Distinguished Guests, and Colleagues,
It is with much pleasure as Director of the SOPAC Division of the SPC that I present to you this morning the Keynote Address for this Conference, the largest of its kind in the region and which was set up to showcase new tools and concepts for improved data collection, capabilities and analysis in GIS and Remote Sensing. At the same time I would like to acknowledge Dr Jimmie Rodgers the Director General of the SPC who is unable to be here today and he gives his apologies.
Firstly, however it would be remiss of me if I did not take this opportunity to welcome all of you here this morning to the Opening of this Conference, and in particular I extend a welcome to all representatives of island governments and administrations, donor partner representatives, representatives of CROP organisations, educationalists, scientists and technologists from other stakeholder groups including NGOs. In particular I would like to highlight the many representatives from the private sector, including satellite data providers, image resellers, software and hardware companies. We have gathered here in the room a genuine mix of providers, developers and users.
Secondly, I would like to acknowledge that I have been privileged to address this conference over recent years and in that context I would like to acknowledge that having something different to say has never been difficult. GIS and Remote Sensing is one of the fastest developing technologies no matter whether you are a provider, and developer or a user.
On the global agenda the outcome of Rio+20 this year "The Future We Want" contains a particular paragraph of relevance:
274. We recognize the importance of space-technology-based data, in situ monitoring, and reliable geospatial information for sustainable development policy-making, programming and project operations.
And here in the region and for the benefit of Pacific island countries and territories we are all striving to stay at the "cutting edge of the technology." In this regard the theme of this Conference focusing on mapping Pacific resources is very timely.
For example, I recall a few years ago conferences were dominated by utility and disaster risk applications whereas this conference land cover and vegetation mapping stands out amongst the number of presentations. Indeed this reflects a reality where utility application of GIS has been stabilised and Pacific island countries and territories are now more concerned about the adverse impacts of climate change on development issues such as food and water security.
As a second example, LiDAR (light detection and ranging) datasets are now available for selected area in Tonga and Fiji and soon in parts of Papua New Guinea and Vanuatu. This data allow the extraction of the digital terrain model in coastal areas. During the conference it will be shown that LiDAR in western Viti Levu allows both to create the digital surface model, which is the top of the canopy, and the digital terrain model of the ground floor as the laser beams can penetrate the canopy.
Colleagues,
Let me assure you all that the SOPAC Division will continue to act as a centre of excellence central hub for GIS and remote sensing and will continue to work with all stakeholders to ensure optimum benefits accrue to Pacific island countries and territories. I will touch on some of our recent work shortly but at this time I would emphasise that we do store a safety copy of all image data sets as experience has shown all too often that data can be lost easily due to in-country local conditions which are not yet adequate for storage of digital images.
I am happy to note that other organisations increasingly recognise this central hub role that the Division plays. In doing so I would like to acknowledge our growing partnership with USAID and GIZ, and inform that FAO wants to establish at the Division a remote sensing based forest and vegetation monitoring centre.
May I now turn to some of the specific technical work that the SPC SOPAC Division has been engaged in during the past twelve months.
Vegetation and Land Cover Mapping
All the low lying atoll islands of Kiribati are now mapped at 1:10,000 scale with very high resolution image data. The image data can "see" individual coconut trees and all have been counted. The results of this work will be presented during the conference. With the support of GIZ the same resource mapping exercise is in progress for Tuvalu.
This type of resource assessment work is being carried out in the SPC and at the same time in- country with country counterparts who spend part of their time here in Suva. They return to their country after a few months to continue the work in their own department while other staff are sent to SOPAC.
Currently, in the SOPAC Division are 8 GIS officers from the Solomon Islands performing satellite interpretation and vegetation delineation under guidance of SOPAC specialists. Although carrying out “only” visual interpretation they are learning the complete handling of digital image data from pre-processing, image enhancement up to image analysis. At the same time SPC is supporting two GIS local consultants who are coordinating the mapping activities in Honiara and developing networking within the country and within SPC.
I am expecting other countries to follow including the Federated States of Micronesia, and Tonga.
Vegetation Mapping: Monitoring and Change Detection
It is essential to document the existing land cover. Subsequent monitoring will determine changes or indeed confirm that there are no changes. Key in this context is that without this baseline data the effects of climate change and the determination of necessary adaptation measures cannot be addressed.
Some Pacific island countries and territories are already covered at least twice with VHR image data and this work can be carried out. Some of these results will be presented during the conference. However, within three or five years changes are limited. There is much, much more work to be done to cover the region satisfactorily and develop historic change databases.
In Kiribati, Tuvalu and the Marshall Islands there are old maps showing the vegetation cover 40 years ago. Image analysis techniques allow exact superimposition of recent image data over the historical maps. Results of some of this work will also be presented during the conference. To date work indicates that the vegetation cover such as coconut palms but especially mangroves has increased on nearly all Kiribati atolls investigated so far. The reasons for this are yet to be determined, but please let no-one misunderstand these early results in the context of the impacts of climate change and sea level rise.
From Visual Interpretation to Semi-Automatic Analysis
Visual image interpretation is essential to learn better how to handle digital data and how to relate images to the land cover. However, visual image interpretation is becoming more challenging when for example the Worldview-2 satellite provides 4 additional spectral bands, as suddenly more differences within the vegetation are visible than before. At the same time this increases the interpretation time by a factor of four or five.
To assist SPC/SOPAC Division has purchased image segmentation software. This software is an add-on module to the “normal” image analysis ERDAS software we use. It takes both texture and spectral parameters into account from "test areas" selected by the operator. Then, the software segments the images from these selected "test areas" and delineates the different vegetation types. As a next step the operator has to either justify or correct these results. Once we are confident that this approach is working well, it will be trialled region wide. I am happy to see that this new development will be presented at the conference, and I look forward to hearing of the fruitful discussions I hope it will generate.
From Land Cover to Topography
For Pacific island countries and territories many projects are now emerging on issues such as climate change linked to food security. For example, a large part of the land cover mapping activities financed by a new USAID project being carried out by SPC requires maps in six countries Fiji, Vanuatu, Solomon Islands, Samoa, Tonga and Kiribati.
The adaptation option must address this issue; If climate change affects crops in a particular location, then an alternative location or alternative crops must be determined. The answer requires more than just the information about the current land cover. It is essential to know for every square metre of the landscape the height above sea level, the slope and the exposure to the sun.
Digital terrain models (DTMs) provide exactly this information, but there is a problem, most existing DTMs are at 1:50,000 scale and the necessary agricultural planning scale is 1:10,000.
The required DTMs can be created in different ways.
First there is LiDAR. LiDAR is the most sophisticated and expensive and examples will be shown in presentations during the conference. The problem for Pacific island countries and territories is that LiDAR costs between $500 and $1000 USD per km2 largely due to the high mobilisation costs.
As an alternative, it is possible to create DTMs from space borne images where DigitalGlobe has agreed to provide special pricing for Pacific island countries and territories through SPC/SOPAC Division of less than USD $20 per stereo pair. But a stereo pair is not a DTM. It requires analysis software, and special hardware such as a stereo computer screen, together with photogrammetric knowledge to convert a stereo pair into a DTM. Nevertheless, the SOPAC Division is currently investigating setting up this capacity. Once again I am pleased to note that this likely development will also be the subject of a presentation during the conference.
New image data and pre-processing requirements for Pacific conditions
Pan-sharpened image data is data which combines the lower resolution colour channels with the higher resolution black and white channel. The result is colour data with high resolution. The disadvantage is that this data cannot be atmospherically corrected, a correction which is essential for high volcanic Pacific islands. Clearly, a pre-processing requirement that takes topography into consideration is necessary. To overcome this problem image pre-processing is now carried out in house at the SOPAC Division, where the pan-sharpening process is done after the original data is atmospherically corrected.
When purchasing pan-sharpened image products “off the shelf” the user has to decide if they want normal colour combination or false colour infrared combination. SOPAC now purchases bundle products which is the delivery of all four colour bands together with the higher resolution black and white band and produces both combinations. This is available for the member countries for a price lower than one pan-sharpened product directly from the company. The Division now produces the following products out of one dataset:
• Pan-sharpened visual band combination
• Pan-sharpened false colour infrared combination
• Pan-sharpened false colour infrared red edge combination
• Vegetation index near infrared
• Vegetation index red edge
The interpreter then toggles between them for optimal vegetation type separation. This requires more time as mentioned earlier however, it improves the separation of vegetation cover. Presentations on image pre-processing and image interpretation will demonstrate the potential of the new data and also the new pre-processing in the SOPAC Division.
New Partnership between SOPAC and DigitalGlobe
Before I close and on behalf of the SPC-SOPAC Division and DigitalGlobe, I would like to invite you to a cocktail at the SOPAC Division to launch a new partnership. DigitalGlobe has agreed to provide very cost effective image data for Pacific island countries and territories through SPC. This partnership does not mean that SOPAC pushes the end-user to purchase DigitalGlobe products. On the contrary, we will still tender for the best possible data offer on a case-by-case basis which gets best product to address the needs of a country.
Colleagues,
In closing and wishing you all well for the week I want to acknowledge the old as well as many new faces attending this Pacific Regional GIS and Remote Sensing Conference. For the new faces, let me recall as I did last year and the year before that in 1993, almost 20 years ago, Philipp Muller the then Director of SOPAC wrote (and I quote) “Geographic information systems and remote sensing have become essential technologies that the SOPAC Secretariat is rapidly adopting…These development efforts although showing little immediate return will have tremendous benefits in the future”. Philipp was a visionary, and history has proved him correct. He would today be the first to acknowledge the unprecedented impact the growth in GIS and Remote Sensing technologies have had.
I appeal to all of you here attending this conference to uphold the spirit of Philipp’s vision and work together to progress ownership and sustainability that the opportunities GIS and Remote Sensing tools have to offer at the national level for Pacific island countries and territories.
Finally, as an acknowledgement of Philipp's vision may I go on record in suggesting that the organising committee of future conferences rename this keynote address as "The Phillip Muller Memorial Address" in recognition of the vision that Phillip had two decades ago.
Thank you.
