SMOS Fast Reprocessing Platform at ESAC
The Data Processing Ground Segment (DPGS) of SMOS mission for the study of the water cycle is located at the European Space Astronomy Center (ESAC), in Villafranca del Castillo (Madrid).
The satellite was launched into space in November 2009 with an expected initial life time of 3 years, but its excellent technical and scientific status has allowed it to expand until 2019 and its continuity is currently being evaluated in order to be extended until 2024.

Indra’s role in DPGS
Indra, as the company that led the development of the Data Processing Ground Segment of SMOS, has been in charge of its corrective and preventive maintenance since the beginning of operations phase in June 2010. Besides this work, Indra is in charge of the engineering tasks carried out for supporting the operation of the DPGS Fast Processing Centre platform (FPC, for routine operational processing to L1 and L2 of satellite downlinked data) as well as for the execution of tests of new DPGS subsystems versions in the Integration and Maintenance Platform (IMP) hosted at Indra’s facilities in San Fernando de Henares (Madrid), before their deployment in DPGS’ FPC at Villafranca.

Reprocessing Platform
Being SMOS’ instrument MIRAS the first of its kind, in the nine years of the current Mission duration the processing algorithms have had to be evolved constantly by experts in seek of the best SMOS data scientific value for the international climate research community.
For such purpose, it has been fundamental the use of the specific SMOS DPGS Fast Reprocessing Platform (FRP), developed by Indra in 2010 under ESA contract and evolved in 2012 and 2015. The intensive processing capacity of this platform, along with its production orchestration capabilities, has allowed supporting the scientific validation of the processing baselines (to evaluate long-term trends, seasonal effects, etc.) before the complete reprocessing campaigns were launched. This, without affecting at all the operation of the FPC, as they are separate platforms.
When a processing baseline is considered validated and stable, it is deployed in the nominal platform for routine production, and the SMOS data since the start of the Mission up to the routine production starting date is reprocessed to L1 and L2 levels in order to achieve a consistent and homogeneous dataset that can be used by experts for scientific climate research. Currently, the processing baseline V7 has been integrated in the IMP, and several metrics exercises have been executed in FRP to support its scientific validation. When the experts consider it stable enough, the full 9 years dataset will be reprocessed.

FRP Scalability
This Fast Reprocessing Platform is based on the same software used by the nominal centre, i.e. the PDPC Core developed by Indra for FPC. However Indra provided critical improvements of this SW that make it possible to reprocess SMOS data up to 36 times faster than FPC production, depending on the level of the final product required.
In fact the speed is currently limited by the orchestration requirements and the number of dedicated processing machines, not by the FRP’s PDPC-Core SW scalability. For example, the orchestration requests consolidating and post-processing all L2 data of one day to generate a set of daily auxiliary files to be used as input for the L2 processing of the next day (i.e. the daily production is forced to be sequential). The reprocessing instance triggers in parallel the L2 processing tasks for all products from one day (29, one for each semiorbit), then consolidates all the outputs and provides them as input dataset to L2 post-processors, that generate the auxiliary daily files for the next day. The reprocessing speed is limited by the time that the slowest L2 processor task from each day takes to process L1 data to generate the L2 products, and no further parallelisation can be applied in that reprocessing instance.
Indra has overcome this difficulty by allowing to set-up several reprocessing instances in the FRP, each with its own processing machines farm, but managed coherently. The SMOS data timeline is divided in several segments, and each is processed in a different reprocessing instance. This compensates the limitations imposed by the orchestration to each instance, and allows achieving reprocessing speeds of e.g. 36:1 for an FRP with 3 reprocessing instances, each with a speed of 12:1 thanks to the L2 processor speed.
The platform is also equipped with a set of tools to simplify management and analysis of results of different instances of Fast Reprocessing Platform (FRP). This set of tools allows parallel execution of different periods for same Reprocessing Campaign by handling of Baselines (orchestration, processing, static auxiliary data, etc.).

Evolution to G-CORE
As part of the preparation for the mission extension, the evolution of DPGS to Indra’s G-CORE product is being analysed. G-CORE is an evolution of the SMOS PDPC CORE software that was developed for SMOS mission (based on the heritage of Helios I Data Processing Centre also developed by Indra), to become a flexible and easily customizable software solution intended to cover many of the processing management related functions of the GS PDGS.

- G-CORE provides high customization and configuration capability. The G-CORE can be configured to manage the data of several missions; ready to support the multi-mission concept.
- G-CORE is ready to support highly complex orchestration schemes.
- Its configurability capability allows the Operations team the easy integration of new versions of current processors or new processors.
- Reliable solution well proven in several scenarios: G-CORE is currently deployed on PAZ Mission GS (already operational) and INGENIO GS (in development), and other Indra products such as Multimission DAT for PAZ, TerraSAR-X and TamDEM-X, Land Analytics EO Platform and PAZ Users Services Provider.
- G-CORE will also allow the deployment of the reprocessing platform in a cloud including the capability of auto-diagnostic for auto-provisioning and additional activities to monitor and procure additional resources in different types of CSP; public, private or hybrid. The inclusion of the capacity to reduce and increase automatically the processing capacities (elasticity) according to the necessities of the moment (satellite pass, for example) can reduce the cost of the service and allow to offer a best price to the client.

Area C - Board 242

European Space Agency’s Soil Moisture Ocean Salinity (SMOS) satellite was launched on 2nd November 2009. Its payload MIRAS (Microwave Imaging Radiometer with Aperture Synthesis) is an L-band interferometric radiometer which achieves unprecedented brightness temperature (BT) multi-angular measurements with a spatial resolution in the range of 30 – 60 Km. After about ten years of successfully operation, the SMOS mission has provided high-quality, global maps of soil moisture over land and sea surface salinity and sea ice thickness over ocean. Recently, wind speed maps over ocean and freeze/thaw soil state maps have been added to the products’ portfolio of this versatile mission.

This paper reports on the result of the assessment of the long-term temporal and spatial stability of the SMOS L-band BT dataset. Instrument stability might be impacted by several factors such as seasonal changes in the thermal state of the satellite, instrument evolution, aging and calibration drifts. In order to characterize the long-term temporal and spatial stability of SMOS L-band BT, the emission from well-known, stable geophysical targets, as well as the results of internal instrument calibration procedures, has been used.
Three BT stability metrics are presented: i) Stability over ocean, inferred by comparison with an ocean model over a large area of the Pacific Ocean without significant land bodies; ii) stability over ice by comparison of the SMOS L-band brightness temperature with DOMEX-3 experiment (over Dome-C base in Antarctica, a well-known stable target) iii) stability over cold sky by monitoring the calibration parameters and cold sky BT observations. The combination of this three metrics provides a good overview of the stability of the SMOS L-band BT for different scenarios: cold, warm and hot targets .
The SMOS L-band BT evolution is also compared with the evolution of the main instrument calibration parameters: power measurement system (PMS) gain and offset, visibility offsets, and fringe washing function amplitude. These parameters provide an overview of the instrument stability along the mission.
It is shown that SMOS L-band BT stability performs well in general, without any major drifts attributable to the instrument. In fact, it is demonstrated that some minor observed deviations are coming from geophysical changes in the surface, uncertainties in the models, or drift in calibration parameters not properly compensated. Nevertheless, few relationship between SMOS L-band BT stability and instrument calibration parameters are not fully understood, and need further analysis.

Area C - Board 243

ESA’s Soil Moisture and Ocean Salinity (SMOS) mission is dedicated to making global observations of soil moisture over land and salinity over oceans. By consistently mapping these two important components in the water cycle, SMOS is advancing our understanding of the exchange processes between Earth’s surface and atmosphere and is helping to improve weather and climate models.
SMOS satellite was launch on November 2009, and after 9 years of successful Operations, the status of the SMOS Data Processing Ground Segment (DPGS) is excellent, providing very reliable acquisition, nominal and NRT data processing, reprocessing, archiving, and dissemination services with several hundreds of end-users around the entire world.
Against the common perception that the Operational Ground Segments are static and conservative by its own nature, the architecture of the SMOS PDGS has been always evolving to respond to new operational and science requirements, the incorporation of new science products, and to improve the robustness, maintainability, and efficiency of the current system using the latest available techniques.
This paper will summarize the efforts of the SMOS DPGS team in three different areas: the flexible provision of the required storage and processing power; the improvements in the monitoring and control systems; and the detection, analysis, and reporting of the Radio Frequency Interference (RFIs) affecting the SMOS observations.
The accumulated SMOS data volume combined with the necessity of periodic full mission data reprocessing, requires an increasing SMOS processing power and storage. To cope with these needs, the DPGS team is constantly evolving the SMOS IT infrastructure, deploying new physical and virtual reprocessing nodes in a private cloud. To add more flexibility and scalability, DPGS Team is designing a complementary Docker based platform and its corresponding orchestration tools to adapt its infrastructure for new processing challenges such as world RFI detection and monitoring in near real time.
Based on the DPGS team's knowhow and accumulative knowledge, new tools are being developed to monitor systems and manage contingencies and alarms, both in the hardware infrastructure and operational processes, to avoid delays in the NRT chain, which are taking more interest in the meteorological agencies. The current monitoring systems already implemented in the operational ground segment are being adapted to extend its functionality available from internet in a secure and user friendly way.
Since the first SMOS mission observations, its radiometer in the passive band 1400-1427 MHz detected Radio Frequency Interference (RFI) in several world areas, despite the ITU Radio-Regulations (RR No.5340) that prohibit any in-band emissions.
SMOS´ objectives are being disturbed by these interference sources that jeopardize part of its scientific use in certain areas of the world, particularly over continental areas in Europe, Southern Asia, and the Middle East. In order to minimize the impact of RFIs and to improve scientific retrieval, several strategies were initiated shortly after the SMOS operations started in 2010: RFI detection and flagging, image mitigation techniques, increase awareness and support initiatives to improve regulatory framework, and RFI Reporting to the different national telecommunication authorities.
The analysis and processing of the RFI have substantially improved, providing more information for each source detected, and thus helping to locate more efficiently the emitting devices in the field. In addition, cooperation with National Administrations is greater than ever, and the information provided on their interventions in the field is growing.
In order to classify and convert all this data into useful information, the SMOS RFI team is developing different tools. The objective is to monitor the status of RFI in the world in near real time and to provide any interested user (scientific community, national authorities, radio astronomy agencies, Earth observation missions, Satellite Ground Segment Teleports…) with daily information on L-Band interference, for their mitigation, management or study.

Area C - Board 244