Crop identification using harmonic analysis of time-series AVHRR NDVI data
Introduction
Identification and mapping of crop types using coarse-resolution satellite imagery is critical to many applications, from simple estimation of cultivated areas (Loveland et al., 1995, Loveland et al., 1991) to stratification for crop yield models (Lee, 1999, Kastens et al., 1998, Doraiswamy and Cook, 1995) to a critical component of mesoscale storm prediction (Gutman and Ignatov, 1998, Gutman et al., 1989) and hydrologic models (Yin and Williams, 1997). Characterization and mapping of crop types using the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High-Resolution Radiometer (AVHRR) 1.1 km resolution data has typically taken one of two approaches: temporal profiles of crop phenology as manifested in the normalized difference vegetation index (NDVI) (DeFries et al., 1995, Reed et al., 1994, Lloyd, 1990); and classification of multitemporal data using maximum likelihood or other classification algorithms (Loveland et al., 1995, Brown et al., 1993, Loveland et al., 1991). To date, few researchers in remote sensing have applied time-series analysis techniques (harmonic or Fourier analysis) developed for other disciplines, such as electrical engineering, hydrology, or climatology (Legates and Willmott, 1990, Davis, 1986), to the NDVI time series. This paper describes the application of harmonic analysis to a single year (1992) of NOAA-AVHRR data (26 periods) for identification of several common crop types occurring within the western Great Plains. The products of the harmonic analysis (additive, amplitude, and phase terms) are used within a discriminant analysis for mapping crop types within Finney County in southwestern Kansas.
Briefly defined, harmonic (Fourier) analysis permits a complex curve to be expressed as the sum of a series of cosine waves (terms) and an additive term (Davis, 1986, Rayner, 1971). Each wave is defined by a unique amplitude and a phase angle, where the amplitude value is half the height of a wave, and the phase angle (or simply, phase) defines the offset between the origin and the peak of the wave over the range 0–2π (Fig. 1a). Each term designates the number of complete cycles completed by a wave over the defined interval (e.g., the second term completes two cycles, Fig. 1b). Successive harmonic terms are added to produce a complex curve (Fig. 1c), and each component curve, or term, accounts for a percentage of the total variance in the original time-series data set.
Section snippets
Study area
Finney County, the second-largest county in Kansas at approximately 1300 mile2, is located in southwestern Kansas in the High Plains at approximately 100°W longitude. Cropland dominates the relatively flat landscape of Finney County, comprising over 76% of the county. The Arkansas River extends east–west through the central portion of the county (Fig. 2), and irrigated agriculture (predominantly corn, milo, and alfalfa) is clustered along the river, drawing on the Ogalalla Aquifer to supply
Harmonic characteristics of crop types
Irrigated corn and alfalfa in Finney County possess a single distinct growing season, attaining peak greenness during midsummer. This is manifested in the harmonic analysis by a strongly unimodal periodic pattern, with a high amplitude value in the first term and low amplitude values in successive terms (Fig. 4a). The majority of the total variance in seasonal NDVI for corn is contained in the first harmonic term. Alfalfa, although it also exhibits a strong summer-peak greenness pattern (Fig. 4
Acknowledgements
This project was conducted at the Kansas Applied Remote Sensing (KARS) Program (Edward A. Martinko, Director). The research described in this paper was funded by the National Institute for Global Environmental Change (NIGEC) (South Central Regional Center, Tulane University, David Sailor, Director), through the US Department of Energy (Cooperative Agreement No. DE-FC03- 90ER61010). Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors
References (23)
- et al.
Global discrimination of land cover types from metrics derived from AVHRR Pathfinder data
Remote Sensing of Environment
(1995) - et al.
Community type differentiation using NOAA/AVHRR data within a sagebrush-steppe ecosystem
Remote Sensing of Environment
(1993) Red and photographic infrared linear combinations for monitoring vegetation
Remote Sensing of Environment
(1979)- et al.
Mapping sub-pixel proportional land cover with AVHRR imagery
International Journal of Remote Sensing
(1997) - et al.
Accuracy of the AVHRR vegetation index as a predictor of biomass, primary productivity and net CO2 flux
Vegetation
(1989) - et al.
Using multi-source data in global land cover characterization: concepts, requirements and methods
Photogrammetric Engineering and Remote Sensing
(1993) Statistics and data analysis in geology
(1986)- et al.
Spring wheat yield assessment using NOAA AVHRR data
Canadian Journal of Remote Sensing
(1995) - Egbert, S.E., Price, K.P., Nellis, M.D., Lee, R. 1995. Developing a land cover modeling protocol for the High Plains...
The 1990 conterminous US AVHRR data set
Photogrammetric Engineering and Remote Sensing
(1992)
Derivation of green vegetation fraction from NOAA-AVHRR for use in numerical weather prediction models
International Journal of Remote Sensing
Cited by (243)
Improving potato above ground biomass estimation combining hyperspectral data and harmonic decomposition techniques
2024, Computers and Electronics in AgricultureWetland mapping in East Asia by two-stage object-based Random Forest and hierarchical decision tree algorithms on Sentinel-1/2 images
2023, Remote Sensing of EnvironmentMangrove zonation mapping in West Africa, at 10-m resolution, optimized for inter-annual monitoring
2023, Ecological Informatics