The document summarizes a study that aimed to assess the impact of land use on water quality within hydrologically sensitive areas (HSAs) and entire watersheds in New Jersey. Key findings include: - Agricultural land and low-density urban land were primary contributors to nitrogen and phosphorus levels in streams. - Forest cover significantly reduced sediment levels compared to nutrients. - Wetlands unexpectedly increased nutrient levels, possibly by releasing accumulated phosphorus over time. - Future work will develop thresholds for defining HSAs and determine land use impacts on stream integrity at HSA and watershed scales.

1. Managing Critical Source Areas For
Enhancing Ecosystem Services in
Agricultural Landscapes
Subhasis Giri, Zeyuan Qiu
Department of Chemistry and Environmental Science
New Jersey Institute of Technology
July 29, 2015

2. Presentation Outline
Introduction
Research Motivation and Objectives
Methodology
Results and Discussions
Conclusions
Future Work

3. Introduction

4. LandScape Transformation
Source:https://www.google.com/search?q=population+growth+water
+pollution&source=lnms&tbm=isch&sa=X&ei=JcGjVbbnHcWq-
AHJ8rHABw&ved=0CAcQ_AUoAQ&biw=1280&bih=646#tbm=isch&
q=+water+pollution/
Water pollutions are related to
change in land use
characteristics( Gomi et al.,
2002; Kennen et al., 2010)
Increasing global population is
a major driver of landscape
conversion
Human interference is one of
the primary cause of landuse
change (Briassoulis, 2014)

5. Land Use Change
in New Jersey
Source: Hasse and Lathrop (2010)

6.  In U.S., 40.2 million acres (area greater than size of Illinois) were converted
to developed land between 1982 to 2007 (USDA-NRCS,2009)
 During this period, in New Jersey, approximately, 26.8 percent increase in
urban area while 24 percent lost in agricultural lands, 7 percent lost in
forested lands, and 5 percent lost in wetlands
 Newly developed lands are primarily low density residential, parking lots,
roads, and right of ways (NJDEP, 2010; Hasse and Lathrop)
Source: Hasse and Lathrop (2010)
Land Use Change in New Jersey

7. Understanding the Connection

8. Environmental Impact Economic Impact
Consequences
Source: http://begreen.botw.org/2011/12/impacts-of-algal-blooms-in-freshwater-ecosystems/
 Turbidity affects aquatic life
 Sedimentation change flow
direction
 Excessive nutrients causes
eutrophication which
ultimately leads to hypoxia
 Heavy metals threat to both
human and aquatic life
 Increased drinking water
purification cost
 Increased maintenance cost
(dredging )
 Negative effects on recreational
activities

9. Characterization of Landscape
Non-spatial landscape characterization
 Percentage of land uses(Johnson et al., 1997)
 Impervious cover (Schueler et al., 2009)
 Equal potential to affect water quality
Spatial landscape characterization
 Inverse distance weighted (Kennen et al., 2008)
 Riparian zone approach (Barker et al., 2006)
Out of two spatial landscape characterization
methods, riparian zone approach is preferred

10. Hydrologic Sensitive Area (HSA)
Smaller area in watershed having higher
propensity to generate runoff (Qiu et al., 2014)
Facilitates by variable source area hydrology
process (Walter et al., 2000)
Helps in quick movement of pollutants from
landscape to waterbodies

11. Hydrologic Sensitive Area (HSA)

12. Hydrologic Sensitive Area (HSA)

13. Research Motivation and
Objectives

14. Research Motivation
Studies have done to understand the connection
between water quality degradation and landscape
change. However, spatial variability of hydrological
connectivity using HSAs have not been
comprehensively reviewed
Specific contributor of urban land to pollution should be
identified
 Till date, no threshold is developed for topographic
index in defining HSAs

15. To assess the impact of land use to water quality
within HSAs and watershed scale using a linear mixed
model
To determine the relationship of land use to stream
integrity based on HSAs level and watershed scale
 To develop a threshold of topographic index in defining
HSAs in landscape that lead to ecosystem degradation
Main Research Objectives

16. Methodology

17. Study Area
Total of 28 watersheds
included in this study
Belongs to Valley and
Ridge, Highlands, and
Piedmonts
Minimum and maximum
watershed area was
5,930and 509,530 acres,
respectively
36% forest, 34% urban,
13% agriculture, 14%
wetland, and rest are
water

18. Finalization of water
quality station
Watershed delineation
Creating soil
topographic index
Creating hydrologic
sensitivity area
Extracting landuse of
hydrologic sensitivity area
Calculating area of each
landuse
Download data
Determine relationship
between water quality and
land use using linear
mixed model
Procedure at a Glimpse

19. Data Source
LIDAR DEM (10 ft × 10 ft) from New Jersey
Department of Environmental Protection (NJDEP)
Soil survey geographic database (SSURGO) soil from
USDA Geospatial Gateway
Modified Anderson classification land use form NJDEP
 2007-Landuse
Water quality data (suspended solid, nitrogen, and
phosphorus) downloaded from National Water Quality
Monitoring Council portal
Water quality station shape file obtained from NJDEP

20. Soil Topographic Index (STI)
It is the likelihood of a point in a watershed to generate
runoff (Qiu, 2009)
STI index identifies spatial distribution of runoff
contributing areas in a watershed (Walter et al., 2002)
 ln ln ………(1)
 α =upslope contributing area per unit contour
length(m)
 β=local surface slope (mm-1)
 = saturated hydraulic conductivity (m/day)
 D= depth to restrictive layer(m)
STI calculation is two fold processes:
 Creating soil transmissivity
 Formation of wetness index

21. Soil Transmissivity
Download soil data
Install soil data viewer
Import data into ArcGIS
Clip transmissivity
layer based on
Watershed boundary
Create saturated
hydraulic conductivity
shapefile
Convert shapefile to raster
Create soil depth
shapefile
Convert shapefile to raster
Weighted average
method
Cell size same as
LIDAR DEM
Multiplied
Transmisivity layer
(County basis)
Merged Transmisivity
layers
Re-project to
LIDAR DEM
projection

22. Soil Transmissivity Layer

23. Clip LIDAR DEM
based on Watershed
boundary
RSAGA in R
Wetness Index
Raster to Ascii
Fill LIDAR DEM
Slope calculation
Catchment area
calculation
Add transmissivity
layer
STI index
Wetness index

24. STI Index

25. Formation of HSA
HSA can be created using a threshold STI index
STI index targeted 20% of the watershed area used
as threshold value in a buffer study (Herron and
Hairsine,1998)
STI index 10 was selected as threshold and
approximately, 27% of total watershed area fall
under HSA
Extraction of HSA for 28 watersheds was performed
using python 2.7.3

26. Formation of HSA
Whole Watershed HSA

27. Land use Matrix
Land use of HSA and whole watershed was
extracted from 2007- land use
Extracted land use categories were: agricultural
land, forest, urban land-high medium density,
urban land-low density, rural residential, wetlands,
and water
Water quality data between 2006 to 2008 was use
to reflect the effect of 2007- land use on water
quality

28. Statistical Analysis
A linear mixed model was used in R using lme
function by Maximum likelihood method
 = β + +
 Yij = response variable (TSS/TN/TP) for
watershed i with j as repeated measures
 Xij = predictors (agricultural land, forest, urban
land- high medium density, urban land-low
density, rural residential, wetlands, and water )
 β =fixed effect among the predictors
 random effect due to unique characteristic of
watershed I
 εij is the residuals
AIC, BIC, and Loglik were estimated to compare
between HSA level and watershed scale model

29. Results

30. Visualization of Data
Spearman Correlation matrix based on whole watershed land use

31. Visualization of Data
Spearman Correlation matrix based on HSA land use

32. TN and land use matrix
Watershed-Scale HSA-Scale
Predictors β‐value p-value β‐value p-value
Intercept 0.308 0.000 0.304 0.000
Agricultural land 0.263 0.017* 0.205 0.085*
Urban land- low
density
0.424 0.006* 0.336 0.026*
Urban land-high
medium density
0.033 0.811
Wetland 0.053 0.536 0.090 0.375
Forest -0.108 0.391
Model Evaluation Statistic
AIC 349.37 351.01
BIC 375.21 376.84
Loglik -167.68 -168.50
* Represents statistically significant at 10 percent level of significance

33. TN and land use matrix
Agricultural land and urban land-low density have
significant positive impact on TN concentration based
on both watershed scale and HSA level analysis
Urban land- high medium density and wetland have
positive impact on TN concentration on watershed
scale
Forest is negatively contributing to TN concentration
based on HSA analysis
Tsegaye et al.(2006) and Wilson and Weng (2010)
also found that agriculture and urban area are primary
source of nitrogen in Wheeler Lake Basin Northen
Alabama and Southern Tennessee and Greater
Chicago area, respectively.

34. TP and land use matrix
Watershed-Scale HSA-Scale
Predictors β‐value p-value β‐value p-value
Intercept -2.821 0.000 -2.859 0.000
Agricultural land 0.301 0.066* 0.143 0.434
Urban land- low
density
0.683 0.000* 0.401 0.085*
Wetland 0.275 0.039* 0.293 0.077*
Forest -0.272 0.181
Model Evaluation Statistic
AIC 729.73 734.65
BIC 753.00 761.80
Loglik -358.86 -360.32
* Represents statistically significant at 10 percent level of significance

35. TP and land use matrix
Agricultural land, urban land-low density, and wetland
have significant positive impact on TP concentration
based on both watershed scale and HSA level
analysis
Contradictory wetland characteristic was may be due
to surpass of phosphorus storing capacity leading to
release of phosphorus into stream
Ardon et al.(2009) also found that wetland produced
greater soluble reactive phosphorus and total
phosphorus compared to agricultural land in Timber
land Lake restoration project in North Carolina
Forest shows a negative contribution towards
phosphorus concentration in the stream based on
HSA level analysis

36. TP and land use matrix
Pratt and Chang (2012) and Wan et al.(2014)
observed that agricultural land and urban land showed
primary contributor of phosphorus to stream around
Metropolitan area in Oregon and Xitiaoxi River
watershed in China, respectively
Tu(2011) found that forest is significantly decreasing
phosphorus concentration in the stream around
Boston Metropolitan area in Eastern Massachusetts

37. TSS and land use matrix
Watershed-Scale HSA-Scale
Predictors β‐value p-value β‐value p-value
Intercept 1.302 0.000 1.303 0.000
Agricultural land -0.248 0.108
Forest -0.505 0.009* -0.254 0.020*
Urban land- high
medium density
-0.446 0.036* -0.161 0.132
Model Evaluation Statistic
AIC 657.55 658.42
BIC 678.61 675.97
Loglik -322.77 -324.21
* Represents statistically significant at 10 percent level of significance

38. Forest and urban land- high medium density are
negatively correlated to sediment concentration in the
stream on both watershed scale and HSA level
Agricultural land is negatively correlated to sediment
concentration in the stream, however, it is insignificant
as the p-value is greater than 0.1
Negative correlation to all land use matrices suggests
that sediment concentration in the stream mostly due
to instream processes rather than overland processes
Qiu and Wang (2014) also found that more than 60
percent of the sediment load to stream was from
stream bank erosion and streambed sediment in
Neshanic River Watershed, New Jersey
TSS and land use matrix

39. Conclusions

40. Conclusions
Agricultural land and urban land-low density are
primary contributors to TN and TP concentration in the
stream
Forest reduced significant amount of sediment
compared to nutrients concentration
Increasing TN and TP concentration in stream by
wetland was may be due to release of nutrients by
wetland after accumulating for longer period
None of the land use contributes positively to
sediment concentration in the stream which suggests
that sediment concentration in the stream depends on
instream process rather than overland process

41. Conclusions
If sediment control is the objective, afforestation
should be recommended whereas If nutrient control is
the objective, targeting agricultural land and urban
land-low density would be performed
HSA level analysis showed close connection between
land use and water quality with more meaningful
information such as impact of urban and forest land
use

42. Future Work
Develop a threshold of topographic index in defining
HSA in landscape that leads to ecosystem
degradation using a Bayesian hierarchical model
Determine the relationship of land use to stream
integrity based on HSA level and watershed scale
analysis

43. Acknowledgement
This study was supported by Agriculture and Food
Research Initiative Competitive Grant no. NJW-2011-
03976 from the USDA National Institute of Food and
Agriculture

44. Thank you!
Question?

45. Watershed Delineation

46. 2007- land use
Urban land-high density: single or multiple unit on
1/8 to 1/5th of acre
Urban land-medium density: residential unit > 1/8th
1/2 acre
Urban land-low density: residential unit > 1/2 acre
to 1 acre
Rural residential: residential unit on 1 to 2 acre

47. Linear Mixed Model Assumption
Assumption:
 εij ~ N(0, δ2 ), δ2 = measures unexplained variation
 ~ N(0, ζ2), ζ2 = measures unexplained variation
due to watershed