Estimating Water Use Through Satellite Remote Sensing
Type: Technical Report
Date Published: December 2010
Authors:
Aldo R. Piñón-Villarreal
Zohrab A. Samani
A. Salim Bawazir
Max P. Bleiweiss
Rhonda Skaggs
Vien T. Tran
Abstract:
The Mesilla Valley, New Mexico is a major producer of pecans in the southwestern United States. Irrigation or pecans constitutes the largest consumer of surface water from the Rio Grande. In order to develop an effective water budget for the region, improved estimates of regional evapotranspiration (ET) are needed. The Regional ET Estimation Model (REEM) using satellite imagery from ASTER and Landsat-7 sensors, and ground-measured meteorological data were used to determine ET of pecan orchards in the Valley for the year 2002. Geographic information systems (GIS) vector files, each representing a single pecan orchard, were superimposed over the REEM output maps to obtain estimates of orchard area and ET. The total areas for the orchards greater than 4 ha (10 acres) and less than 4 ha (10 acres) were determined as 7,969 ha (19,691 acres) and 1,851 ha (4,573 acres), respectively. The weighted average annual ET of the larger orchards was determined as 991 mm using ASTER data and 1,018 mm using Landsat-7 data compared to 800 mm and 852 mm for the smaller orchards using ASTER and Landsat-7 data respectively. The spatial and temporal variability in ET was analyzed among 280 sampled orchards greater than 4 ha (10 acres). Standard deviation of ET using ASTER and Landsat-7 data ranged from 3.45 to 28.40 and 1.70 to 27.13 mm/month in December and in May, respectively. A method to estimate ET in pecan orchards was developed by relating the fractional cover (fc), the normalized vegetation index (NDVI) and REEM annual ET for 280 pecan orchards. Orchard fc was estimated using supervised classification of aerial photography (color infrared-Digital Ortho-photo Quarter-Quadrangles or CIRDOQQs). This method compared well with field measurements of fc carried out in thirteen pecan orchards from the Valley (mean percentage error = 14%). Linear regression of the fc determined by supervised classification and the NDVI resulted in a coefficient of determination (R2) of 0.71 (n = 170). This relationship was then used to relate fc to the REEM annual ET via NDVI for 280 pecan orchards for the Mesilla Valley.
Download the full report in PDF format:
Attachment | Size |
---|---|
tr357.pdf | 3.02 MB |