Figure 1. ERS Payload Data Ground Segment
The mission planning activities which are performed at ESRIN in order to schedule data acquisitions for the Synthetic Aperture Radar (SAR) instruments on board of ERS-1 and ERS-2 (alone and in tandem) are described starting with their initial organisation and then their evolution based on experience and change in requirements. This is preceded by a short description of the ERS Ground Segment and an introduction which lists other key mission factors and summarises ESRIN's responsibilities in Earth Observation (ESA and Third Party Missions).
The major topics discussed are: user interface, user requests and their conflicts, baseline plans, data policy, mission guidelines; platform, sensors, ground segment and exploitation constraints; as well as planning tools, manpower needs, and interfaces with ground stations. The experience gained can be used for future missions in the identification of realistically achievable objectives, the definition of the offers to users, and the design of the mission planning system, in particular for user interfacing, mission planning tools and preparation of agreements with ground stations.
The ERS-1 and ERS-2 satellites, launched respectively on 17 July 1991 and 21 April 1995, currently represent a unique case of a dual remote sensing mission and have provided results beyond expectations, in particular for tandem data acquisition with a one-day difference. This provides data to many user categories such as real-time operators involved in meteorological, oceanographic and environmental applications, long-term research groups working off-line, commercial users, etc. The two satellites, carrying the set of instruments listed in Table 1 on board, have been exploited in the different mission phases listed in Table 2.
Table 1. ERS-1 and ERS-2 On-board Instruments
Instruments ERS-1 ERS-2 Active Microwave Instrument SAR Image Mode X X SAR Wave Mode X X Wind Scatterometer X X Radar Altimeter X X Along Track Scanning Radiometer-1 X Along Track Scanning Radiometer-2 X Global Ozone Monitoring Experiment X
Table 2. ERS-1 and ERS-2 Mission Phases
Mission Phases Start Cycle SAR Mission Objectives ERS-1 - Launch 17-Jul-91 - Payload switch-on & verif. 17-Jul-91 A Commissioning 25-Jul-91 3 days all instruments; until 10-Dec-91 B Ice 28-Dec-91 3 days ice & pollution; interferometry possibility R Roll-tilt (experimental) 02-Apr-92 35 days different SAR incidence angle (35°) C Multi-disciplinary 14-Apr-92 35 days AO; land & ice mapping; consistent set in regular intervals D 2nd Ice 23-Dec-93 3 days see Phase B E Geodetic 10-Apr-94 168 days radar-altimetric mission; SAR as Phase C F Shifted Geodetic 28-Sep-94 168 days 8 km shift vs. Phase E for denser grid G 2nd Multi-disciplinary 21-Mar-95 35 days see Phase C G Tandem 17-Aug-95 35 days interferometry & mapping G Back-up 2-Jun-96 35 days ERS-2 - Launch 21-Apr-95 - Payload switch-on & verif. 21-Apr-95 35 days A Commissioning 02-May-95 35 days SAR commissioning A Tandem 17-Aug-95 35 days see ERS-1 Tandem Phase G A Multi-disciplinary 3-Jun-96 35 days see ERS-1 Phase C Note: 3 days = 43 orbits; 35 days = 501 orbits; 168 days = 2411 orbits
The ERS Payload Data Ground Segment, sketched in Figure 1 and providing SAR coverage as per Figure 2, is managed by ESRIN, via its Earth Remote Sensing Exploitation Division (RS/E), and is composed of:
Figure 1. ERS Payload Data Ground Segment
Figure 2. Network of all stations (ESA & NFSs)
In this paper, the term 'mission planning' indicates only those activities performed at ESRIN for planning SAR acquisitions. Other linked activities, such as planning of LBR instruments (mostly performed by default at ESOC) or production planning are not discussed.
SAR acquisition planning, even if a complex task, has not caused bottlenecks or problems for the ERS mission. The overall effectiveness of the ERS SAR mission was, or is, affected much more by:
ESRIN is in charge of the planning and handling of the Earth Observation data from ESA and Third Party Missions (TPMs). For TPMs (e.g. Landsat and JERS-1), planning is limited to the collection of user needs, the definition of other potentially relevant acquisitions, the transmission of the resulting plan to the satellite operator, and the scheduling of ESA stations. On the other hand, ERS-1 and 2 SAR activity planning is far more complex because, over and above the number of on-board instruments and the parallel activities of two satellites, it must match user requirements (user requests), gathered through the user interface, and the baseline plan (derived from mission/data policies, anticipated user needs and contingency planning), with the system constraints. This activity is supported by a balanced combination of dedicated planning tools and manpower, and relies on interfaces with the ground stations. All these elements are discussed below, with a description of their initial implementation and their evolution in line with the experience gained and the changing requirements.
The user interface has been organised around three service desks: the ESA Help Desk (for information, documentation, tools, etc., to all users), the ERSC Customer Service (for commercial users) and the ESA Order Desk (for non commercial users); correspondence can be exchanged through fax, telephone, letter, e-mail, etc., as selected by the user.
Currently 3378 users are registered, of which only 922 have on-line access (mainly e-mail). Of the total number, 841 have submitted at least one user request (of these only 323 have e- mail). It is evident that normal correspondence media are still widely used. Their use will continue in the coming years especially in view of the opening of Eastern and African markets (telecommunication links to be set-up).
User interfacing for ERS was designed to serve a variety of user categories (see Table 3):
The service was set-up to treat all users on an identical footing.
Table 3. Current number of users per user category
User Category No. of Users AO / PP 1113 NA / FO 32 Planning 7 Commercial 291 ESA 239 No Project 2140 --------------------------------------------- Total number of users 3378 (some users belong to several user categories)
After the start of the mission, it was evident that interaction with users was more difficult and demanding than expected, despite the fact that some key documents describing the system had been prepared and widely distributed (some users had the impression that we could move the satellite wherever necessary). It became essential, particularly for planning, to improve the user's "visual" knowledge of the mission, and to ensure that the user and our service desks clearly understand each other. Therefore the graphical, simple and powerful Display ERS-1 SAR Coverage (DESC) tool, running on PCs, was developed and distributed. It was enhanced over time through valuable user feedback, up to the most recent Display ERS Swath Coverage for Windows (DESCW), which is multi-mission, supports quick-look display, provides on-line help, etc.
DESCW shows graphically the coverage of the various sensors in the future and/or in the past (through inventory search and filter). It is based on visibility files for possible future acquisitions and on compressed inventory files for past and planned acquisitions. The inventory files are either historical (past years) or updated weekly and are available online for free-of-charge downloading via FTP or Internet, together with the software and all supporting data. The entire software, the basic files, the help text, the inventory files (about 30 years of inventory data in total for ERS-1, ERS-2, JERS-1 and Landsat) take less than 4 Mbytes and, therefore, are also distributed on three PC diskettes on user request.
Over time, DESCW has been more and more used by our service desks and also by the mission planner, particularly to identify possible acquisition conflicts with other missions (the ERS mission planning system is not multi-mission), to derive rough indications useful for detailed mission planning and to quickly check future planning over small areas.
The system was designed to permit formalisation of user needs through user requests, which, for acquisition planning, mainly define the area and time period of interest and can equally accurately identify single frames and very large acquisitions (e.g. the full station visibility area for some months).
Large sensing requirements must be submitted for planning about one month ahead of acquisition, limited requests up to five working days ahead and exceptional cases have been handled up to two to three working days ahead (uplinking of the spacecraft telecommands is done one day before the acquisition). User requests can be submitted and their status verified on-line through dedicated forms, via X.25 and VT200 terminals. Users are also actively informed via e-mail or fax of major status changes.
Figure 3 shows the total number of user requests per user category since mission start and Figure 4 shows their variation over time. It must be noted that:
Figure 3. Total number of user requests per category
Figure 4. Number of user requests per user category vs. time
A few months after exploitation started, it was realised that some NFSs and most of the investigators were submitting large acquisition requests, causing extra overheads in mission planning and a possible waste of satellite resources. Since most of the investigations had a production quota defined, the investigators were asked to limit their acquisition requirements to those which could be associated to a future product. This simple measure permitted a drastic reduction in the size rather than the number of user requests. However, when justified, the excess acquisitions were accepted within the baseline plan (see below).
The need to speed up the provision of information to users in case of sensor unavailability soon emerged (some users take in-situ measurements during satellite passes). Therefore, an automatic procedure was added to inform all affected users, immediately, via fax.
Before the launch of ERS-1, it was realised that there was a need for an ESA baseline plan (a set of mission planner user requests) for the implementation of a data policy and mission guidelines (see Table 4 for the most relevant ones), and for the collection of data of potential commercial, operational or scientific interest. In particular, the mission guidelines, defined for each mission phase in the high level operations plan, influence planning over selected areas depending on the phase, while data policy has large impacts on data requests from NFSs.
Table 4. System Constraints
(- = initial;* = close to ERS-1 launch; / = not implemented; + = during exploitation)
Data Policy:
- national stations (with signed agreement) can acquire data on a non-interference basis
- stations with approved agreement can request data acquisition (at no cost for national stations
and for foreign stations with no-exchange of funds agreements)
Mission Guidelines:
- adhere to mission objectives (phase dependent)
- SAR has priority in descending passes, Wave and Scatterometer in ascending passes (night)
- solve conflicts applying priorities to the user categories and then to users according to past allocation
/ allocate acquisitions for AOs within the assigned quota, in a 6 months moving window, varying
their priority according to remaining time, past allocation for country and application category
+ LBR activity has priority over SAR in ascending passes, every other cycle
Platform and Sensor Constraints:
- SAR can be activated only within visibility of a ground station (no HR tape recorder on-board)
- in each 100 min. orbit, SAR can be activated <12 minutes in total, <10 minutes per segment
on descending passes, <4 minutes in eclipse (in addition, merge gaps <30 seconds)
- maximum number of SAR on/off switches = 6 per orbit
- SAR imaging mode of AMI mutually incompatible with SAR Wave mode and Windscatterometer
- Windscatterometer must be switched on 128.2 seconds (850 km) before and after the site
of interest
Ground Segment Constraints:
- take into account the real station visibility mask in planning SAR sensing
- instrument planning and Kiruna station scheduling must follow defined time constraints
* schedule SAR sensing from 5 to 2 degrees above horizon
* handle station unavailability at major subsystem level
+ adhere to ground station specific operational constraints, like: working hours (depending on campaign,
country or religion), conflicts, available tapes, interval between adjacent passes, etc.
+ schedule all stations within visibility of planned segments, unless no agreement exists, unless there is a
national or commercial request, or a natural disaster unless there is no hope to serve the user
+ schedule overlapping stations depending on reliability and on station or PAF processing capability
+ some stations report on acquisitions with a variable delay (even months, causing loss of replanning opportunities)
and occasionally their reports are discovered to be incorrect
+ some MoUs signed later than expected or signature proceeding with difficulties
+ reduce number of HDDTs avoiding overlapping acquisitions and minimise night passes to save manpower
Exploitation Constraints:
* avoid bridging of specific segments (precise start flag)
/ monitor and control energy and thermal balances over and across orbits
/ handle SAR gain setting at user request level
/ permit planning of sensor modes (e.g.: OGRC / OBRG)
/ handle solar panel occultation of downlink antenna (changing over the year and with latitude)
+ 12 SAR minutes not per orbit, but from eclipse start to eclipse start (changing with seasons)
+ apply 'common sense (strict application of HLOP rules prevents optimised use of resources)
+ assign higher priority to 'production requests' requiring new planning over 'acquisition only' ones
+ assign higher priority to requests over stations working in campaigns
+ change confirmed requests only in case of natural disasters or calibration needs
+ ensure proper and complete tandem planning (no multi-mission planning tool, user requirements might conflict with
tandem mission, ground stations operational constraints more difficult to match, etc.)
+ keep to a minimum the number of IDHT on/off switches (from June 1996, to extend lifetime)
+ 'keep alive' scenario requires planning of at least two segments per day, about 12 hours apart
The baseline plan became more defined and complex, being centred on acquisitions for:
During the Tandem phase, data acquisition from both satellites was implemented through a special baseline planning considering:
The baseline planning is currently kept to a minimum in order to extend satellite lifetime.
Some of the system constraints (the major ones are listed in Table 4) are imposed by the physical characteristics of the instruments, spacecraft or orbit, while others are derived from ground segment and exploitation possibilities (of course, more detailed constraints are taken into account at ESRIN and ESOC). Some of the listed constraints make the planning process complex, their relative emphasis changing over time in relation to, for example, day/night, season, mission phase, etc.
The constraints marked with an asterisk in Table 4 were defined a few months before the ERS-1 launch, after a pre- release of the ESRIN mission planning system was delivered, and, therefore, induced late changes. Those with a slash were defined around the same period but were not implemented. The constraints marked with a plus have been encountered during exploitation.
The basic implementation of the ESRIN mission planning system was embedded in the development of the Central User Service (CUS) by MacDonald Dettwiler. In this type of core sub-system, the planning is based on user requests shared with other sub-systems (user request handling, order handling, production planning, etc.), while a specific set of tools (forms, graphics and reports) assists the planner in his activities.
Prior to the mission, ESRIN developed, in conjunction with Advanced Computer Systems, a mission analysis tool to verify possible use of SAR sensing in various mission scenarios (different launch dates and cycles). Since the major concern was related to the probability of acquisition conflicts, this tool was upgraded to test a simple algorithm for possible conflict resolution, reduction, or at least, identification. The problem was extremely simplified, generating for three key types of user requests, all the visible (by a ground station) orbit and frame combinations, with all their possible alternatives. The algorithm was designed to allocate acquisitions starting from the less critical orbits (those with more frames available and less requests) and propagating the effects to all involved user requests (the algorithm was designed to minimise conflicts and not to optimise planning, allocating the minimum number of sensing segments). The results were promising, since, on feeding the tool (which could also have been easily interfaced with CUS) with the available AO user requests, practically no conflict was detected. It was decided to verify CUS planning in practice before connecting this algorithm.
An analysis was made of the real conflicts experienced among user requests. From Table 5, related to Phase C, it is evident that the limited commercial (top priority) requirements could not cause conflicts, while investigators have larger conflict probabilities, even if, with a share of only 1/5 of the total allocation, other resources could have been freed for them if necessary. But this was not the case, since only 0.25% (5 out of 2024) of the requests were in conflict and therefore marginally descoped. During Phase D the percentage of requests in conflict grew to 1.75 % (13 out of 743), because the short repetition cycle (3 days) and phase (3 months) forced the grouping of large requirements from ice scientists over fewer orbits (43 against 501).
Table 5. Frames allocated during ERS-1 Phase C vs. user category (as of June 1996) User Category No. of Frames -------------------------------------- AO / PP 66331 NA / FO 80028 Planning 171116 Commercial 1479 ESA 3305
Currently, the orbit configuration for both satellites is as for Phase C, while large investigator requests are tending to decrease. Therefore, even fewer conflicts are being experienced.
Acquisition planning is currently performed at ESRIN by one contractor supervised by an ESA staff member (part time only), who ensures backup during working days, but also contributes to the preparation of planning documents, defines the detailed acquisition strategy in line with mission guidelines, sets-up the baseline plan, follows specific cases, contacts the stations for special arrangements, ensures correct reporting, etc. This manpower level is only just adequate and, consequently, in periods of a particularly higher load, such as during a Tandem mission, some low priority activities have to be descoped, deferred or cancelled (e.g. internal reporting, analysis of station reports, etc.). The use of an expert system would not have reduced the manpower requirements below this limit, since, in addition to the planner, there would have been a need for an expert-system specialist for adapting the rules according to the constantly varying mission needs (and possibly additional manpower for corrections and tuning).
Besides the internal ESA interface between EECF and MMCC for mission planning, the EECF has planning interfaces with acquisition stations, mainly for sending acquisition schedules and spacecraft ephemerides, and receiving acquisition reports. This loop is essential for user request satisfaction, since, in case of lost acquisitions, the sensing must be replanned if this is still acceptable to the user. This interface was defined in two documents, one for ESA stations and a simplified one for NFSs, both based on files exchanged through telecommunication links using two file transfer protocols (FTAM and FTSV) over X.25.
When NFSs started to join the ground segment, few of them had interfaces in line with the specifications and new interfaces and procedures had to be quickly defined and implemented, based initially on telefax communication. Slowly, over the years, stations started to migrate towards online connections, but using their own preferred protocols. We therefore had to progressively add new protocols to our system in order to simplify our operations, but at the expense of complexity.
Table 6 shows the number of SAR frames acquired over all stations for both missions.
-------------------------------------------------------------------
ERS-1 ERS-2
-------------------------------------------------------------------
Ground Stations Phases A-G Phase A Total
Fucino 99198 24536 123734
Kiruna 252740 38716 291456
Maspalomas 32559 6578 39137
-------------------------------------------------------------------
Total ESA Stations 384497 69830 454327
Aussaguel 4843 0 4843
Gatineau 101314 12708 114022
Libreville 5975 3711 9686
Neustrelitz 3783 3320 7103
O'Higgins 37080 11247 48327
Prince-Albert 150371 19638 170009
Tromsø 215452 27996 243448
West Freugh 54119 6128 60247
-------------------------------------------------------------------
Total National Stations 572937 84748 657685
Alice Springs 22372 5889 28261
Bangkok 5858 0 5858
Beijing 7834 4814 12648
Cotopaxi 7058 407 7465
Cuiaba 19445 3204 22649
Fairbanks 212328 18933 231261
Hatoyama 20874 0 20874
Hobart 3753 2661 6414
Hyderabad 24012 3961 27973
Johannesburg 5870 2919 8789
Kumamoto 18561 0 18561
Norman 2588 2287 4875
McMurdo 12851 12858 25709
Parepare 8395 2 8397
Singapore 2924 2765 5689
Syowa 6700 1183 7883
Taiwan 3990 1256 5246
-------------------------------------------------------------------
Total Foreign Stations 385413 63139 448552
Grand Total 1342847 217717 1560564
Distinct frames 831824 152809 984633
Distinct/Grand Total (%) 61.94% 70.19% 63.09%
Due to the experimental nature of the mission, the complexity of the ground segment, as well as the political drive and sensitivity, it was not possible in the ERS SAR mission planning to anticipate many of the eventual ground- segment constraints and the evolution of requirements during the mission. Therefore, the planning system for future missions of this type should be designed starting with the basics and adding complexity over time, when the real constraints, requirements and possibilities are known. The initial system should be sufficiently flexible, and enough resources should be foreseen, for this expansion/adaptation.
Since the user is an integral part of the system, a large amount of information exchange is necessary. This should take place according to the user's preferred method(s) and possibly supported by a graphical, powerful and user-friendly tool, running in the most popular environment. This tool should be precise and comprehensive enough to also be used internally, in order to communicate with the user on the same basis. The resources (facilities and manpower) necessary for proper interaction should be carefully evaluated and allocated since they are essential to the reduction of problems and workload, and to the improvement of overall service quality.
The initial forecast of a practically conflict-free 12 minutes of SAR per orbit was confirmed by experience, verifying that the decision to implement neither an expert system nor a special conflict resolution tool was correct. The absence of conflicts, and the variability of constraints and rules, make flexibility more important than plan optimisation, and favour mission planning based on natural rather than artificial intelligence (supported by powerful tools). A skilful mission planner can anticipate and resolve conflicts before they become critical, can judge new requirements against knowledge of the constraints, and can learn from the past and dynamically adapt procedures to the changing environment.
In conclusion, SAR mission planning has never been a limiting factor for the ERS mission. As has been shown, other factors have had a much greater overall impact.
ESA Bulletin Nr. 90.