Multi-time scale analysis of sugarcane within-field variability: improved crop diagnosis using satellite time series? |
| |
Authors: | Agnès Bégué Pierre Todoroff Johanna Pater |
| |
Institution: | (1) CIRAD, UMR TETIS, Montpellier, 34093, France;(2) CIRAD, UR SCA, Petit Bourg, Guadeloupe, 97170, France |
| |
Abstract: | Within-field spatial variability is related to multiple factors that can be time-independent or time-dependent. In this study,
our working hypothesis is that a multi-time scale analysis of the dynamics of spatial patterns can help establish a diagnosis
of crop condition. To test this hypothesis, we analyzed the within-field variability of a sugarcane crop at seasonal and annual
time scales, and tried to link this variability to environmental (climate, topography, and soil depth) and cropping (harvest
date) factors. The analysis was based on a sugarcane field vegetation index (NDVI) time series of fifteen SPOT images acquired
in the French West Indies (Guadeloupe) in 2002 and 2003, and on an original classification method that enabled us to focus
on crop spatial variability independently of crop growth stages. We showed that at the seasonal scale, the within-field growth
pattern depended on the phenological stage of the crop and on cropping operations. At the annual scale, NDVI maps revealed
a stable pattern for the two consecutive years at peak vegetation, despite very different rainfall amounts, but with inverse
NDVI values. This inversion is linked with the topography and consequently to the plant water status. We conclude that (1)
it is necessary to know the crop growing cycle to correctly interpret the spatial pattern, (2) single-date images may be insufficient
for the diagnosis of crop condition or for prediction, and (3) the pattern of vigour occurrence within fields can help diagnose
growth anomalies.
|
| |
Keywords: | Sugarcane Remote sensing Diagnosis Satellite time series Spatio-temporal variability Topography NDVI Growth anomaly |
本文献已被 SpringerLink 等数据库收录! |
|