This article has been commissioned by the IOCCG and has appeared in the backscatter magazine, published by the Alliance of Marine Remote Sensing (AMRS)

Introduction
Imaging Spectrometers in the VIS/NIR-region are now an accepted tool in the investigation of the ocean-atmosphere system. They can provide many narrow spectral channels at medium spatial resolution and excellent co-registration of all spectral channels. This hyperspectral data gives a new and improved potential to transfer the measured data into spectral characteristics of the ocean and a very precise and quantitative retrieval of geo-/bio-physical properties of the water.

The German Aerospace Research Establishment (DLR) has developed a visible/near-infrared (VIS/NIR) imaging spectrometer named MOS (Modular Optoelectronic Scanner). Currently two MOS-instruments are operating in orbit: one on the PRIRODA/MIR station and the other on the Indian remote sensing satellite, IRS-P3.

The MOS Instrument
The (MOS) is a spaceborne imaging spectrometer for the VIS/NIR-spectral range. It is designed for remote sensing investigations of the Atmosphere-Ocean-System, especially coastal zones. The measurment philosophy follows the tried and tested concept of having separate spectrometer blocks for estimating the atmospheric turbidity and for measuring a set of spectrally dense data of the object signature. This concept has been tested on three missions (on Russian satellites and space stations Salyut 7/MIR) with non-imaging spectrometers.

Following this concept, the MOS-complex consists of two imaging spectrometers. MOS-A is designed for measurement of scattered atmospheric light (turbidity) in one window with 3 absorption channels of = 1.4 nm in the O₂A-band near 760 nm (Note: is wavelength). The ratios of this channel give a measure of the aerosol content in the atmosphere and the related aerosol optical thickness. MOS-B is the main block for measuring the target spectra with 13 channels in VIS/NIR-range between 400-1010 nm, with a halfwidth of = 10 nm.

The design criteria were dictated by the remote-sensing objectives and some technological reasons. The -channels were selected for ocean targets to minimize the influence of atmospheric gas absorption, even though it is difficult to avoid the influence of O₃- and water vapour. For water vapour we use two channels to estimate the columnar water content and to correlate it to aerosol influence on diffuse scattered light. Table 1 gives the figures together with other technical parameters.

In the design of the instrument, the main focus was on the spectral and radiometric resolution. The spatial resolution was of secondary importance and, with a pixel size of 500 - 700 m, gives an acceptable value for ocean targets. For physical reasons, the field of view can only have small angles. Both MOS-spectrometers reach ±7.0° across track.

The overall optical-mechanical design follows the low cost philosophy. In the conventional optical mounting, commercial lens-optics were used for entrance, collimator and imager and plane grating for spectral dispersion. A more sophisticated part is the specially made focal plane hybrid, which consists of a staircase mounting block for 13 CCD-lines to compensate the chromatic aberration of the lenses. The CCD's are specially made lines of 512 elements of size 23 x 480 µm², with electronic exposure control.

Both MOS-instruments on each mission can be calibrated using the sun as a reference - at PRIRODA via a transmission diffusor and at IRS-P3 via a diffuse SPECTRALON reflector.

Table 1b: Technical characteristics of MOS-IRS

Modular Optoelectronic Scanner MOS-IRS
Parameter	MOS-A	MOS-B	MOS-C
Spectral Range [nm]	755-768	408-1010	SWIR
No. of Channels	4	13	1
Wavelengths [nm]	756.6; 760.6; 763.5; 763.5; 766.3; O2A-band	408.0; 443.6; 484.6; 520.8; 570.5; 615.3; 650.3; 685.3; 749.7; 868.3; 1011.1; 814.1; 942.5; (H2O vapour)	1600
spectral halfwidth [nm]	1.4	10	100
FOV along track x [deg]	0.344	0.094	0.14
FOV across track [deg]	13.6	14.0	13.4
Swath Width [km]	195	200	192
No. of Pixels	140	384	299
Pixel Size x*y [km²]	1.57x1.4	0.52x0.52	0.52x0.64
Measuring Range Lmin..[µWcm-2nm-1sr-1]	0.1	0.2	0.5
Lmax [µWcm-2nm-1sr-1]	40	65	18

The PRIRODA mission is a Russian multisensor remote sensing mission on board a special module of the MIR station. The module was launched on April 23, 1996. Its payload consists of different active and passive sensing instruments in the UV-Visable -Thermal Infrared-Microwave range. It is in orbit at a height of ~ 400 km, at an inclination of 51,6°. The main goal of the PRIRODA programme is the development and testing of different remote-sensing methods and data interpretation algorithms by using measurements in different spectral regions. MOS-P plays the role of providing high dimensional multispectral data in the VIS/NIR range for ocean remote-sensing. The MOS-P instrument tests are a little delayed and are currently going on.

MOS on IRS-P3-mission

The satellite IRS-P3 is an Indian remote-sensing mission on a sun-synchronous orbit at 817 km and 10:30 am descending node crossing time. It was launched on March 21, 1996. The Indian Space Research Organization (ISRO) has installed a Wide Field Camera (WiFS) with 3 channels in NIR/SWIR range. Table 2 gives some specifications of the IRS-P3 satellite and of WiFS. MOS-IRS is the advanced version of the two instruments. This is true of the spatial and radiometric resolution as well as other technical details. To expand the spectral range MOS-IRS has been equipped with a 1 channel SWIR-camera (Fig. 1). The parameters of this channel are adopted to MOS-B, forming a "14th channel" for target data. The advanced MOS- IRS specifications are given in Table 1. The first image was received on March 23, 1996 and since this date, the instrument has been working well.

Figure 1
Modular Optoelectronic Scanner(MOS) - IRS complex.©DLR

There is no data recording device on board of the IRS-P3. Data are transmitted in real time to the ground stations Hyderabad in India and Neustrelitz in Germany. These ground stations cover the Indian subcontinent with the adjacent Arabian Sea and Gulf of Bengal and the whole European land region and coastal zones. Currently under discussion is an extension to third party ground stations in other regions of the world of high oceanographic interest. One of the first stations will be on Maspalomas/Canary Island. Other interests from South Africa, Australia, Argentina, Brazil are under negotiations. With the US/NASA a Memorandum of Understanding is in preparation.

Data policy

IRS-P3 is a bilateral scientific experimental mission of DLR and ISRO. Data acquired from that mission are available to both internal science teams for scientific applications within national and joint activities. As far as WiFS is concerned, for DLR, this is valid only over German territory, since WiFS data outside India are taken care of by special commercial contracts.

DLR is interested in a broad application of MOS data, at least in case studies for a wide variety of relevant remote sensing questions. However, since the resources of the DLR mission team are limited, a step-wise approach to include a broader community is considered. During the early phases of the mission, data from MOS- IRS has been available to selected scientific partners exclusively in the frames of joint activities such as ground truthing and calibration campaigns or joint studies for algorithm development. These "pilot" applications in the next step may serve as a nucleus for broader data access to the scientific community. Future joint work will also be discussed during a special workshop on MOS-IRS utilisation (April 28-30, 1997, Berlin).

For the MOS payload all health parameters such as temperature and power showed up normal; there was no problem in instrument operation. The overall functioning of the instrument and its performance in orbit is quite excellent. Because there were no problems with the instrument, the redundancy units were not tested for safety reasons.

Over the German ground station in Neustrelitz, data from roughly 3 orbits per day were received and processed to generate the Level-1B radiance data files. Up until the end of February 1997, the database contained about 600 passes. After reaching the nominal orbit, 26 sun calibration measurement sequences were recorded.

Internal mini-calibration lamps operating at different illumination levels of the focal planes are used to check the spectral alignment, the radiometric stability and the linearity of the CCD-lines. Usually such a control is performed at the beginning of each data take giving also actual measurements of the dark values for correction during ground processing. All received internal control data showed very good consistency compared to the pre-launch values; deviations are below 1.5%.

Sun calibration measurements are realized every 2 weeks during terminator crossing over the North pole using the external calibration unit of MOS by special command. The data are recorded in an internal memory of the MOS board computer and transmitted later in the orbit, when the satellite reaches the radio zone of Neustrelitz. The analysis of the data showed a good consistency and stability over time: deviations between recorded data sets are below 0.5%. Detailed analysis of the sun measurements are used to update the calibration of the MOS instrument.

MOS interpretation algorithm

Spectral high resolution measurements offer the chance to use the fine-structure of the spectral signatures of remote sensing objects to derive geophysical parameters quantitatively.

For the ocean colour applications of MOS at DLR, an algorithm was developed that makes use of the full information content of the spectral high resolution data to derive different water constituents and atmospheric turbidity from the VIS/NIR bands. Theoretical basics and the detailed approach used for this algorithm are described in Krawczyk et.al. 1995 and Neumann et.al. 1995. The idea is to use principal component analysis of modelled top-of-atmosphere radiance data to derive weighted coefficients for each measurement band. These coefficients represent the contribution of information of the corresponding wavelength to the estimate of a given parameter. Based on modelling and inversion for different geophysical situations (i.e. types of atmosphere, different dominant factors in the water, regional specific models for the inherent water-optical properties etc.) look- up-tables of coefficients are derived to select suitable sets for the interpretation of an actual scene. The selection process of the coefficients requires a classification scheme that uses a priori knowledge as well as parameters derived from the actual scene. Figure 2 shows the schematic approach.

The algorithm can be applied to MOS-B data to derive pigment concentrations (see the example from the Baltic Sea in the February issue of backscatter, pg 16), or to sediment distribution. Figure 3 illustrates the changes in sediment distribution in the Northern Adriatic sea. The May image was taken after heavy rainfalls causing the large plume in the mouth of the Po river. The water model was provided by JRC Ispra.

Krawczyk H., Neumann A., Walzel T., 1995: "Interpretation Potential of Marine Environments Multispectral Imagery", Proceedings of the Third Thematic Conference on Remote Sensing for Marine and Coastal Environments, Seattle, Sept. 1995, pp. II-57 - II-68

Neumann A., Krawczyk H., Walzel T., 1995: ,A Complex Approach to Quantitative Interpretation of Spectral High Resolution Imagery", Proceedings of the Third Thematic Conference on Remote Sensing for Marine and Coastal Environments, Seattle, Sept. 1995, pp. II-641 - II-652, 1995.