering 3,585 km 2 Encompasses parts of Suwannee, Gilchrist, Columbia, Union, Bradford, Alachua, Baker and Clay Administratively, Suwannee River Water Management District (SRWMD)

3. 1995 Land Use / Land Cover (LULC) classes

4. Soil Orders

5. Environmental Geology

6. 1.Depth to water 2.Net recharge 3.Aquifer media 4.Soil media 5.Topography 6.Impact of the vadose zone 7.Hydraulic conductivity DRASTIC Index

8. Non-point source pollutants are the major source of surface and ground water pollution in U.S today. Increasing concentrations of nitrate-nitrogen are observed in the surface water, ground water and springs in the SRWMD. Contribution of the SFRW has increased by 4% from 2001 to 2002. 2002, the Suwannee River Basin: 2,971 tons nitrate-nitrogen. SFRW (5.7% of the Suwannee River Basin): 19.6% of the N loads.

9. Hypotheses Spatially distributed patterns of land resources and land cover dynamics are useful proxies providing information about nitrogen levels in soils and surface water Land use / land cover (LULC) and soils are the major factors impacting soil and water nitrogen in the SFRW

10. Characterize the land cover dynamics in the SFRW from 1990 to present Quantify the spatial distribution of soil nitrate-nitrogen across the SFRW Investigate the spatial relationships between watershed characteristics and soil and water quality

11. Module 1 Land cover dynamics in the SFRW

12. Objective Identify recent changes within land cover classes Quantify the areal extent of these changes Assess the trend or nature of change within land cover classes

13. Materials BandWavelength (µm) Spectral location 10.45-0.52Blue 20.52-0.60Green 30.63-0.69Red 40.76-0.90Near-infrared 51.55-1.75Mid-infrared 610.4-12.5Thermal infrared 72.08-2.35Mid-infrared Landsat Satellite Series NASA and Dept. of Interior Spatial resolution – 30m

14. Path 17, Row 39 Landsat TM August 26 th 1990 August 13 th 2000 Landsat ETM+ February 11 th 2003

15. Methods 1.Design of a land cover classification scheme 2.Ground truth data collection 3.Image processing 4.Change trajectory analysis

16. Design of a Land Cover Classification Scheme Four levels of land use / land cover classification –Aggregation of level 2, 3 and 4 to create level 1 Land cover classes used for the analysis Coniferous pine, Upland forest, Agriculture, Rangeland,Urban,Wetland,Water

17. Ground Truth Data Collection 487 Ground Control Points (GCP’s) Categorization into training and accuracy assessment sites (60% / 40%)

18. Image Processing 1.Preprocessing –Geometric correction –Atmospheric correction –Noise removal 2.Pre-classification scene stratification 3.Image classification (Supervised approach) 4.Accuracy assessment

19. Preprocessing:Geometric Correction 2000 Landsat image imposed over the 2003 image RMS error: 0.5 pixel Correction for distortions in platform attributes

20. Preprocessing:Atmospheric Correction Dark object subtraction technique Based on the assumption that the reflectance from water bodies is close to zero. R DOSN = R * (R DO )/ ((Cos (90-θ)*  )/180) To account for atmospheric attenuation factors

21. Splitting the image into individual bands Header file R DOSN = R * (R DO )/ ((Cos (90- θ)*Π)/180) R DOSM = R * (R DO )/ ((Cos (90- θ)*Π)/180) Layer stacking the individually calibrated bands Atmospherically corrected Landsat image. R DOSN R DOSM Θ values Raw Landsat image Pixel value of the dark object in the particular band Identifying a dark object, like a water body Pixel value of the dark object in the particular band

22. Preprocessing: Noise Removal Masking cloud and cloud shadow Cloud / cloud shadow infested image Cloud / cloud shadow mask Cloud / cloud shadow masked image of SFRW

23. Pre-Classification Scene Stratification To separate spectrally similar classes of urban, agriculture and rangeland

24. Image Classification

25. Image Classification: Training Stage Numerical descriptors of land cover classes Two sets of spectral signatures were developed Summer scene Winter scene

26. Image Classification: Classification Stage Minimum Distance to Mean Classifier (MDM)

27. Image Classification: Output Stage 1990 2000

28. Image Classification: Output Stage 2003 Overall classification accuracy: 82%

29. Change Trajectory Analysis Three data change image of land cover change classes

30. Trajectories of Land Cover Change

31. Conclusions The multi-temporal change detection analysis indicates a increasing trend in agricultural intensification in the watershed Western part: expansion of agriculture on Ultisols and karst topography Eastern part: moderate to weak expansion in agriculture on Spodosols and clayey sand

32. Module 2 Quantify the spatial distribution of soil nitrate-nitrogen across the SFRW

33. Tasks Develop a site selection protocol to address the spatial variability of nitrate-nitrogen across the watershed using GIS techniques Soil sampling Laboratory analysis of nitrate-nitrogen Compare different interpolation techniques and identify the method with lowest prediction error Interpret soil nitrate-nitrogen in context of land resources within the SFRW

34. Land-use and soil combination raster (Illustrated here are the soil orders present under the urban land use class) Stratified Random Sampling Design

35. 101 sites were approved for September 2003 sampling Soil samples were collected at Layer 1 (0 to 30 cm), Layer 2 (30 to 60 cm), Layer 3 (60 to 120 cm) and Layer 4 (120 to 180 cm) Soil nitrate-nitrogen values (  g/g soil)

36. Layer 1 Spline with tension RMSPE: 1.455 Layer 2 Spline with tension RMSPE: 1.369

37. Layer 3 Inverse Distance Weighted RMSPE:1.904 Layer 4 Inverse Distance Weighted RMSPE:1.462

38. Average profile concentrations Spline with tension RMSPE: 1.306

39. Pixel Based Prediction of Soil Nitrate-Nitrogen Average nitrate-nitrogen profile values for each LULC-soil combination OPixel soil-N PPixel soil-N Based on LULC-soil combinations

40. Pixel-Based Prediction of Soil Nitrate- Nitrogen

41. Conclusion This analysis is the first step in characterizing the spatio-temporal variation of nitrate-nitrogen at a watershed scale The LCLU and the soil data support developing predictive models of soil nitrate-nitrogen in the SFRW

42. Water Quality Analysis Module 3

43. Objective Characterize the geographic position and distribution of land resources to understand spatial relationships between watershed characteristics and water quality data Materials Surface water and ground water quality data from SRWMD

44. Surface Water Quality Observations Time frame of observations: 1989 to 2003

47. Sub-Basin Attributes Land use / land cover class (2000) Soil order (SSURGO) Geology Mean, maximum and minimum DRASTIC values Mean, maximum and minimum soil organic carbon Mean, maximum and minimum population Mean, maximum and minimum elevation Mean, maximum and minimum slope

48. Results N-NO 3

54. Conclusion Results indicate that multiple factors contribute to elevated nitrogen found in soils and water Karst terrain, soil material, and agricultural and urban land uses pose the greatest risk for nitrate leaching In addition the geographic position and spatial distribution of land resource factors and spatial interrelationships between factors influence nitrogen levels observed in soils and surface water Understanding the interrelationships between land cover / land use, soils, geology, topography and other factors in a spatially-explicit context support ongoing efforts to improve the water quality in the SFRW

55. Acknowledgement My parents Dr. Sabine Grunwald (Chair) Dr. Mark Clark Dr. Michael Binford (Dept. of Geography) Christine Bliss and Isabel Lopez Sanjay Lamsal Kathleen McKee and Rosanna Rivero