The European Mediterranean has been under cultivation since Neolithic times, and extensive grazing following transhumance strategies have ever since shaped Mediterranean rangelands. However, shifts in land tenure have in recent times increased anthropogenic pressure, often associated with stocking numbers exceeding the carrying capacity or the concentration of animals in piospheres. Annual time series of Landsat-derived vegetation fractional cover have been used to map long-term vegetation dynamics and past studies have concluded that a major portion of rangelands was showing symptoms of land degradation associated with declining vegetation cover or shrub encroachment, resulting from over- or under-utilization, respectively. However, these analyses were restricted to a limited amount of carefully selected images close to the peak of the season, only.
The opening of the Landsat archive in 2008 as well as the recovery of a vast amount of historic Landsat data into the USGS archive boosted data availability considerably. However, data captured between 1986 and 1999 over Europe was unavailable, as it could not be processed due to missing Payload Correction Data (no-PCD). In early 2018, USGS successfully unlocked no-PCD imagery, which together with advances in High Performance Computing, the emergence of Analysis Ready Data, and rapidly developing methodology marked a game changer for taking European land use analyses to the next level. We now have 1000s of observations available, which are processed automatically and allow for a completely data-driven analyses.
This presentation provides a comparison of past and current state-of-the art methodology that was/is used for observing long-term land use changes using Landsat time series. We revisited a case study analyzing long-term vegetation dynamics in the Mediterranean rangelands of Crete done in 2003 to compare the results derived from methodology and data available then and now. We employed linear trend analysis based on green vegetation fractions derived from linear spectral unmixing. We compared long-term trends using a small number of manually selected Landsat acquisitions vs employing Land Surface Phenology metrics derived from all available observations. We found large differences in the extent of rangelands subject to land degradation, as well in the strength of the trend. In general, overall spatial patterns of the trend direction are similar. However, the new assessment indicates that vegetation cover decline was less pronounced than previously assumed. In 2003, it was assessed that 41% of Cretan rangelands showed symptoms of land degradation, while our new findings only indicate 18%. The findings and methods are discussed with respect to current data availability, prospect of integrating Sentinel-2 data, as well as past limitations in data access that may have violated some assumptions that the initial study was based on. The main conclusion of this analysis is that studies relying on few images only should be read and interpreted with caution.

Crop growth stages (progress) in the United States are reported weekly at state or district level by the U.S. Department of Agriculture (USDA) National Agricultural Statistics Service (NASS) based on the field observations. However, the crop progress and condition ground data collection is time consuming and samples are sparsely distributed within the statistic unit. Furthermore, the NASS weekly report provides summaries at the district and state level but does not reflect the spatial variability within the unit. Large scale dense field date collection is operationally un-realistic. Satellite remote sensing has the potential to add value to NASS observations, providing spatially distributed assessments of crop condition in an objective and cost-effective way. Remote sensing data have been used to extract vegetation phenology for decades. Because remotely sensed phenologic metrics (e.g. green-up dates and dormancy dates) are different from crop physiological growth stages (e.g. planting, emergence and harvest dates), their connections need to be established for use the remote sensing derived phenology in crop modeling systems. Several remote sensing approaches have been developed to extract vegetation/crop phenology using vegetation index (VI) time-series. The basic steps of these approaches are: 1) to build a series of mathematical functions to fit the observations; and 2) to detect the transition dates in the functions by using either a predefined threshold of change or curvature points from the functions. The results are somewhat dependent on the availability of satellite data especially during rapid growth or harvest stages. Daily observations or multi-day composites from coarse resolution sensors such as MODIS and VIIRS can provide dynamics of crop growth and have been used to generate the global vegetation phenology data product. However, the spatial resolutions of these sensors are too coarse to capture spatial variability within the field and among fields that are smaller than a coarse resolution pixel. Landsat and Sentinel-2 spatial resolutions have been shown sufficient for field scale mapping in the U.S. but their data acquisitions are not as frequent as MODIS and VIIRS. In this presentation, we will review current crop phenology mapping approaches and demonstrate a data fusion method to build daily VI time-series at 30m spatial resolution using multiple remote sensing data from Landsat, Sentinel-2, MODIS and VIIRS. Crop phenology extracted from Landsat, Sentinel-2, MODIS/VIIRS will be compared to the results derived from the fused VI time-series. Added value by using the recent Harmonized Landsat and Sentinel-2 (HLS) data product will be examined. High spatial and temporal resolution VENUS (Vegetation and Environment monitoring on a New MicroSatellite) images collected over the experiment site will be used to assess the VI and crop growth stages from multi-sensor fused data. The challenges to map the near real-time crop growth stages at field scales will be further discussed.

The Joint Research Centre operationally runs the early warning decision support system ASAP (Anomaly hot Spot of Agricultural Production), that provides timely information of crop production anomalies based on remote sensing and rainfall data, with a focus on food insecure countries. ASAP’s framework consists of two levels of spatial and temporal aggregation, with the first level referring to the automated classification of meteorological and crop conditions at the sub-national administrative levels into four warning categories every ten days. The second level involves the monthly crop condition assessment at national level made by agricultural analysts, assisted by the automatic warnings as well as by ancillary information (European Media Monitor, high-resolution imagery, maps and graphs). Among the aforementioned ancillary information, crop calendars are a crucial input for the analysts performing the monthly assessments but also for the users of the system. A crop calendar lists crop types with associated typical growing periods. Different sources (e.g. FAO; USDA) make available crop calendars but only at national level. Relevant subnational information regarding the growing cycle of the crops can be retrieved from satellite data. Satellite-based Land Surface Phenology-LSPcan be derived from the analysis of NDVI time series to map the Start of the Season (SOS) and End of the Season (EOS).
The objective of this study is to establish a link between the main agricultural vegetation growth stages from LSP and the commonly used FAO national level crop calendars, in order to evaluate the consistency of the two datasets and to derive crop specific calendars at both national and sub-national level for the 80 ASAP countries.
The satellite-based SOS and EOS timings were extracted for crop areas in the ASAP units of interest and then plotted as density scatterplots of SOS vs. EOS. The timing is expressed in dekads, meaning 10-day periods covering the full calendar year with 36 dekads. The data points in the scatterplots tend to cluster in some regions of the SOS vs. EOS space. These clusters are interpreted as main crop seasons. In order to attribute a crop type listed in FAO crop calendar to a main crop season recognized on LSP, the compatibility between the time of occurrence of SOS and EOS, derived from LSP, with the period of sowing and harvesting derived from the calendars, was visually evaluated according to a set of simple rules. The resulting matches represent crop types in the original crop calendars that agree with local satellite-based phenology. Such hybrid calendars are made available in the on-line system of ASAP in both graphic and tabular format.
It was observed that in some cases a season identified in the LSP cannot be matched with a FAO crop calendar and vice versa. This can be due to the fact that LSP method can handle one or two growing seasons per solar year, so crops with more than two cycles in a year cannot befully described by LSP. Additionally, it cannot be excluded that FAO crop calendars may be outdated or incorrect.
The full automation of the proposed procedure is currently being developed. This will include the use of a clustering algorithmfor detecting the crop seasons in LSP, the use of agricultural statistical data available from FAOSTAT at national level for computing the probability for a crop type to be assigned to a cluster.To conclude, observing LSP at improved spatial resolution, provided to this novel application the necessary datain order to assign crop types to the crop seasons recognised in LSP, thus providing crop development information that depicts the variations of the phenology, especially in the subnational level.

Remote sensing of vegetation has been used extensively to study the spatio-temporal changes in phenology. Most of these studies, using leaf area index (LAI) or Fraction of Absorved Photosyntethic Active Radiation (FAPAR), have been limited by the spatial (in the order of 10 kilometers) and especially temporal resolutions (in the order of weeks). However, recent advancements in the land surface analysis products derived from geostationary satellites provide data at much higher resolutions. These data, therefore, have a large potential to improve our understanding on local-scale ecohydrological processes at continental to global scales.

In this study, we use the Fractional Vegetation Cover (FVC) data derived from SEVIRI sensor aboard the European geostationary satellites. SEVIRI is a multispectral optical sensor with 12 spectral channels with a spatial resolution of 3 km at sub-satellite point and a temporal resolution of 15 minutes for the whole Meteosat disk. The Meteosat disk is centered at 0° longitude covering whole Europe and Africa. The FVC is available from The Satellite Application Facility of EUMETSAT on Land Surface Analysis (LSA-SAF). Temporal resolution of the data is daily with the native spatial resolution of SEVIRI sensor.

In particular, we used the FVC data to investigate the spatio-temporal variations of the decay in phenology in dry seasons by characterising the decay of the data. The rate of decay in FVC is calculated for each pixel separately by fitting an exponential decay function to FVC time series for the dry seasons. Using an exponential decay function characterises the decay behaviour independent of the amplitude and duration of the decay; which makes a fair spatial analysis possible. Such characterization of the dry down allows us to identify the regions where the vegetation activity is supported by secondary (non-precipitation) sources of moisture. As mapped in the attachment, our preliminary analysis shows that the decay constant is larger in the dry regions than the temperate regions. Tropical regions are also showing high decay constants which is likely because of the low intensity of dry seasons but still to be investigated.

Vegetation is a highly dynamic land surface parameter and plays an important role in regional and global hydrological-, energy- and carbon cycles. Vegetation phenology, the timing of vegetation phases, is a sensitive indicator of terrestrial ecosystem response to climate change, and changes herein, e.g. lengthening of the growing season, can influence terrestrial carbon uptake and thus, depending on the net effect, either exacerbate or dampen global warming. Because of the crucial role of vegetation in the Earth’s system, there is an urgent need to study the response, resistance and resilience of vegetation to a warming climate. Remote sensing of phenological metrics, e.g. start and length of season can be used to assess the response of vegetation. Classically, studies use optical based satellite products, e.g. Normalized Difference Vegetation Index or Leaf Area Index, to monitor vegetation and land surface phenology. However, optical based remote sensing is often hindered by cloud cover, smoke. Microwave remote sensing provides a complementary source of data to monitor vegetation especially in crucial areas with extensive cloud cover, such as the tropics and high latitudes. Due to the longer wavelength of microwaves they penetrate the canopy deeper and are sensitive to the total water content in both the crown as well as the woody part of the vegetation. Often Vegetation Optical Depth (VOD) is retrieved from active and passive microwave observations using a radiative transfer model. The VOD parametrizes the attenuation, through absorption and scattering, of the microwave signal within the vegetation layer caused by the water in the vegetation. Studies have used VOD to monitor vegetation dynamics and land surface phenology and have found strong correspondence between VOD and optical based products (Jones et al. 2011, Liu et al., 2013, Vreugdenhil et al., 2017). However, the disadvantage of microwave remote sensing for land surface phenology monitoring has always been the low spatial resolution of microwave observations compared to optical based remote sensing. And to identify the response of different land cover types, the spatial resolution of the EO datasets need to be sufficient.

Sentinel-1 provides for the first time backscatter at a spatio-temporal resolution suitable for land surface phenology monitoring. The Sentinel-1 satellites provide co- and cross-polarized backscatter observations from the C-band SAR instruments at 5 x 20 m resolution. The co- and cross-polarized data enables the calculation of the VH/VV ratio, the cross-ratio (CR). Generally, cross-polarized backscatter increases stronger with volume scattering and soil vegetation interaction than co-polarized backscatter. Since vegetation is a volume scatterer, cross-polarized backscatter increases stronger with vegetation than co-polarized backscatter, hence the CR increases with increasing vegetation. Studies have already shown that Sentinel-1 CR is sensitive to vegetation dynamics (Veloso et al., 2017, Vreugdenhil et al., 2018). On a larger scale, at 1°, Rötzer et al. (2017) found a strong linear relationship between HV cross-polarized backscatter coefficients from Aquarius and VOD from SMOS. This study demonstrated the potential of merging cross-polarized backscatter with VOD products. However, an extensive comparison between CR and VOD, and their potential for land surface phenology monitoring is still missing.

Here we present the results of two evaluation studies that were performed to asses synergies between CR and VOD; a. between Aquarius based CR, Metop ASCAT based VOD and MODIS Leaf Area Index (LAI) over the Continental USA (CONUS), and b. between Sentinel-1 CR, Metop ASCAT based VOD and MODIS LAI over Europe. The temporal correspondence between all products was assessed by temporal Spearman rank correlation (Rs) analysis and the spatial comparison of phenological parameters such as start and peak of season. The temporal correlation analysis between Aquarius CR, Metop ASCAT VOD and MODIS LAI over CONUS showed high correlation between the microwave based products and LAI, especially for VOD where Rs was 0.77. Moderate correspondence was found between CR and VOD with Rs of 0.43. Over areas where a strong correlation (Rs>0.5) between CR, VOD and LAI was observed, start and peak of season from microwave observations correspond to start and peak of season from LAI. Differences between start and peak of season were observed, where in most regions CR and VOD were concurrent, but LAI preceded both microwave based products in southeastern CONUS. The results demonstrate the complementary information in microwave based products next to optical based products. In addition, the study shows the similarities between CR and VOD and the potential of combining Sentinel-1 based CR with VOD products for vegetation and land surface phenology monitoring.

Jones, M.O., Jones, L.A., Kimball, J.S., McDonald, K.C., 2011. Satellite passive microwave remote sensing for monitoring global land surface phenology. Remote Sensing of Environment 115, 1102–1114. https://doi.org/10.1016/j.rse.2010.12.015
Liu, Y.Y., de Jeu, R.A., McCabe, M.F., Evans, J.P., van Dijk, A.I., 2011. Global long-term passive microwave satellite-based retrievals of vegetation optical depth. Geophysical Research Letters 38.
Rötzer, K., Montzka, C., Entekhabi, D., Konings, A.G., McColl, K.A., Piles, M., Vereecken, H., 2017. Relationship Between Vegetation Microwave Optical Depth and Cross-Polarized Backscatter From Multiyear Aquarius Observations. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 10, 4493–4503.
Veloso, A., Mermoz, S., Bouvet, A., Le Toan, T., Planells, M., Dejoux, J.-F., Ceschia, E., 2017. Understanding the temporal behavior of crops using Sentinel-1 and Sentinel-2-like data for agricultural applications. Remote Sensing of Environment 199, 415–426.
Vreugdenhil, M., Hahn, S., Melzer, T., Bauer-Marschallinger, B., Reimer, C., Dorigo, W.A., Wagner, W., 2017. Assessing vegetation dynamics over mainland Australia with metop ASCAT. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 10, 2240–2248.
Vreugdenhil, M., Wagner, W., Bauer-Marschallinger, B., Pfeil, I., Teubner, I., Rüdiger, C., Strauss, P., 2018. Sensitivity of Sentinel-1 Backscatter to Vegetation Dynamics: An Austrian Case Study. Remote Sensing 10, 1396.

High spatial resolution missions as Sentinel-2 open opportunities to set-up operational Earth Observation (EO) services at local scale. However, due the frequent cloud coverage in Western Europe, combined with the lower revisit time of high spatial resolution sensors, operational services have to combine data from different missions in order to have sufficiently frequent observations. This is for instance done within the WatchITgrow platform, an EO-based service for potato monitoring and yield prediction, where Sentinel-2 and DEIMOS-1 are combined in order to ensure capturing changes in the potato fields.
A seamless combination of EO products coming from different missions is not straightforward. The joint use of data from different sensors raise some clear concerns about data consistency. There might be radiometric biases at TOA (Top-of-Atmosphere) level. Furthermore, even for a perfectly cross-calibrated constellation of instruments, intrinsic differences in the Relative Spectral Response Functions (RSRFs) of comparable bands might cause discrepancies in the final products. Finally, biases in the products can also be introduced through the use of different processors and algorithms (e.g. for the atmospheric correction). One option is to normalize time series at the product level following an empirical approach. The drawback of such an approach is that the root cause of an observed bias is not identified. Furthermore, a non-linear empirical relationship could be found with associated large uncertainty. Finally, such an empirical relationship is only applicable to the sites or time periods for which it has been derived.
Within the BELHARMONY project (https://belharmony.vito.be/), which we present here, a bottom-up approach is used instead in order to assess and to improve multi-mission high resolution time series consistency.
First, vicarious radiometric calibration methods are used to assess the consistency at the L1 TOA level. This is done for the following sensors: Sentinel-2A/B MSI, Landsat-8 OLI, PROBA-V, and Deimos-1. The bias assessment is performed over targets with low (Bassani and Sterckx, LPS2019) as well as targets with medium radiance levels (Wolters and Sterckx , LPS 2019) . This distinction is required (but often neglected) as, particularly at very low (and also extremely high) scene brightness, the radiometric response of the optical system might not be linear.

Next, to adjust the spectral response of one sensor to another, we model the difference that is related to the difference in the RSRFs , allowing to derive band- and/or index-dependent spectral adjustment functions. For this, we use simulated vegetation spectra derived from physically-based radiation transfer models that consider the leaf optical properties, the canopy structure, and the background reflectance. These simulations are completed by adding spectral libraries of non-vegetated surfaces.

Finally, we generate enhanced Level 2 and Level 3 time series over the BELAIR sites using MORPHO, a multi-sensor Level 1 to Level 3 processing chain. In the frame of BELAIR, test sites have been set up in Belgium, for which targeted EO data and other measurements are collected on behalf of the Belgian and international research communities, and which may be used as calibration and validation sites for new EO missions, data, and products. Using the extensive set of EO data and in-situ reference data which have been systematically collected over the BELAIR urban, agricultural, and coastal waters sites since 2013, we will evaluate the impact of the different harmonization measures.