land cover

Ground-Truthing Satellite Imagery with Phenological Observations: Visual Observations from Grasslands at the Sevilleta National Wildlife Refuge, New Mexico (2007-2008)

Abstract: 

Phenology is the study of recurring natural phenomena. The seasonal "greening-up" and "greening-down" of dominant vegetation can be used as a predictor for a variety of processes and variables at local to global scales. The use of satellites to monitor land surface phenology is important for understanding local and regional ecosystem variability, identifying change over time, and potentially predicting ecosystem response to short and long-term changes in climate. However, the relationship between how phenology is expressed on the ground and how it is interpreted from satellites is poorly understood because phenological stages do not always correspond well to changes in spectral reflectance. In this study, we explored the relationship between greenness as measured by digital camera, the human eye, and ASTER imagery in two perennial grasslands at the Sevilleta National Wildlife Refuge in central New Mexico.

Core Areas: 

Data set ID: 

214

Additional Project roles: 

176
177

Keywords: 

Data sources: 

sev214_remotesensing_06252009

Methods: 

Visual observations: While facing south, a "niner" (i.e., a 30 cm x 30 cm frame partioned into nine squares, each equal to an area of 1% of 1m2) was placed with one corner at a pinflag to the north and the other at a pinflag to the south. Four substrates were measured: bare Soil (S), brown vegetation (B), green vegetation (G), and green forbs (F).

The total cover of substrates at a sampling location equalled 9% (i.e., the entire niner). If the substrate had < 0.25% cover, T (i.e., trace) was recorded. If the cover of a substrate was > 0.25% it was rounded up to 1. If a substrate did not exist at a location, a zero was recorded

Time of day: The date/time stamp on a digital camera was used so that date/time was recorded on each photo.

Plot layout: Two sites were chosen to represent semi-arid grasslands. Site B was dominated by blue grama and site G by black grama. At each site six plots were established. Each plot was comprised of twelve quadrats that measured 50 X 50 cm for digital photographs and 30 X 30 cm for visual observations in a 3 X 4 grid.

Digital images: A 50 X 50 cm pvc frame was used to delineate area. Laminated markers placed next to the frame denoted the plot and quadrat number. A camera was held directly over a quadrat to get the entire PVC frame in the picture. Every effort was made to minimize shadow.

Maintenance: 

All data was checked for errors. Both years were combined and put online. 6/25/09, KRW.

Additional information: 

Data Collection Period:

5/3/2007, 7/2/2007, 7/16/2007, 7/31/2007, 8/17/2007, 9/4/2007, 9/18/2007, 10/4/2007, 4/14/2008, 5/30/2008, 8/4/2008, 8/19/2008, 9/5/2008, 9/22/2008, and 10/6/2008.

Multi-temporal TM-NDVI Vectors at Rodent Webs from the Sevilleta National Wildlife Refuge, New Mexico (1984-1993)

Abstract: 

This database contains mean NDVI values for 200 m diameter circles encompassing Rodent Webs on the Sevilleta National Wildlife Refuge (NWR), for 21 Landsat TM scene dating from 1984 to 1993. These NDVI vectors were generated as part of cooperative project between the Sevilleta LTER and the Indian Health Service, to study the 1992 Hantavirus outbreak.

Core Areas: 

Data set ID: 

107

Additional Project roles: 

21
22

Keywords: 

Data sources: 

sev107_rodenttm-ndvi_08122011.txt

Methods: 

The following Landsat Thematic Mapper (TM) scenes (from the Sevilleta Information Management System (SIMS) image archive) were rectified to UTM (NAD-27) coordinate system, radiometrically calibrated and converted to reflectance, then transformed into Normalized Difference Vegetation Index (NDVI) images. Characteristics of the imagery were: 28.5 m cell resolution; UTM NAD-27 coordinate system; bounded by ULx = 303973 Easting, ULy = 3819519 Northing, LRx = 346752 Easting, LRy = 3761122 Northing.

84jun22 87sep10 88sep28 89mar07 89may19 89oct10 90jan14 90may06 90sep1191apr23 91sep30 92apr09 92jun04 92jul06 92jul14 92aug15 92oct02 93may3093sep03 93sep19 93oct05

Rodent Web coordinates were extracted from the SIMS GPS Master Archive database, converted into an Arc/Info point coverage, then buffered to create circular polygons of 200 m diameter centered on the Web center stake. Then Arc/Info base and GRID routines were used to overlay the Web polygon coverage on each NDVI image, and mean NDVI values for each 200 m polygon were calculated; this generates an NDVI vector at each of the Web locations. The exact procedure log is included in the Additional Information below.


Maintenance: 

06/10/97 - date this file created. More to be added soon. G. Shore.

06/11/97 - reformatted file. G. Shore.

07/02/97 - added more documentation. G. Shore.

07/04/97 - added more documentation. G. Shore.

09/19/97 - dataset code corrected from SEV103 to SEV107. G. MacKeigan.

Additional information: 

Additional Study Area Information

The Sevilleta National Wildlife Refuge (NWR) is located in Socorro County, New Mexico, in the United States of America. The Sevilleta LTER was initiated as the Sevilleta National Wildlife Refuge, a former Spanish land grant now administered by the U.S. Fish and Wildlife Service. The LTER recently has been expanded to a research area of approximately 3,600 km2 that ranges from Rio Grande riparian forests ('bosque') and Chihuahuan Desert up to subalpine forests and meadows. Four dedicated research areas comprise the core sites; Sevilleta National Wildlife Refuge (100,000 ha), Bosque del Apache National Wildlife Refuge (25,300 ha), Sierra Ladrones Wilderness Study Area (28,390 ha) and the Magdalena Mountains Research Area (15,000 ha). The research region spans the Rio Grande basin with elevations ranging from 1,350 m at the Rio Grande to 2,195 m in the Los Pinos Mountains in the east, to 2,797 m at Ladrone Peak in the northwest, and to 3,450 m in the Magdalena Mountains to the southwest.

Climate is characterized by an intriguing combination of abundant sunshine, low humidity and high variability in most factors. The site exists in the boundary between several major air mass zones which contributes to the dynamics of the local climate. Precipitation ranges from <100 mm to 600 mm with an average of 280 mm. Summer precipitation occurs as intense thunderstorms often accounting for over 1/2 of the annual moisture. El Nino and La Nina events influence winter precipitation and marked variations occur on an inter-annual basis. Mean monthly temperatures range from -2.5 C to 27 C. Topography, geology, soils, and hydrology, interacting with major air mass dynamics, provide a spatial and temporal template that has resulted in the region being a transition zone for a number of biomes.

The region contains communities representative of, and at the intersection of, Great Plains Grassland, Great Basin Shrub-steppe, Chihuahuan Desert, Interior Chaparral, and Montane Coniferous Forest. The elevational gradient of the Magdalena Mountains provides further transitions for Interior Chaparral, Pinyon-Juniper Woodland, Petran Montane Conifer Forest, Petran Subalpine Conifer Forest, and Subalpine Grassland. The regional location at the junction of a number of biomes is critical for quantifying (1) gradient relationships with distance, (2) the scale-dependent or independent nature of spatial variability, (3) how steep gradients influence system properties, (4) integrated responses across the region, and (5) biome responses to climate change. Many species of these communities are at their distributional limits. For example, 54 plant species terminate their distributions within the Sevilleta and some represent major life forms and physiologies, such as the C3 perennial grasses. Reptiles provide a dramatic example as 47 of the 58 species end their distributions in the vicinity of the Sevilleta (33 are northern limits of desert species). An important feature of the biodiversity of this region is the number of examples of sympatric swarms of closely related species. This sympatry affords opportunities for studying the evolutionary differential of species.

Problems You Should Know About

A) ***NOTE***: that images were examined for Cloud and Cloud-shadow problems and the following Scene_Dates/Rodent_Webs were determined to partially or wholly impacted by Clouds and Cloud-shadows (and thus probably unsuitable for analyses purposes): tm91sep30: FPL-1, FPL-3, FPG-1, FPG-3 thru FPG-5, SC-2 thru SC-5, BX-1-1, BX-1-3, BX-1-4, BX-2-2, BX-2-3, BX-3-3, BX-3-4, BX-4-1 thru BX-4-3; tm92jul14: 222-1 thru 222-5

B) ***NOTE: an NDVI value of -9999.0 indicates MISSING Satellite data for that TM scene date for that Web location

C) Information regarding attempted error correction of NDVI vectors:

From gshore@sevilleta.unm.edu Wed Jul 2 16:40:05 1997
Date: Wed, 2 Jul 1997 00:20:08 -0600 (MDT)
From: Greg Shore
To: "Robert R. Parmenter"
Cc: Greg Shore
Subject: Sev/IHS Hanta Project

Hi Bob,

Well, I've tried several different approaches to developing correction factors for the NDVI values at the Web locations on the Sev, but have been foiled everytime. I am convinced that we'll just have to go back to the original bands and perform atmospheric correction on them utilizing the NASA correction algorithms Eric Vermote is developing. Unfortunately, that's probably a month or two away. In the meantime, I'm not sure what to think of the Landsat 4 imagery (6/4/92 and 7/6/92 scenes) since they seem to show unusually high NDVI values in relation to the Landsat 5 scenes just before and just after those dates (4/9/92 and 7/14/92). I know we received a lot of rain in May 1992, so some of this could be real but not sure how much. In addition, I've noticed that NDVI values generally seem higher starting in 1992, then before; and this could be due to a change in radiometric calibration procedures that EOSAT implemented in 1992. This points out the difficulty with doing multi-temporal analyses (looking for absolute changes at a location) with TM imagery. The next generation sensors (e.g., MODIS) will eliminate much of these variations and unknowns but that doesn't help us now.

Anyway, I've also taken a look at each image for cloud and cloud shadow effects, and have the following list of Webs and dates that should be excluded:

tm91sep30: FPL-1, FPL-3,
FPG-1, FPG-3 thru FPG-5,
SC-2 thru SC-5,
BX-1-1, BX-1-3, BX-1-4,
BX-2-2, BX-2-3,
BX-3-3, BX-3-4,
BX-4-1 thru BX-4-3

tm92jul14: 222-1 thru 222-5

Gregory A. Shore
GIS Analyst Programmer
gshore@sevilleta.unm.edu
http://sevilleta.unm.edu/~gshore
phone: (505) 277-2109 begin_of_the_skype_highlighting     

Additional Metadata- Command Log

List of processing commands for IHS/SevLTER Hanta Virus Project

NOTE that indicated Unix shell scripts, "C" programs, Arc/Info
AML's, etc. are located in /net/sevilleta/export/db/work/ihs_hanta_proj/
on sevilleta.unm.edu.

# BEGIN LOG

cd /net/sevilleta/export/db/work/ihs_hanta_proj/arcinfo/

#
# Extract coordinates and attribute data for Sev Web Sites from GPS file
#
arc2rdb /db/work/geodesy/sev_CAcode_gps_pts.dbf \
| select 'Project_Area == "Sevilleta National Wildlife Refuge"
&& Study_Name != "Neotoma Sympatry" && Plot_Type == "Rodent Web"
&& Study_Name != "Hantavirus Enclosure"' \
| project Study_Name Site_Name Block_Num Plot_Num NAD27utm_easting
NAD27utm_northing \
| listtotable \
> web_coors.rdb

sed -f make_plotcode.sed web_coors.rdb > web_coors2.rdb

project Study_Name Site_Name Block_Num Plot_Num NAD27utm_easting
NAD27utm_northing plotcode \
< web_coors2.rdb \
| compute 'if (Block_Num == "na") plotcode = Site_Name "-" Plot_Num;
else plotcode = Site_Name "-" Block_Num "-" Plot_Num' \
| project NAD27utm_easting NAD27utm_northing plotcode Study_Name \
| tail +3 \
| sort +3 -4 +2 -3 \
| nl \
| expand \
| awk '{print} END {print "END"}' \
> web_coors3.gen

grep -v 'END' web_coors3.gen \
| awk '{printf ",,,,\n",$1,$4, $5, $2, $3}' \
> web_coors3.info

#
# Generate the point coverage of Sev Web Sites, and add attribute data
#
Arc: precision double double
Arc: generate sev_webs_pt
Generate: input web_coors3.gen
Generate: points
Generate: q

Arc: projectdefine COVER sev_webs_pt
Project: PROJECTION UTM
Project: ZONE 13
Project: UNITS METERS
Project: DATUM NAD27
Project: PARAMETERS

Arc: build SEV_WEBS_PT points

Arc: info
ENTER COMMAND >DEFINE SEV_WEBS.DAT
ITEM NAME>SEV_WEBS_PT-ID,4,5,B
ITEM NAME>PLOT,6,7,C
ITEM NAME>STUDY,25,26,C
ITEM NAME>EASTING_NAD27UTM,8,12,F,3
ITEM NAME>NORTHNG_NAD27UTM,8,13,F,3
ITEM NAME>

ENTER COMMAND >ADD FROM ../web_coors3.info
57 RECORD(S) ADDED

Arc: joinitem SEV_WEBS_PT.PAT SEV_WEBS.DAT SEV_WEBS_PT.PAT SEV_WEBS_PT-ID ~
SEV_WEBS_PT-ID

Arc: additem SEV_WEBS_PT.pat SEV_WEBS_PT.pat id 4 5 b # SEV_WEBS_PT-ID

Arc: info
ENTER COMMAND >SELECT SEV_WEBS_PT.PAT
ENTER COMMAND >CALC ID = SEV_WEBS_PT-ID

#
# Buffer the Sev Web center stakes to create 200m diameter circular polygons,
# and add attribute data
#
Arc: buffer SEV_WEBS_PT SEV_WEBS_POLY # # 100 0.01 point

Arc: additem SEV_WEBS_POLY.pat SEV_WEBS_POLY.pat id 4 5 b # SEV_WEBS_POLY-ID

Arc: info
ENTER COMMAND >SELECT SEV_WEBS_POLY.PAT
ENTER COMMAND >CALC ID = SEV_WEBS_POLY-ID

Arc: joinitem SEV_WEBS_POLY.PAT SEV_WEBS_PT.PAT SEV_WEBS_POLY.PAT ID INSIDE
Arc: dropitem SEV_WEBS_POLY.PAT SEV_WEBS_POLY.PAT SEV_WEBS_PT#
Arc: dropitem SEV_WEBS_POLY.PAT SEV_WEBS_POLY.PAT SEV_WEBS_PT-ID

#
# Uncompress the 21 TM-NDVI Imagine scenes, convert them to Arc/Info GRID's
# (and fix them up for processing), then calculate mean of NDVI's within
# 200m circular zone around each Web.
#
# NOTE: creates GRID's called TMyymmmddNDVI
#
# Uncompress images
foreach f ( /db/work/vegmap_proj/ndvi_images/*.Z )
set fn = `basename $f | sed 's/\.Z//'`
uncompress -c $f > $fn
end

# Convert to GRID's
image2grid.csh

# Fix map projection in header
projectdefine.csh

# Create multi-layer GRID stack
/bin/ls -d *ndvi | wc -l > tm_stacklst.txt
/bin/ls -d *ndvi >> tm_stacklst.txt
Grid: makestack tm_ndvi file tm_stacklst.txt

# Convert -1.1 values to NODATA values
fix_nodata.csh

# Copy Webs point coverage template to working coverage
Arc: copy SEV_WEBS_PT sevwebspt200m

# Create Web Zonal GRID from 200m diameter Webs polygon coverage
Grid: setcell TM84JUN22NDVI
Grid: setwindow TM84JUN22NDVI TM84JUN22NDVI
Grid: webzonegrid = polygrid(SEV_WEBS_POLY, ID)

# Finally, Calculate NDVI means of each Web Buffer Zone
# NOTE: mean NDVI values, by scene date, stored in attribute table
# of point coverage SevWebsPt200m.
calcmeans.csh webzonegrid sevwebspt200m

# END COMMAND LOG

Subscribe to RSS - land cover