| |
Background
Over the past 50 years surface water rights have been transferred off
of a number of land areas in the Middle Rio Grande Region (See Appendix
B for a definition of terms, including “water right”). In
some cases the rights have been purchased by municipalities who instead
draw the
equivalent amount of groundwater. What is not known is how the “land
sans surface water rights” is being used and whether it is still
being irrigated, for instance, using MRGCD water.
Objective
Evaluate the “greenness” of areas that have sold off surface
water rights.
Methods
Study Area
The study area is the Middle Rio Grande Region of New Mexico, comprised
of Bernalillo, Sandoval, and Valencia counties.
Assumptions
- Parcels from which water rights were sold (WTP) were agricultural
plots. NDVI is not an effective measure of greenness in forested or “bosque” area
because satellite data used does not detect understory growth (Cleverly,
2004).
- Areas that are not being irrigated will show a lower “greenness” than
ones that are still being irrigated.
- The water transfer and shapefile data provided by the Office of
the State Engineer is accurate. (Actually, there are known problems
with the data. See below.)
Analysis
- Projection – I used UTM 13N only because that was the projection
that the OSE data came in. For this project, the projection is not
so important since I’m not particularly concerned with either
shape, direction, distance, or area.
- Software Environment – Most of the work, except conversion
of the OSE shapefiles to coverages, was done in ArcMap. I used raster
data since that is the native format of satellite imagery.
- Landsat7 data -- Using the satellite image taken
on July 18, 2002, a “leaf on” condition, I converted bands
3 and 4 to NDVI (See Appendix B). Individual rasters get a value between –1 and
1. I multiplied the values by 100 (a range of –100 to 100) since
ArcMap can interpret integers as shades of gray or colors (Figure
G). According to the NOAA
website, values less than –10 are “not very green” and
values greater than 60 are “very green.” I assigned colors
to the map as follows:
| Range |
Description |
| < -10 |
Not green (shades of gray) |
| -10 thru 18 |
Slightly green (shades of brown) |
| 18 thru 61 |
Green (shades of light green) |
| > 61 |
Very green (dark green) |
- Note that the Landsat7 data used does not cover the entire area of
the 3 counties and does not need to, since all water rights transfers
have taken place in the “green” corridor along the Rio
Grande. Thus, the effective study area is the Rio Grande corridor in
these three counties.
- OSE data -- The water transfer files from OSE contain many columns
of data as well as shape data. Of interest for this study were:
| Database Column |
Description |
| File |
An identifier, usually a well number |
| Permit |
OSE permit number |
| Transferred |
1 = full transfer
2 = partial transfer |
| Date_Trans |
Date of the transfer |
| Lease_Back |
If filled in, date water was leased back to property owner |
I ran a zonal function to derive the mean greenness of each parcel. Then
I created a buffer zone 200m in diameter around each parcel and calculated
the mean greenness of the buffer. I then took the differences of the “means” of
each parcel and its buffer to determine how much the greenness varied between
the parcel and the surrounding area.
- Counties and Roads -- I added layers outlining the three counties
and major roads in the counties to help the viewer see the location
of parcels.
Results
- The “half-decade” for 2001-2005 actually only contains
data for a year or year and a half, since the satellite readings took
place in July 2002. All 2003 transfer data has been excluded from analysis,
as well as data from the second half of 2002 (Figure
C & Figure
E).
- About a third of the transfers in the files do not have a transfer
date associated with them (Figure
A).
- Just under half of the transfers are transfers of part of the surface
water rights belonging to the “partial transfer” parcels
(Figure A & Figure B). In these
cases I would expect to see less of a change in greenness compared
to the
surrounding parcels. In fact, the
average change in greenness across all “full transfers” is
.3 percent and the change for “partial transfers” is .6
percent. Both numbers are so small and so similar as to be statistically
insignificant.
- The overall
difference in greenness between the parcels and the surrounding areas
is very slight, in the 1%-3% range, except for a
couple outlier transfers with leaseback (Figure
D). The greatest AVERAGE
variance for a half-decade appears in the transfers with leaseback;
however, in both cases this is a situation where there are only one
or two transfers contributing to the mean (Figure
C & Figure
E).
Conclusions
What does this mean? One interpretation is that the data support the
contention that the amount of water being used on the parcels has not
changed perceptibly since the surface water rights were sold off.
This may, in fact be a correct interpretation of the data. However, there are
enough questions about the accuracy of the water transfer data, that additional
data collection and verification are warranted. For example, in Figure
F the
parcel identified as File 4432, Permit RG-0620 is in the Valencia county shapefile
which is known to have alignment problems. After moving the parcel shape to
where it looks like it belongs, the greenness variance is more than 3 times
higher (5.9% rather than 1.6%). Corresponding corrections to the Valencia shape
file would likely show better greenness variances than is currently the case.
This is just one example of inaccuracies in the OSE data.
Future Work
I see a number of directions for further work.
- Increase data accuracy -- First, obtain accurate water transfer
data from the OSE by going through the paper permits. Update the shapefiles
accordingly, especially for Valencia county.
- Increase data accuracy – Get accurate information about the
size of the parcels, size of water right, and size of right transferred.
Use this information to make a weighted greenness analysis of the partial
transfers.
- Evaluate greenness over time -- Do a comparison of the 2002 satellite
data with earlier files to see how greenness has changed over time
in the parcels of interest.
- Improve the interpretive framework of NDVI -- Obtain actual water
and plant coverage data from a sample of parcels and “calibrate” the
NDVI readings against this.
References
Normalized Difference Vegetation Index. CGIS at Towson University.
http://chesapeake.towson.edu/data/ndvi2.asp (Accessed April 23, 2004).
What is NDVI? NOAA Coastal Services Center. http://www.csc.noaa.gov/crs/definitions/NDVI.html
(Accessed April 23, 2004; Last Updated April 27, 1998).
A conversation with James Cleverly on April 21, 2004 regarding differences
in the accuracy of NDVI as a measure of water use in forested and agricultural
areas. Albuquerque, New Mexico.
Dataset References
Landsat7 images for July 18, 2002, for portions of Bernalillo, Sandoval,
and Valencia counties. Of interest for this project is the portion of
each county along the Rio Grande where water rights transfers have taken
place.
Source: Obtained from EDAC, UNM.
Shapefiles of Water Transfer Data 1950’s thru 2003 for Bernalillo,
Valencia, and Sandoval counties.
Source: NM OSE.
Notes: Per OSE, many of the data columns in the shapefile are inaccurate. The
columns used in this analysis are thought to be about 70% accurate. The Valencia
county data, however, has additional problems. The polygons are not aligned
correctly with the other layers. OSE intends to update this, but the updated
shapefile wasn’t yet available at the time of this project.
Shapefile of major roads in the 3-county area.
Source: ESRI ArcGIS 3.2. Data & Maps, Western United States. Copyright 1996, 1998, 1999, ESRI.
Shapefile outline of Bernalillo, Sandoval, and Valencia counties
Source: ESRI ArcGIS 8; US Map Template Series.
Appendices
Appendix A - Steps to Create This GIS
Appendix B - Pertinent Terms
|
|
|
