< Back to previous page

Project

Multi-temporal, multi-source and multi-scale remote sensing for the monitoring of Conference pear tree phenology

The general objective of the present research is to describe and monitor Conference pear tree phenology, by means of air- and spaceborne multispectral and hyperspectral imagery. Specific focus will be on a multi-temporal approach. The influence of pedo-climatic and agronomic factors on phenology variability will be investigated. To achieve this objective, the following steps have been set: 1. Characterizing Conference pear tree phenology by means of multi-temporal hyperspectral imagery, collected via ground and airborne measurements. Images acquired with remote sensing measurements are not results in themselves, but rather data that need to be analysed and interpreted. To this end, it is important to investigate in detail the characteristics of the object of the image, the processes that take place and their influence on the information contained in the image. This is the approach that will be used to achieve the objective: RPAS data will be acquired and analysed; at the same time, spectroradiometer measurements will be made to help in the interpretation of the RPAS data. The PCFruit experimental fields will be surveyed with an automated RPAS operated by the Dronegrid company. The hyperspectral camera of the UAV will gather reflectance in the RGB, NIR and Red Edge regions; measurements will be acquired with a daily temporal resolution during the four key developmental stages as defined by the BBCH code. Concurrently, regular spectroradiometer measurements will investigate the changes in leaf reflectance that occur in time. In this way, data will be collected about the reflectance at the leaf level and at the orchard level. Measurements at the leaf level can help in interpreting measurements at the orchard level, but they need to be compared and linked. The gap between the two classes of data will be filled by using Plant Radiative Transfer Models, which have been extensively used to simulate the behaviour of solar radiation when it enters a leaf or a canopy. This step will be performed several times: first, it will be repeated in time, on each of the four phenological stages surveyed with the RPAS; secondly, it will be repeated in space, to test the effect of different fertilization and growth regulators regimes. 2. Characterizing Conference pear tree phenology with Sentinel-2 multitemporal multispectral data, at the field and country level. By using satellite-derived imagery it is possible to describe the entire crop development, by calculating its temporal profile (e.g. Vina et al., 2004). This research will employ ESA Sentinel-2 data: Sentinel-2A was launched in June 2015 and Sentinel-2B was launched in March 2017 (ESA, 2000). The potential of the sensors carried by the satellites, the MSI, has yet to be fully explored. Its 10m spatial resolution can open new doors in the study of orchards: in these contexts, where vegetation never covers the soil completely, spatial resolution can make a difference. Up to now, there have been attempts to classify tree species with S2 data (Immitzer, Vuolo, & Atzberger, 2016), but no one has tested the MSI sensor for the detection of phenological stages in orchards. However, the potential of Sentinel-2 lies not only in its spatial resolution, but also in its temporal resolution: its 5-day revisit time is unprecedented and represents a new opportunity for multitemporal studies. 3. Developing a country scale monitoring tool for Conference pear phenology. The information provided by Sentinel-2 data gives the possibility of building a monitoring tool to be used at the country scale. The phenological data calculated with S2 imagery will for sure have a certain degree of variability: but up to what point a temporal profile can be considered “normal” and when does it represent an anomaly? There is a need to interpret, by explaining, the phenological variability found at the country level. In general, phenology is determined primarily by pedo-climatic conditions; in addition to this, agronomic practices may have an influence on certain phenological stages; finally, a condition of biotic or abiotic stress may alter the normal growing cycle of the crop. Thus, there are mainly three components in the phenological variability quantified with satellite imagery. The amount of variability due to pedo-climatic conditions can be accounted for by employing an agro-ecological zonation, i.e. the division of a territory into homogeneous pedo-climatic areas. The satellite-derived phenological data will be studied separately for each zone, dividing orchards into groups that have a homogeneous pedo-climatic profile. Following, the remaining variability will be studied to estimate the effect of biotic or abiotic stress on phenological changes. The research will be conducted thanks to the support of two bodies: the Centre for Remote Sensing and Earth Observation Processes of VITO; the Division of Forest, Nature and Landscape, belonging to the Department of Department of Earth and Environmental Sciences (E&ES), of the KU Leuven.

Date:19 Oct 2017 →  24 Sep 2019
Keywords:pear, phenology, remote sensing, satellite, drone, senescence
Disciplines:Landscape architecture, Art studies and sciences, Physical geography and environmental geoscience, Communications technology, Geomatic engineering, Forestry sciences, Ecology, Environmental science and management, Other environmental sciences
Project type:PhD project