Drought is the primary concern of small holder, subsistence farmers in Kenya. When a catastrophic drought occurs small holder farmers and pastoralist’s communities are severely impacted. They are challenged to dispose of their remaining assets (e.g. cows, land, etc.), seek temporary employment or migrate.
The frequency and severity of adverse events like drought and floods are on the rise partially as a result of global warming. Ironically, it is not uncommon for both drought and floods to occur in the same year, in the same area in different seasons
Farmers have already identified strategies to combat drought, such as efficient irrigation systems, infrastructures for water storage, farming good practices, genetically resistant crop varieties, etc. Among these options - not all applicable everywhere, Agricultural insurance is a risk management option for farmers and pastoral communities to transfer the risk of crop damage to the insurer. In turn, the insurance transfers the same risk to the international reinsurance market, and in that process a nationwide safety net is created

Index-based approaches used thus far by the insurance sector in various schemes (e.g. Kilimo Salama, Takaful) are related to meteorological data collected by ground stations. Examples of meteorological observations are rainfall, temperature, relative humidity and sunshine hours. These stations are expensive in terms of maintenance and, in developing countries, are not always automated (and can be subject to interference or vandalism). In addition, due to their actual poor spatial distribution, especially in remote areas, such a network either doesn’t exist or is not fit for purpose.
As drought is an extended peril affecting extremely large areas, sometimes nationwide, the government may accept considerable quotas of risk transfer and provide government intervention to manage large scale disasters. Given the geographic scale of drought, satellite based technologies providing EO (Earth Observation) data are suitable for mapping out areas exposed to risk. With regard to the timeframe EO technologies may find unparalleled applicability in the following cases:

• Pre-event: archive of damaged areas, needed for actuarial purposes to generate risk maps;
• During event: maps provided in near real time to monitor the situation and take decisions in emergency time;
• Post-event: maps and statistics to monitor the recovery phase.

Unlike an emergency service, a satellite based index service for farmers, insurers and government aims at managing the risk before the damage occurs, thus using EO products to monitor and potentially forecast a certain crop failure for an upcoming drought event.
The goal here is to move from the concept of emergency management to that of risk management.
This study proposes a new high-resolution Vegetation Health Index (VHI) as an indicator of the vigor of the vegetation cover. This index is a function of the vegetation index (Normalized Difference Vegetation Index - NDVI) and thermal properties (land brightness temperature), minima and maxima in a certain period. Successfully applied in numerous case studies (Kogan, 1995, 1997), the VHI has been already tried experimentally for African environments and conditions (Rojas et al. 2011) where it exhibited its performance, making it ready to be tested for insurance purposes.
VHI drought maps, derived from Copernicus Sentienl-2 (for NDVI) and MODIS-Terra (for thermal information), were generated every 10 days at 20 m resolution for large areas in Kenya. The database is structured as a GIS capable datastore for time series data with an online dashboard which offers to the identified stakeholders, a drought monitoring service. The service is undergoing in-country trail for deployment to Kenyan insurances, farmer cooperatives and Ministry of Agriculture in Kenya. Additional capabilities of the system include analytics from multi-temporal statistics at field level, drought early warning and agricultural consultancy.

A leading part of the SIBELIUs project is delivering improved severe-weather resilience for Mongolian herding communities using environmental rangeland products derived from satellite Earth Observation data. In Mongolia, extreme weather events called ‘dzuds’, can result in high livestock mortality which can be devastating for herders and for the wider economy. A dzud can occur because of: drought, poor pasture quantity in summer, heavy snow in winter and low winter temperatures. Often, a combination of these factors will cause a particularly bad dzud and the effects of this phenomenon is amplified by livestock overgrazing and climate change.
SIBELIUs will provide stakeholders with access to data from the Sentinel-2 and Landsat 8 satellites, to deliver regular information on pasture, drought and snow conditions across the vast rangelands in Mongolia. SIBELIUs is working with the National Agency of Meteorology and Environmental Monitoring (NAMEM) to improve the systems already in place in the country and to ensure the sustainability of the project after funding ends.
Indices such as Normalised Difference Vegetation Index (NDVI), Normalised Drought Index (NDDI) and Normalised Difference Snow Index (NDSI) will be derived from Sentinel-2 and Landsat 8. These products, together with comprehensive ground measurements conducted by NAMEM will allow time series derived products such as pasture trend and anomaly products, drought dynamics and snow cover maps to be provided.
A key challenge for SIBELIUs is delivering environmental products on a regular basis over such a large country (over six times the area of the UK). To address this issue, medium and high-resolution Data Cubes have been developed which sit at the heart of the data processing and handling system. Initially SIBELIUs is focussing on three test sites but will roll-out the service to the whole of Mongolia before the project’s end in 2021.
SIBELIUs will also help to improve the dissemination of information through existing Mongolian channels, including: the insurance sector, governmental agencies and more directly to the herders themselves and via regional governors. Currently being investigated are systems to send SMS messages to herders on request, getting SIBELIUs products on television weather reports watched by herders, physical paper copies of maps and web applications to local government staff.
The use of freely available Sentinel and Landsat data supported by new infrastructure within established Mongolian institutions will ensure the long-term sustainability of the SIBELIUs services. Furthermore, the use of a Data Cube will make this rich data source easily accessible to key Mongolian stakeholders, therefore potentially stimulating a range of possible further environmental, scientific and commercial applications.
SIBELIUs is supported by the UK Space Agency as a part of its International Partnership Programme.

Landslides represent one of the main hazard in dissected landscape, especially in tropical areas where a combination of active tectonics, steep topography, intense rainfall and high population density is found. Many regions of the Global South are facing such a situation, with landsliding resulting every year in fatalities, structural and functional damage to infrastructures and serious disruptions of the organisation of societies. The rapidly expanding city of Bukavu (DR Congo), located in the Kivu Rift, is experiencing such a challenging combination. Here, we analyse the kinematics of a large (2km²), densely inhabited deep-seated landslide situated in the city. We use satellite interferometry (InSAR), archive aerial photographs (50’s and 70’s), dGPS and multi-temporal topographic reconstructions obtained from UAV-SfM to resolve both the long-term and the seasonal behaviours of the slide. The Multidimensional Small Baseline Subset (MSBAS) method is used to obtain a temporally dense, 4-year multi-sensor InSAR time series that combines Sentinel 1 and more than 400 COSMO-SkyMed SAR images from both ascending and descending sets. Validated by in situ dGPS measurements, these deformation measurements provided by InSAR are also complemented with very high spatial resolution multi-temporal topographic reconstructions from UAV-SfM. This allows for exhaustive validation of remote sensing products, but also provides complementary information in zones of rapid deformation where the InSAR measurements are restricted by coherence loss. We show that the landslide displays different deformation patterns with contrasting mean velocity of up to one meter/year and movement styles. By confronting the deformation time series to on-site rainfall and seismic records, we show that the seasonality of the precipitation pattern influences the slide behaviour. The analysis of these deformation patterns with regard to the environmental changes derived from the aerial photographs allows to further explore the interplay between natural and human controls over several decades. The outputs of the research should help improve the evaluation of landslide hazard and mitigation in the area, but also across other regions where similar environmental conditions are met. This research is carried out within the framework of the Belgian Science Policy (BELSPO) RESIST and MODUS projects.

Mapping landslides after major triggering events such as earthquakes and large rainfalls is crucial for disaster response, hazard assessment, as well as for having benchmark inventories on which landslide models can be tested. Numerous studies have already demonstrated the utility of very-high resolution satellite and aerial images for the elaboration of inventories based on semi-automatic methods or visual image interpretation. However, while manual methods are very time consuming, faster semi-automatic methods are rarely used in an operational context after major triggering events, partly caused by data access restrictions on the required input (i.e. VHR satellite images) and by the absence of dedicated services (i.e. processing chain) available for the landslide community. Several on-going initiatives should help to go beyond these limitations.
First, from a data perspective, the free access to the Sentinel-2 and Landsat-8 missions offers opportunities for the design of an operational service that can be deployed for landslide inventory mapping at any time and everywhere on the globe. Second, from an implementation perspective, the processing platforms (such as for instance the Geohazards Exploitation Platform –GEP– of the European Space Agency –ESA–) allows the integration and diffusion of on-line processing algorithms in a high computing performance environment. Third, from a community perspective, the Landslide Pilot and the Recovery Observatory of the Committee on Earth Observation Satellites (CEOS), has targeted the take-off of such service as a main objective for the landslide and risk community.
Within this context, we present a largely automatic, supervised image processing chain for landslide inventory mapping from bi-temporal (before and after a given event) multi-spectral images. The processing chain combines change detection methods, image segmentation, higher-level image features (e.g. texture, shape) and topographic variables. Taking as input a sample of manually mapped landslide polygon, a random-forest classifier is trained and subsequently used to distinguish newly triggered landslides from other landscape elements.
The processing chain is tested with both medium resolution (Sentinel-2, Landsat-8) and high-resolution (Spot-7) images on three case studies recently impacted by strong landslide disasters (e.g. New Zealand, Myanmar and Haiti).
The landlside inventory maps are provided along with an uncertainty map that allows identifying areas which might require further considerations. Effects of the segmentation features and on the size of the training samples are also discussed.

The primary goal of volcano monitoring is to provide timely warnings of volcanic hazards to decision makers and the public. Volcano observatories around the world have traditionally relied on ground-based monitoring networks of seismic, gas and geodetic instrumentation to provide this advice. In recent years, data acquired by remote sensing satellites has played an increasing role in volcano monitoring, facilitating measurements at multiple volcanoes simultaneously without requiring equipment or personnel on the ground. Remote sensing observations are especially valuable in cases where there is no in-situ monitoring capability, as they provide the only means of identifying precursors to eruptions. One such remote sensing technique, satellite radar interferometry (aka InSAR), makes use of repeated radar images to map ground displacements. The occurrence of ground displacement signals at volcanoes has been shown to strongly correlate with the occurrence of volcanic eruptions in many cases. InSAR has therefore become a valuable tool to provide evidence of impending eruptive activity at remote volcanoes, and those volcanoes previously thought to be inactive. Despite these successes, the majority of volcano observatories do not readily use InSAR to support their operational activities. In this presentation, we will discuss the barriers faced by volcano observatories in using InSAR to support their operational monitoring mandate. In an effort towards overcoming these barriers, we will discuss how InSAR can be applied to deformation monitoring of active volcanoes and present a simplified InSAR processing workflow designed to be used directly at a volcano observatory in any resource-constrained environment. As a case study, we will focus on Papua New Guinea, a lower middle income country with over 50 Holocene volcanoes and a single volcano observatory responsible for all monitoring activities. We will show results obtained from the simplified InSAR processing workflow for the Rabaul Caldera, Papua New Guinea’s most high risk volcano, including a comparison between InSAR and GPS observations. We will also present the results of a country-wide L-band InSAR survey, which we use to identify targets for future operational InSAR monitoring of volcanoes in Papua New Guinea.

The Kivu region is a densely populated area hosting two very active volcanoes, Nyiragongo and Nyamulagira, which require continuous surveillance using the widest means of observation as possible.
The Belgian BELSPO RESIST project aims at developing advanced algorithms and data networks for improving volcanic monitoring in the region, and ultimately explore the added-value information in combining different types of data. This study presents an overview of space-based sulfur dioxide (SO2) data record obtained during RESIST, form nearly 15-years of AURA/OMI and recent Sentinel-5 Precursor/TROPOMI observations. For OMI, short- and long-term changes in volcanic SO2 emissions are investigated and satellite data oversampling is used to discriminate the volcanic sources for the full mission. For TROPOMI, we show that the improved detection limit and spatial resolution of the instrument (3.5x7km² compared to 13x24km² for OMI) enables to discriminate volcanic sources on a nearly daily basis and that small changes in degassing patterns are better tracked with TROPOMI than with any other satellite sensor.
We explore the potential of the obtained satellite SO2 dataset by combining the results with in-situ observations of seismicity and lava lake levels, from the ground network that operated continuously during that period. The overarching aim being to improve our ability to interpret signs of volcanic unrest on a daily time scale.