As in the BELAIR 2013 campaign, the main aim of the 2015 campaign was to gather remote sensing data to improve the monitoring, steering  and managing of capital intensive perennial fruit orchards.

1. Focus on Optimizing Fruit Orchard Management

Executive summary

The potential yield is rarely attained in crop production systems due to physiological limitations and impacts of biotic and abiotic stress factors. A sub-optimal yield implies a vast economic loss, especially in capital-intensive multi-annual crop systems. The development of efficient management strategies is therefore endeavored, which demands an accurate monitoring of the impact of the damaging factors. This is however not straightforward since the symptom distribution following a given stress varies according to foliage exposure, species and foliage physiological characteristics. Another issue lies in identifying the causal agent of stress symptoms since different kinds of stress inductors (e.g. ozone, heavy metal pollution and drought) cause many similar biochemical changes. The temporal dimension is therefore often indispensable to distinguish stresses (e.g., varying climatological conditions, phenological phase). Proper understanding of plant physiological processes in relation to the environment is as such essential to enable early detection and more time and site-specific treatment against production limiting factors, with the goal of optimizing yield. Due to the intensification of agriculture and the response of plants to environmental variables that vary on fine scales, specialized techniques are needed that retain the benefits of large-scale monitoring, but recognizes local variation, i.e. site-specific stresses. Non-intrusive techniques are therefore essential for capturing data in a continuous manner, thereby enabling rapid response and minimized unintentional impacts. Moreover, these techniques have to allow a presymptomatic monitoring of changes in the plant physiological state since by the time that the first visible symptoms of stress can be observed, a plant can already be adversely affected. Spectral reflectance, chlorophyll fluorescence, and thermography are currently the most advanced techniques to detect early signs of stress.

However, in spectral remote sensors, usually a trade-off exists between SNR, spatial and spectral resolutions due to physical limitations, data-transfer requirements and some other practical reasons. High potential therefore lies in the fusion of multiple data sets from different sensors which combines complementary information of the same scene or identical information with complementary spatial domain or with a better temporal description.

Data acquisition

The same nutrient field trial of 2013 was again measured in 2015. The drought stress trial of 2013 was excluded from further investigation due to the difficulty to induce a severe drought stress on that location. This trial was replaced by a growth regulation trial on pears. Furthermore, due to the economic importance of strawberries, remote sensing and in situ data were acquired on an irrigation and fertigation strawberry trial as well.

Next to the measurements on the fruit trials, additional hyperspectral  measurements were performed on a large homogeneous wheat field. This dataset would later serve as reference spectral database for the validation of 100m Proba-V products.

  • Hyperspectral ASD measurements (reference + objects of interest)
  • Flir Thermal measurements of strawberry
  • Reference GPS measurements
  • Description of the health state of the fruit orchards
  • Physiological measurements
  • Chlorophyll fluorescence imaging
  • Fast fluorescence induction kinetics and quenching analysis
  • Pictures
  • RGB: RGB UAV data gathered by VITO
  • Multispec and Red edge UAV imagery gathered by VITO
  • Hyperspectral APEX images of Hesbania were acquired on July 1st 2015. Quicklooks are available on

2. Focus on Crop Monitoring from Parcel to Local and Regional Scale


  •  to acquire UAV’s observation needed in the the SIII (1) BELCAM (collaborative system for agricultural monitoring) and (2) iPot (industrial potato monitoring for the Belgian potato sector) projects (details below)
  • to validate PROBA-V 100 m reflectances
  • to continue BELAIR 2013 research related to the optimization of nitrogen fertilization (CRA-W 2015 potato trial) through a fine assessment of in-season crop nitrogen status (CNS) and to evaluate the potential use of optical devices to assess potato crop nitrogen status. This is achieved by the use of leaf/canopy optical properties such as crop canopy reflectance, leaf transmittance or leaf fluorescence. These plant-based methods have been developed since decades and can help in decision on the need for supplementary N fertilizer application.

Executive Summary

BELAIR 2015 has allowed obtaining the required observations to meet the research objectives of the projects listed below.

(1) Ipot project: development of geo-information products for both the Belgian potato processing industry and research centres on potato crop, as an answer to the industry’s specific needs regarding crop condition, growth monitoring and yield estimates and forecasts.

Although the products are satellite-based, detailed UAV images are used as an intermediate level between field and satellite and help to define the required characteristics of the input satellite images and serve as a reference for high resolution satellite data interpretation.

The UAV flights enable further studies that have begun in 2014 and that need to be followed in 2015 with UAV flights to answer the main questions. These studies aim to validate phenology products derived from HR satellite images with the objectives to : (1) determine when we can detect emergence with HR satellites (2) compare absolute fCover estimates, (3) compare fCover patterns (UAV vs. satellite), (4) to determine for crop senescence to which extent can satellites measure decrease in chlorophyll content. Phenology is not only provided as information product for iPot but serve also as input for yield modelling. The availability of UAV’s time series images is crucial around phenological periods (emergence/senescence) that are of prime importance to predict final potato yield and to determine the optimal date for harvest). There is a HUGE need for UAV’s data to calibrate/validate satellite data that is the basic source of information across the whole Belgian territory to manage the potato crop fields at the end of the project through the Geo-web-based platform.

(2) BELCAM project: design of collaborative information system built on the complementarity of local (professional crowdsourcing) and satellite remote sensing technologies for crop monitoring.

By fully exploiting the red-edge capabilities of Sentinel 2 and the wide and frequent coverage of Sentinel 1 and 2 as well of PROBA-V, different information will be delivered along the season to the farmers: 1) the provisional annual Nitrogen balance-sheet, 2) the field zoning, 3) the crop Nitrogen status before the 3rd application for winter wheat and before a potential second application for potato at the parcel level. In addition, the overall crop status along the season, the major water stress/pest/diseases damages and the yield estimate will be provided at the district level.  Existing methods will be applied to Sentinel 2 for fAPAR and LAI, to Sentinel 1 for biomass while the crop status and intermediate crop mapping will rely on both, combined with Proba-V at 100 m.  New or adjusted methods will be developed to derive from the red-edge bands the leaf Chlorophyll “a” content and then to relate this to the Nitrogen status.  The most original part is to apply the retrieval algorithm for these biophysical variables over a large scale (Belgium) in order to identify the main sources of errors across various agricultural landscapes. Regarding the assessment of the crop status, and linked to the Hesbania APEX Flight 2015, hyperspectral UAV’s acquisition data during the same period over BELCAM potato fields located in the BELAIR zone presents a high value to correlated SPOT5Take5 and Rapid-Eye images acquisition with simultaneous ground-based measurements.

(3) PROBA-V 100 m reflectance validation (with VITO).

(4) CRAW 2015 potato trial located in the BELAIR area (Gembloux-Ernage) and crossing nitrogen rates and irrigation modalities are available with ground-based measurement (chlorophyllmeter, radiometer, fluorimeter) to validate satellite, APEX flight and UAV’s flight data.

Data acquisition

In 2015, one winter wheat field and three potato fields in Gembloux were monitored intensively in support of BELCAM and iPot projects.

Another potato field was monitored for the CRAW 2015 potato trial.


The phenological development stage was recorded (using the 2-decimal BBCH code) and hemispherical pictures were taken on by CRA-W to estimate fCover, fAPAR and LAI. These data were collected as close as possible to the UAV flight dates (24/04, 22/05, 5/06, 11/06, 24/06, 01/07, 07/07, 09/07, 10/07, 06/08, 20/08, 03/09, 04/09).

Ground data in field trials and at field scale were collected to retrieve the leaf chlorophyll content for both crops. One protocol of data acquisition was elaborated by crop.

The winter wheat trial, called ‘Les Pompes’, was composed of 40 varieties and 2 levels of Nitrogen (recommended application and additional 50 kg of Nitrogen) in two repetitions. The half of each trial plot (1.5m x 5m) received the additional 50 kg of Nitrogen. Five varieties (Edgar, Expert, Cellule, Sahara, Tobak) were selected for the study according to the following criteria: (1) representative variety of those used by the Belgian farmers, (2) Variability in the varieties in terms of precocity, resistance, structure having a potential influence on the chlorophyll content.

The data in the winter wheat trial was collected on the 20th of May.

The CRA-W 2015 potato trial was based on one variety (Charlotte), two levels of water (irrigated, non-irrigated) and four levels of fertilisation (0, 100, 150, 250 kg N/ha). In this study, we consider one repetition, 3 levels of Nitrogen (1, 100, 250kgN/ha) and the two levels of water. Nitrogen rates were applied at planting. Irrigation scheduling was conducted based on the soil water content which is monitored using soil tensiometer probes (Watermark probes, Irrometer Company, California).  The split plot design consisted of four replications of 12 rows of potato with a length of 12 m. The density of plantation was 0.26 cm*0.90 cm. Potato was planted on 18 April. The data, described in the following sections, were collected on the 26th of June and the 7th of July.

Ground-based Optical measurements were taken fot this potato trial :

Near sensing measurements:

The chlorophyll-meter Hydro N Tester (SPAD/HNT, Yara, Oslo, Norway) is a handheld leaf clip sensors (corresponding respectively to 2-3 mm2 of leaf area reading). The measurements were made on distal leaflet of the first fully developed leaf from the top of the canopy (corresponding to the 4th or 5th leaf from the apex of a main stem) avoiding midribs. The measurements concern the upper face of the leaf. 30 individual readings were collected for each plot on the 29 June. The Hydro N Tester provides a HNT index (related to leaf chlorophyll content). The fluorimeter Dualex (Force-A, Orsay, France) is a hand-held leaf clip sensors (corresponding respectively to leaf area of 2-3 mm2 and 19-20 mm2 per reading). As with the chlorophyll-meter, the measurements were made on distal leaflet of the first fully developed leaf from the top of the canopy. The measurements focused on the upper face of the leaf. 30 individual readings were collected for each plot on the 29 June. The Dualex provide a CHL index (related to leaf chlorophyll content), FLV index (related to leaf flavonoids content), and NBI index (related to both chlorophyll and flavonoids content).

Near remote sensing measurements:

The Cropscan (Cropscan Inc, Rochester, USA) is a passive near remote sensing radiometer measuring crop light reflectance. The radiometer is extended on a boom to a height of 2 m above the ground (±1.5 m above canopy) providing a circular field area of 1 m2.

The radiometer was faced downward perpendicular to the crop surface (nadir view). On each date, 5 measurements were made and averaged across the replications. The sensor allows the use of eight wavebands from 460 nm to 810 nm at 50 nm intervals. Irradiance and radiance were stored for each waveband allowing the calculation of canopy reflectance as the ratio between radiance and irradiance.

LAI assessment:

Hemispherical images of plant canopies were taken by a camera equipped with a fish-eye lens (Besel, Japan). The hemispherical images estimates the leaf area index (LAI).

Plant sample collection and analysis

After the optical measurements have been performed, plant samples were collected on the first of July. Within each plot, plants were collected on 1.04 m length and taken to the laboratory. The plants were washed, air-dried and weighed in order to determine separately aerial parts (leaves + stems), roots and tubers fresh weight. Subsamples of each plant part were then dried at 80° C up to constant dry weight. The dry matter (DM) content of each sample was then calculated. From the various plant parts (leaves + stems, tubers and roots), finely crushed samples with a Cyclotec 1093 sample mill (FOSS .Tecator) were subjected to analytical measurements of total N plant concentration conducted with NIR spectroscopy using a FOSS-NIR Systems 6500 scanning instrument (NIR-Systems, Silver Springs, MD) and calibrated using the Dumas Combustion method (LECO, St Joseph MI, USA).

At field scale, data were also collected in a winter wheat field of 4,6 ha in Gembloux where a variability study within the field is ongoing via the VISA project (CRA-W). The collection of data was done at two dates (13th of May, 3rd of June) in 3 homogeneous elementary surface units (ESU) of 20 meters by 20 meters to simulate a pixel of Sentinel 2. Inside an ESU 3 x 2 rows of winter wheat were flagged, the measures were systematically taken on these rows. For the potato, the data were collected on three individual plants well dispersed in 2 fields of 2 different varieties (Bintje and Nicolas).

The data collected in field trialds and at parcel level were divided in 2 parts: (i) data collected on the ground and (ii) data measured in the lab. Data collected on the ground include phenological observations (BBCH code) reflectance data (5 measures by plot; 8 wavelengths (from 460 to 810)), LAI (8 hemispherical pictures per plot), Chlorophyll index (SPAD and Dualex scientific Force A, 30 measures on the leaves) and sampling of the aerial part of the plants for measurements in the lab.

Data measured in laboratory include Biomass and water content, LAI (by destructive measurements), reflectance of the leaves (0 to 2500 nm with the ASD Field Spec 3), chlorophyll extraction  and Nitrogen content (Extraction with the Dumas method).


In 2015, the three potato fields (variety Bintje, Fontane and Nicola) in Gembloux mentioned above were monitored in support of the iPot project. In addition to RGB, also RedEdge and Multispec 4C images were acquired. Images are available for the following dates:

Emergence: 22/5, 11/6, 24/6, 1/7 ;

Senescence: 5/8, 21/8, 2/9.

A CRA-W winter wheat trial field and the neighbouring  was monitored as well with UAV. RedEdge and Multispec4C images were taken on 24/4, 11/5, 22/5, 11/6, 7/7.

The images were processed by VITO using Agisoft Photoscan (RGB and RedEdge) or PostFlight Terra software (Multispec4C). This includes generation of ortho-photographs and digital terrain models.

For the CRA-W 2015 potato trial, UAV data were acquired simultaneously to the ground-based measurements.


An APEX flight was carried out on 1st July 2015 in the Hesbania zone. The images were processed by VITO.


Time series of high resolution DMC/Deimos satellite images were used for crop growth monitoring. These images cover the entire Belgian territory at a spatial resolution of 22m. DMC/Deimos ortho-rectified images for the potato growing season (mid-May till mid-September) were ordered via BELSPO (EO helpdesk) with a frequency of 2 weeks. Images are available for the following dates in 2015: 4/6, 11/6, 14/6, 1/7, 11/7, 1/8, 7/8, 21/8, 31/8, 10/9.

RapidEye images were acquired in support of BELCAM, for field zoning and LAI estimation. The images were ordered via BELSPO’s EO helpdesk. Available dates: 13/5, 30/6, 17/7, 31/8, 11/9/2015.

The DMC/Deimos and RapidEye satellite images were pre-processed with VITO’s MORPHO chain. This includes atmospheric correction, cloud and shadow detection and calculation of biophysical parameters.


The in-situ measurements as well as the processing of the data was carried out by CRA-W, UCL and VITO.

UAV flights were carried out by VITO.

3. Soil


This project aims to acquire up-to-date and standardized soil property data through calibrations of the signal similar to the one from the future generation of satellites, without the need of field sampling campaigns.

Executive summary

Optical satellite sensor systems are now reaching the spectral resolution necessary to derive critical topsoil properties that have previously been demonstrated with laboratory, field and airborne sensor systems.

Forthcoming high spectral resolution satellites, such as the German EnMAP, have the greatest potential for high quality mapping of soil properties over extensive areas with a uniform methodology that hitherto was not possible.

A processing chain of hyperspectral satellite imagery is being derived based on a hyperspectral flight campaign on the HESBANIA flight strip. This procedure includes the following steps: i) building a soil spectral library compatible with the LUCAS European spectral library, ii) spectral transfer functions allowing a meaningful comparison between the spectral library, field, airborne and simulated satellite spectra, iii) algorithms for soil discrimination based on the spectral library, iv) spectral rules for the exclusion of pixels that do not show a soil signal (e.g. too much vegetation, moisture or roughness).

Data acquisition


During the APEX flight spectra of reference surfaces as well as bare cropland  soils are acquired with an ASD Fieldspec Pro and soil samples are taken.

The team carried out a survey to select as many bare cropland fields at the time of the overflight as possible. These fields are in seedbed condition mostly for carrots and beans. The coordinates were registered and photos were taken. As the flight was quite late in the season the number of bare fields that we detected was limited (n=12), although distributed throughout the flight lines. The fields are all in seed bed condition, free of residues and the soil surface is uniformly dried out. A sampling scheme was set up allowing us to collect around 50 samples randomly in the fields. The number of samples in each field was weighted against its area. 


Not applicable


The APEX campaign Hesbania was carried out on 1 July 2015 around 3pm under clear sky conditions.

Auxillary data pre-processing will be carried out in order to allow the image processing by VITO to be optimized for soil spectra.


Not applicable




VITO Remote Sensing - Senior R&D Professional