Fusion of remote sensing dataset with heterogeneous spatio-temporal resolution: simulation of sentinel-2 time series of vegetation indexes for agricultural monitoring (Abstract/Poster in convegno)

Type
Label
  • Fusion of remote sensing dataset with heterogeneous spatio-temporal resolution: simulation of sentinel-2 time series of vegetation indexes for agricultural monitoring (Abstract/Poster in convegno) (literal)
Anno
  • 2014-01-01T00:00:00+01:00 (literal)
Alternative label
  • G. Candiani G. Bordogna G. Manfron M. Boschetti M. Pepe (2014)
    Fusion of remote sensing dataset with heterogeneous spatio-temporal resolution: simulation of sentinel-2 time series of vegetation indexes for agricultural monitoring
    in SENTINEL-2 for Science Workshop, ESA - ESRIN, Frascati (IT), 20-22 May 2014
    (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#autori
  • G. Candiani G. Bordogna G. Manfron M. Boschetti M. Pepe (literal)
Note
  • Poster (literal)
Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#affiliazioni
  • Consiglio Nazionale delle Ricerche - IREA, via Bassini 15, 20133 Milano Consiglio Nazionale delle Ricerche - IREA, via Bassini 15, 20133 Milano Consiglio Nazionale delle Ricerche - IREA, via Bassini 15, 20133 Milano Consiglio Nazionale delle Ricerche - IREA, via Bassini 15, 20133 Milano Consiglio Nazionale delle Ricerche - IREA, via Bassini 15, 20133 Milano (literal)
Titolo
  • Fusion of remote sensing dataset with heterogeneous spatio-temporal resolution: simulation of sentinel-2 time series of vegetation indexes for agricultural monitoring (literal)
Abstract
  • Agricultural monitoring needs updated information on both type and dynamic of cultivated varieties. Usually, the agricultural environment is fragmented into fields with heterogeneous vegetation dynamics due to differences in cultivated varieties, agro-practices and weather influence. Thus, a suitable monitoring system, exploiting remote sensing capabilities, requires data featuring both high spatial resolution and high revisiting time. Currently, only heterogeneous data - in terms of spatial and temporal resolutions - are available for operational monitoring purposes. Typically, high spatial resolution (HR) data (< 30 m) feature low revisiting time (16 to 26 days) while, daily data are available at low or very low spatial resolution (LR) (250-1000 m). Till now, most of the analyses for operational land cover mapping have been performed through multi-temporal approaches using high spatial resolution satellite data. The Sentinel-2 mission will offer the chance for the coexistence of high spatial and temporal resolution for the first time. Theoretically, this key feature will introduce the opportunity for agricultural monitoring to move from multi-temporal approaches to time series analysis. However, cloud obstruction represents a major constraint reducing the potential revisiting time, and actually decreasing the number of scenes per season exploitable for land cover mapping purposes. In this perspective, in order to understand in advance the potentiality of Sentinel-2 data, it is important to find a method suitable to assimilate data from sensors with high spatial resolution and sensors with high revisiting time, fusing them into a new time series dataset featuring both high spatial and time resolutions. The SPOT4 Take5 experiment - proposed by CNES - gives the opportunity to simulate Sentinel-2 data for a short period (from February to June 2013), providing the best benchmark to evaluate the improvements offered by the coexistence of high spatial and temporal resolution. In this study, we propose an approach which perform the fusion of the available heterogeneous satellite data to generate new time series characterized by better spatio-temporal information. More specifically, the SPOT4 Take5 (S4T5) level 2 dataset, acquired on the Provence area in France, together with a MODIS dataset acquired on the same area and in the same period, have been divided into training and testing dataset. The training dataset has been used to perform several tests of the proposed fusion procedure with different settings, to assess the sensitivity of the method to several parameters. Basically, for a given timestamp t, a vegetation index (VI) image will be simulated using the available dataset from both S4T5 and MODIS with different timestamps, weighted by a function of their temporal validity with respect to t. The time series resulting from different tests have been validated through the testing dataset not used in the fusion process. Finally, the most accurate simulated areas of the VI time series have been tested in the agricultural area of Provence for the classification of wheat cultivated area and time occurrence of crops phenology. This contribution describes i) the procedure for the generation of VI time series through the fusion of information coming from heterogeneous satellite data, ii) the tests conducted with different settings for the proposed method and iii) an example showing the utility of the method for operational agronomic monitoring. (literal)
Prodotto di
Autore CNR
Insieme di parole chiave

Incoming links:


Autore CNR di
Prodotto
Insieme di parole chiave di
data.CNR.it