A moving window analysis for exploring landscape and geologic controls on spatial patterning of streambank groundwater discharge

Jackson, Kevin1 ; Briggs, Martin2 ; Helton, Ashley3

Published Feb 29, 2024 on Dryad. https://doi.org/10.5061/dryad.xsj3tx9mn

Data files

Feb 29, 2024 version files 340.95 MB

Jackson2023.zip

340.92 MB
README.md

27.50 KB

Abstract

Preferential groundwater discharge to streams and rivers is important for a range of hydrologic and ecological functions, from providing baseflow and thermal refuge to delivering excess nutrients and other contaminants. Improving our understanding of the spatial distribution and longitudinal extent of these groundwater expressions necessitates updated mapping and modeling techniques. Here we provide the dataset and code that enables the research community to replicate our moving window analysis which was used to analyze riverscape and landscape attributes along 5th order study river sections of the Farmington and Housatonic Rivers (a 26km river section near Tariffville, CT; and a 12km river section near Ashley Falls, MA, USA, respectively).

README

This zip directory contains the data, R code, python code, and outputs associated with our moving windows analysis reported in the associated paper (Article DOI: 10.1002/hyp.15112).

Contact: Ashley Helton (ashley.helthon@uconn.edu) or Kevin Jackson (kevin.jackson@umces.edu).

This dataset enables the replication of a moving window analysis used to calculate riverscape and landscape geomorphometrics, land use land cover characteristics, and subsurface characteristics for two river study sections along the Farmington and Housatonic Rivers.
This README file was generated on 2024-02-22 by Kevin Jackson.

#########################################################################

GENERAL INFORMATION

Title of Dataset: A Moving Window Analysis for Exploring Landscape and Geologic Controls on Spatial Patterning of Streambank Groundwater Discharge

Author Information
A. Principal Investigator Contact Information
Name: Kevin Jackson
Institution: University of Maryland Center for Environmental Science
Address: Frostburg, MD USA
Email: kevin.jackson@umces.edu

B. Associate or Co-investigator Contact Information
   Name: Martin Briggs
     Institution: United States Geological Survey
     Address: Storrs, CT USA
     Email: mbriggs@usgs.gov

C. Associate or Co-investigator Contact Information
   Name: Ashley Helton
     Institution: University of Connecticut
     Address: Storrs, CT USA
     Email: ashley.helton@uconn.edu

#########################################################################

Description of the data and file structure
The zip file provided for download has a pre-established file structure that enables you to run the scripts contained in ‘scripts/’ sub directory. The purpose of this analysis is to calculate landscape and geologic controls along 50m increments of our Farmington River and Housatonic River study sections. Landscape and riverscape characteristics as well as subsurface geologic variables are calculated along 50m river increments (e.g., ‘FarmSN_pnt.shp’, ‘Hous_pnt.shp’). Scripts are numbered showing you the order of operations. The zipfile excludes unnecessary temporary files and folders created when you run the scripts yourself. Everything included in the zipfile includes both (1) inputs necessary for the successful operation of each script, and (2) outputs most relevant for those interested in the final product of this work (i.e., ‘FarmSN_metrics.csv’, ‘Hous_metrics.csv’). The following lines of text show the folder structure and .csv, .img, .xml, .shp, and .txt files contained within ‘Jackson2023.zip.’

#########################################################################

The descriptions of the scripts you need to run are listed sequentially and are as follows:
0_NLCD_rewrite.py - This script rewrites nlcd_2019_pcs.img to produce impervious lulc and forested lulc rasters for later use in 4_BnksdSlp_RprnCvr_ImprvsCvr.py. Run-time is ~13 seconds.
1_WndwSz.py - This script converts our pre-drawn river reach centerlines (Hous_Cntrln.shp, FarmSN_Cntrln.shp) and creates point (FarmSN_pnt.shp, Hous_pnt.shp) and polygon data (FarmSN_25.shp, FarmSN_250.shp, FarmSN_500.shp, Hous_25.shp, Hous_250.shp, Hous_500.shp) along 50m increments, the latter of which is used as bounding extents for a majority of the subsequent calculations of metrics. Run-time is ~40 seconds.
2_DschrgDnsty.py - This script computes discharge density for each 50m river increment which can be used to categorize river increments based on lateral extent of mapped groundwater. It uses bounding extents produced from WndwSz.py and calculates the percentage of both right-hand and left-hand banksides that exhibited groundwater during our surveys. We consider discharge density in terms of all observed groundwater (Dnsty_v1); seeps >=10m (Dnsty_v2); and seeps <10m of discharge (Dnsty_v3). This script relies on FarmSN_PDPs.shp and Hous_PDPs.shp which differentiate seeps by type (<10m, >=10m); this script rewrites FarmSN_pnt.shp and Hous_pnt.shp with computed metrics and creates the final .csv outputs (FarmSN_metrics.csv, Hous_metrics.csv). Run-time is ~480 seconds.
3_Snsty_Crvtr.py - This script computes profile convexity, otherwise referred to as local curvature (Cnvxty50), and sinuosity (Snsty500). Derivation of sinuosity requires FarmSN_pnt_ext.shp & Hous_pnt_ext.shp. This script updates FarmSN_pnt.shp and Hous_pnt.shp with new metrics and rewrites the final .csv outputs (FarmSN_metrics.csv, Hous_metrics.csv). Run-time is ~120 seconds.
4_BnksdSlp_RprnCvr_ImprvsCvr.py - This script quantifies bankside slope (BnkSlp25), riparian forested LULC (RprnCvr25), and riparian impervious LULC (Imprvs250). A commonality in workflow shared by these metrics is that they exclude information within the river channel and assess zonal information within the intersection of a river buffer and bounding window extent. This script uses FarmSN_Poly_reclass.img and Hous_Poly_reclass.img. This script can be computationally intensive when creating FarmSN_Poly_reclass.img and Hous_Poly_reclass.img (lines 48-52); so, we include the files in case you’d like to bypass this step in the script. This script uses NLCD .img files and LiDAR DEMs (FarmSN_1m.img, Hous_1m.img) as well as updates FarmSN_pnt.shp and Hous_pnt.shp with new metrics and rewrites the final .csv outputs (FarmSN_metrics.csv, Hous_metrics.csv). Run-time is ~10540 seconds.
5_Soils.py - This script computes soil saturated hydraulic conductivity (KSAT250), soil erodibility factor (KFACT250), percent soil sand (SAND250), and water table depth (WTDEP250). This script uses soils_poly_fnl.shp and updates FarmSN_pnt.shp and Hous_pnt.shp with new metrics and rewrites the final .csv outputs (FarmSN_metrics.csv, Hous_metrics.csv). Run-time is ~1790 seconds.
6_Min_Elev.py - This script resamples 1m DEMs (FarmSN_1m.img, Hous_1m.img) to 5m DEMs that are used to find the minimum elevation along each river increment. This results in Hous_Min_Elev.csv and FarmSN_Min_Elev.csv. Run-time is ~1400 seconds.
7_Min_Elev.R - This R script takes elevational data collected from Min_Elev.py for both the Housatonic and Farmington River study sections and uses a fitted linear model (using geom_smooth() and the ‘lm’ method) to assign minimum elevations for each river increment. This information is used when delineating the floodplain extent and valley bottom. This script uses Hous_Min_Elev.csv and FarmSN_Min_Elev.csv and creates Hous_Min_Elev_Est.csv and FarmSN_Min_Elev_Est.csv.
8_VBE_Cnfmnt_FPE.py - This script computes floodplain extent (FPE500v3), valley bottom extent (VBE500v3), and confinement (Cnf500v4). 1m LiDAR drive DEMs are resampled to 30m resolution and reclassified at every river increment into two binary rasters, (1) one identifying elevations within the floodplain (within 3m of minimum elevation) and (2) another identifying the valley bottom (elevations within 10m of the minimum elevation). Minimum elevations are based on the minimum elevations modeled in 7_Min_Elev.R. Floodplain extent is computed as the percent of reclassified floodplain within a 500mx1000m window size. Similarly, Valley Bottom Extent is computed as the percent of reclassified valley bottom within a given 500mx1000m extent. Confinement counts the proportion of bankside records (FarmSN_PDPs.shp and Hous_PDPs.shp) that are within 30m of the reclassified floodplain terrace. This script uses FarmSN_resample_cubic_30m_m.img, Hous_resample_cubic_30m.img, FarmSN_PDPs.shp and Hous_PDPs.shp; and this script updates FarmSN_pnt.shp and Hous_pnt.shp with new metrics and rewrites the final .csv outputs (FarmSN_metrics.csv, Hous_metrics.csv). Run-time is ~7230 seconds.

#########################################################################

Descriptions of directories within Jackson2023.zip:
dem/ - This directory contains .img rasters used in various scripts.
spatial_analysis/ - This directory contains shapefiles created and rewritten by scripts in this analysis (FarmSN_pnt.shp, FarmSN_25.shp) as well as shapefiles provided by publicly accessible sources (soils_poly_fnl.shp). This directory also stores outputs in the form of .csv files as well as Analysis and Pre-Analysis scripts.
surveys/ - this directory contains shapefiles associated with the delineation of the river study sections and the absence and presence of groundwater-surface interactions from our principal survey.
metadata.zip - this zip file contains metadata files associated with downloads from CT ECO, MASSGIS, SSURGO, and NLCD.

#########################################################################

Descriptions of .img files included are as follows (filenames prefaced with 'FarmSN' have an equivalent file associated with the Housatonic which are prefaced with 'Hous' unless otherwise specified):
nlcd_2019_pcs.img - 2019 NLCD LULC raster used for the creation of riparian forested and riparian impervious rasters (see 0_NLCD_rewrite.py)
FarmSN_1m.img - 1m resolution LiDAR derived DEM originating from CT ECO. The analogous 1m resolution DEM for the Housatonic river, Hous_1m.img, was derived from MASSGIS.
FarmSN_resample_cubic_30m_m.img - a resampled DEM at 30m resolution.
FarmSN_Poly_reclass.img - a raster converted from FarmSN_Poly.shp set to null and used to temporarily set, within river, cells for the 1m DEM to null to avoid miscalculating bankside slope (BnkSlp25).

Hous_1m_fill.img - 1m resolution LiDAR derived DEM for the Housatonic river, Hous_1m.img, was derived from MASSGIS, and was subsequently filled using the spatial analyst tool, Fill,' for use in '4_BnksdSlp_RprnCvr_ImprvsCvr.py'.

#########################################################################

Descriptions of .csv files included are as follows (filenames prefaced with 'FarmSN' have an equivalent file associated with the Housatonic which are prefaced with 'Hous'):
FarmSN_metrics.csv - Final output which is a table to table conversion of FarmSN_pnt.shp. This .csv file contains all computed metrics for the river study reach.
FarmSN_Min_Elev.csv - This is the output of 6_Min_Elev.py which is a list of minimum elevation values which are linked to each unique river increment.
FarmSN_Min_Elev_Est.csv - This is the output of 7_Min_Elev.R which models/smooths the minimum elevation values from the previously recorded observations (i.e., FarmSN_Min_Elev.csv).

#########################################################################

Descriptions of .shp files included are as follows (filenames prefaced with 'FarmSN' have an equivalent file associated with the Housatonic which are prefaced with 'Hous'):
soils_poly_fnl.shp - Shapefile of Soil characteristics derived from Soil Survey Geographic (SSURGO) database shapefiles joined with tabular data of area- and depth-weighted averages of SSURGO variables using NRCS map unit keys (see Briggs et al., 2022).
FarmSN_25.shp - polygon shapefile of 50mx200m rectangles which serve as bounding extents through which zonal statistics are computed for bankside slope (BnkSlp25), riparian forested LULC (RprnCvr25), and discharge density (Dnsty_v1, Dnsty_v2, Dnsty_v3).\
FarmSN_250.shp - polygon shapefile of 500mx1000m rectangles which serve as bounding extents through which zonal statistics are computed for soils & subsurface characteristics (KSAT250, KFACT250, SAND250, WTDEP250), and riparian impervious lulc (Imprvs250).
FarmSN_500.shp - polygon shapefile of 1000mx500m rectangles which serve as bounding extents through which zonal statistics are computed for floodplain extent (FPE500v3), valley bottom extent (VBE500v3), and confinement (Cnf500v4).
FarmSN_pnt.shp - shapefile of point data for every 50m of river. This file is derived from armSN_Cntrln.shp.
FarmSN_Cntrln.shp - polyline shapefile of river centerline hand drawn based on LiDAR derived DEM and hillshade rasters.
FarmSN_Cntrln_ext.shp - extended centerline to enable calculation of sinuosity for river increments near the upstream and downstream terminus if river study.
FarmSN_Poly.shp - Polygon shapefile of the river study section.
FarmSN_PDPs.shp - See Moore et al. (2023) for description of this shapefile of point data representing every 1m of bankside (river right and river left). Points in this dataset identify bankside gw discharge and whether it’s associated with a laterally extensive seep face (e.g., is the m increment a part of a seepage face > or < 10m).
FarmSN_pnt_ext.shp - extended shapefile of additional river increments (as points) to enable calculation of sinuosity for river increments near the upstream and downstream terminus if river study.

#########################################################################

Sharing/Access information
Briggs et al., (2022) (https://doi.org/10.3390/w14010011) describes the derivation of Soils Characteristics along the Farmington River study reach. This dataset extends this analysis to include our Housatonic River study reach.
Moore et al., (2023) (https://doi.org/10.5066/P915E8JY) includes a collection of our thermal infrared images used to characterize groundwater surface water interactions along the two study sections. In addition to field notes used to produce our groundwater survey shapefiles (FARMSN_PDPs.shp, Hous_PDPs.shp).
Jackson et al., (2023) (https://doi.org/10.5066/P9FTZ0DK) contains HVSR measurements along the Farmington River study section. These are not used in this particular dataset; however, these measurements are combined for use and analysis in the manuscript associated with this dataset.

#########################################################################

Data used in this analysis was downloaded from publicly available sources in from the following sources including:
1m high resolution DEMs were derived from LiDAR and can be downloaded from (1) CT ECO (see Capitol Region Council of Governments, 2016; https://maps.cteco.uconn.edu/data/lidar/), and (2) MASSGIS (see MassGIS (Bureau of Geographic Information), Commonwealth of Massachusetts EOTSS; https://www.mass.gov/info-details/massgis-data-lidar-dem-and-shaded-relief);
The soils maps used to calculate soils characteristics were taken from SSURGO (seeSoil Survey Staff 2020; https://gdg.sc.egov.usda.gov);
Land use land cover characteristics were downloaded from NLCD (see Dewitz, and U.S. Geological Survey, 2021; https://www.usgs.gov/centers/eros/science/national-land-cover-database)

#########################################################################

Code/Software
Code is in the sub directory, ‘scripts/’ and named to indicate their order of operation. As stated in Usage Notes, this analysis was run using ArcMAP (v.10.8.1; ESRI) using external python scripts ran on Python 2.7.18 Shell. Additional preprocessing was run using R version 4.0.3 (2020-10-10). After downloading, make sure that the correct file paths are being used when running the scripts.

#########################################################################

References
Briggs M.A., Jackson K.E., Liu F., Moore E.M., Bisson A., Helton A.M., 2022, Exploring Local Riverbank Sediment Controls on the Occurrence of Preferential Groundwater Discharge Points. 14(1):11. https://doi.org/10.3390/w14010011
Capitol Region Council of Governments. (2016). 2016 Aerial imagery. Retrieved from http://cteco.uconn.edu/data/flight2016/index.htm.
Jackson, K.E., Haynes, A.B., and Briggs, M.A., 2023, Passive seismic depth to bedrock data collected along streams of the Farmington River watershed, CT, USA: U.S. Geological Survey data release, https://doi.org/10.5066/P9FTZ0DK.
MassGIS (Bureau of Geographic Information), Commonwealth of Massachusetts EOTSS.
Moore, E.M., Jackson, K.E., Haynes, A.B., Harvey, M., Helton, A.M., and Briggs, M.A., 2023, Thermal infrared images of groundwater discharge zones in the Farmington and Housatonic River watersheds (Connecticut and Massachusetts, 2019)(ver. 3.0, January 2023): U.S. Geological Survey data release, https://doi.org/10.5066/P915E8JY.
Dewitz, J., and U.S. Geological Survey, 2021, National Land Cover Database (NLCD) 2019 Products (ver. 2.0, June 2021): U.S. Geological Survey data release, https://doi.org/10.5066/P9KZCM54.
Soil Survey Staff 2020, accessed 4 October 2021, https://gdg.sc.egov.usda.gov

#########################################################################

DATA-SPECIFIC INFORMATION FOR: soils_poly_fnl.shp

Variable List:
* MUKEY: NRCS map unit keys, or ‘MUKEY’ identifications, that link soils characteristics to SSURGO soils.
* AVG_KFACT: soil erodibility factor (AVG_KFACT)
* AVG_KSAT: saturated hydraulic conductivity (μm/s; AVG_KSAT)
* AVG_SAND: percent sand (%; SAND)
* WTDEP_MIN: minimum water table depth (cm; WTDEP_MIN)

#########################################################################

DATA-SPECIFIC INFORMATION FOR: FarmSN_PDPs.shp

Variable List:
* active: Binary indicator of presence of active groundwater discharge regardless of lateral extent, where '1' is a positive indicator and '0' indicates a bank location where no active discharge was detected
* active_2: Binary indicator of active groundwater discharge greater than 10m in lateral extent and considered 'discharge faces', where '1' is a postive indicator
* active_3: Binary indicator of active groundwater discharge less than 10m in lateral extent and considered preferential discharge points, where '1' is a positive indicator

#########################################################################

DATA-SPECIFIC INFORMATION FOR: Hous_PDPs.shp

#########################################################################

DATA-SPECIFIC INFORMATION FOR: FarmSN_Min_Elev.csv

Variable List:
* Min_Elev: minimum elevation (m) calculated from 6_Min_Elev.py.

#########################################################################

DATA-SPECIFIC INFORMATION FOR: FarmSN_Min_Elev_Est.csv

Variable List:
* actual: minimum elevation (m) calculated from 6_Min_Elev.py.
* predicted: modeled/smoothed minimum elevation (m) using 7_Min_Elev.R and 'actual' measurements.

#########################################################################

DATA-SPECIFIC INFORMATION FOR: Hous_Min_Elev.csv

Variable List:
* Min_Elev: minimum elevation (m) calculated from 6_Min_Elev.py.

#########################################################################

DATA-SPECIFIC INFORMATION FOR: Hous_Min_Elev_Est.csv

Variable List:
* actual: minimum elevation (m) calculated from 6_Min_Elev.py.
* predicted: modeled/smoothed minimum elevation (m) using 7_Min_Elev.R and 'actual' measurements.

#########################################################################

DATA-SPECIFIC INFORMATION FOR: FarmSN_metrics.csv

Variable List:
* Dnsty_v1: the fraction (%) of surveyed bankside (river right and river left) exhibiting gw discharge.
* Dnsty_v2: the fraction (%) of surveyed bankside (river right and river left) exhibiting gw discharge >= 10m.
* Dnsty_v3: the fraction (%) of surveyed bankside (river right and river left) exhibiting gw discharge < 10m.
* Cnvxty50: Profile convexity, or local curvature, was calculated by fitting an arc to cartesian coordinates of the center point of each 50m river increment and its upstream and downstream neighboring 50m increments. Convexity is the ratio (unitless) of the distance between (1) the midpoint of the fitted arc and the midpoint of the straight line between center points and (2) the straight-line distance between center points.
* Snsty500: Sinuosity was calculated as the ratio (unitless) between the total stream length within a 1000m x 500m window and the shortest Euclidean distance between the furthest upstream and downstream 50 m center points contained within the window.
* BnkSlp25: Mean bankside slope (percent rise, unitless) was derived from 1m resolution DEMs. Slope was measured within a 10m buffer from the streambank for each 50m stream increment.
* RprnCvr25: Percent riparian forest was calculated as the fraction (%) of forested LULC within the intersection of a 50 x 200m window and a 60 m buffer of the river channel.
* Imprvs250: Percent riparian developed was calculated as the fraction (%) of developed LULC that occurs with the intersection of a 500 x 1000 m window and a 600 m buffer of the river channel.
* KSAT250: soil saturated hydraulic conductivity (μm/s)derived from the average values within a 500 x 1000m window extent.
* KSAT250p: percent of window extent with KSAT values available to average.
* KFACT250: soil erodibility factor derived from the average values within a 500 x 1000m window extent.
* KFACT250p: percent of window extent with KFACT values available to average.
* SAND250: percent soil sand (%) derived from the average values within a 500 x 1000m window extent.
* SAND250p: percent of window extent with SAND values available to average.
* WTDEP250: water table depth (cm) derived from the average values within a 500 x 1000m window extent.
* WTDEP250p: percent of window extent with WTDEP values available to average.
* VBE500v3: Valley bottom extent was calculated as the fraction (%) of valley floor within a 1000m x 500m window. We classified a 30m cell as valley floor if elevations were within 10m of the minimum 50m river increment elevation
* Cnf500v4: Confinement was calculated as the fraction (%) streambank length on either bank within a 1000m x 500m window that is within 30m of the floodplain boundary.
* FPE500v3: Floodplain extent was calculated as the fraction (%) of the floodplain within a 1000m x 500m window. We classified a 30m cell as floodplain if elevations were within 3m of the minimum 50m river increment elevation.

#########################################################################

DATA-SPECIFIC INFORMATION FOR: Hous_metrics.csv

A moving window analysis for exploring landscape and geologic controls on spatial patterning of streambank groundwater discharge

Data files

Abstract

README

Methods

Usage notes