conservation on the radar

Finally, my blog on radar remote sensing and forest conservation at our FFI website is up!

sneak peek: 1st FFI blog post

After the successful training workshop on SAR image processing that we conducted last week, I was inspired to write and submit my first blog to our website at Fauna & Flora International. Aldrin tells me he's already forwarded it to the blog administrator but I'm still waiting for confirmation on when it will be posted online (plus I'd probably have to create an FFI blog account too). While that blog is undergoing approval, here's a sneak peek of that article.

----------------------------

Building capacity on radar remote sensing for Philippine forest conservation

When we hear the term “radar” an image of a circular screen showing blips that detect the presence of aircrafts or ships is usually the first thing that comes to mind. We also think of radar instruments that enable us to forecast the weather, or the speed gun that police use to tell whether a vehicle is moving within or past the speed limit. But the kind of radar I’m interested in, particularly in my work at FFI’s Philippines Programme, involves synthetic aperture radar—a technology which is mainly useful for observing and monitoring changes on the Earth’s surface.

Some noteworthy examples demonstrating the effectiveness of synthetic aperture radar (or SAR) technology include the Global Rain Forest Mapping Project in 1995, which generated semi-continental satellite image datasets for monitoring forests within the tropical region of the Earth. Another previous effort is the Shuttle Radar Topography Mission in 2000, which successfully produced the first high-resolution topographic map of the Earth’s land surface at a near-global scale.

SAR has been widely used for Earth observation. One of its main advantages over optical satellite images (such as those Earth images you normally see in Google Earth) is that it is unaffected by cloud cover, meaning it can “see through” clouds thereby providing complete cloud-free coverage of the surface of the Earth. It has also been proven to be effective for mapping and monitoring changes in the world’s forests. In fact, past and current researches have even shown that SAR can be employed for measuring forest biomass up to a certain extent. This makes SAR very useful for FFI’s global initiative on Reducing Emissions from Deforestation and Forest Degradation (REDD).

2008 ALOS/PALSAR image of Luzon Island, Philippines.

In the Philippines, FFI is currently working in close partnership with Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH and the Department of Geodetic Engineering of the University of the Philippines (UPDGE) to explore remote sensing approaches in support of developing the Philippines’ national REDD+ Monitoring, Reporting, and Verification (MRV) system.

One of the fruits of our partnership was a jointly developed proposal that we submitted to Japan Aerospace Exploration Agency (JAXA) through the ALOS Kyoto & Carbon (K&C) Initiative to gain access to Japan’s ALOS/PALSAR radar imageries, and was fortunately approved in the 3^rd quarter of last year. This is the first time, as far as I know, that FFI will be collaborating with the K&C Initiative, an international remote sensing science community, which will look into the application of SAR technology for forest conservation, particularly for our case.

Through this collaboration with JAXA and the K&C Initiative, we will not only be able to access ALOS/PALSAR data, but also learn from the world’s foremost radar remote sensing scientists on utilizing SAR for conducting historical analysis of forest changes and for assessing forest carbon stock within our sub-national REDD+ pilot areas. In return, we shall then be supporting JAXA’s global forest mapping effort by sharing our field data to validate the forest/non-forest maps generated from the ALOS/PALSAR images.

Recognizing the potential of the application of SAR technology for REDD+ MRV development, FFI together with GIZ and UPDGE also successfully initiated and co-facilitated a 3-day training workshop on SAR image processing, which was recently concluded last week. The Philippine Tropical Forest Conservation Foundation, a local funding organization committed to conserving Philippine forests, also generously supported the activity. The training intended to build the capacity of local institutions on the use of SAR technology for forest applications and REDD+ MRV development.

And so it was really exciting to see a room full of people with almost 35 participants from various institutions across the country including government bureaus, NGOs, universities, research institutions, and energy corporations, who were interested to learn and develop their skills. Some of the initial efforts by different organizations on using SAR data for forest applications were shared and presented. The activity also provided a good venue to discuss the gaps, challenges, and ways forward within the country’s forestry sector; it was a fantastic opportunity as well for different organizations to develop collaborative research ideas on SAR applications. It is my hope that the next time this group convenes we will be sharing results, experiences, and lessons from each of our efforts.

Participants and resource persons during the SAR image processing training workshop at UP Diliman campus, Quezon City, Philippines conducted on 8-10 February 2012 (Photo by Erika Valdueza).

Our next activity: this summer (referring to the hot and humid summer in the Philippines), we will be getting our hands dirty by conducting the first wave of forest carbon inventories and biodiversity assessments in Gen. Nakar in Quezon site through FFI’s EU REDD programme, as well as in Victoria-Anepahan mountains in Palawan (through the EU-funded project of our partner organization, the Non-Timber Forest Products—Task Force). The initial field data that we gather from the forest plots will be used together with the ALOS/PALSAR radar data to generate preliminary baseline results in support of our REDD+ projects.

a thesis proposal, at last

Earlier today, Eric and I spent almost three hours discussing my proposed thesis at the AGST lab. But I finally got together my thesis proposal, which I will hopefully be presenting this March. My proposed thesis is entitled, "Discriminating Philippine forest formations using object-based analysis of radar backscatter and texture analysis of ALOS/PALSAR mosaics." The title sounds much simpler than it really is. 

The primary objective of this study is to assess the performance of SAR image data, specifically ALOS/PALSAR image data (25m; L-band; HH,HV polarization) provided by the JAXA ALOS Kyoto & Carbon Initiative, for discriminating and mapping different forest formations of the Philippines (as described by Fernando et al. (2008), following the scheme and nomenclature of Whitmore (1984)). My study will also be specifically aiming to achieve the following:

[1] To investigate the viability and accuracy of using SAR to distinguish between Philippine forest formations;

[2] To develop a replicable approach using SAR image analysis for mapping forest formations;

[3] To identify the best correlated forest stand parameters to SAR image data applicable to tropical forest setting; and

[4] To evaluate multi-spectral data sensitivity in discriminating forest types based on adopted national forest classification schemes.

To achieve this, I will be implementing textural analysis, modeling and regression analysis of radar backscatter, and object-based image classification techniques on ALOS/PALSAR data. For the last objective, Eric also recommended that I include an evaluation of multi-spectral data, such as Landsat, to assess its sensitivity in discriminating forest types based on the adopted national forest classification schemes, which hopes to settle once and for all how far multi-spectral data can be used for forest type mapping.

I believe my proposed study is significant in the following respects, particularly: 

[1] The Philippines has yet to come up with a map that reflects the true diversity and spatial distribution of Philippine forest formations based on physical and vegetation structural characteristics. Having such a map can improve our appreciation and comprehension of Philippine rainforests, and aid in the formulation of effective forest management and conservation policy.

[2] My study may also contribute to the national forest mapping and classification work of NAMRIA and FMB by providing a different approach for mapping the spatial distribution of the country’s various forest formations using SAR data. An evaluation of multi-spectral data sensitivity in discriminating forest types based on previously adopted national forest classification schemes can also establish how far multi-spectral data can be used for mapping Philippine forest formations.

[3] A review by Langenberger (2004) indicates that Philippine rainforests are poorly represented in research compared to other tropical rainforest areas. My study may not only contribute to the existing body of knowledge that demonstrates the utility of SAR for forest applications, but also to gain a better understanding of complex tropical forest ecosystems.

[4] Finally, the approach that I intend to develop for mapping forest formations may also be applied for stratifying forests based on parameters shaping vegetation structure in support of estimating above-ground biomass, which is relevant to current initiatives on Reducing Emissions from Deforestation and Forest Degradation (REDD+).

My study will look into three sites in the Philippines, namely: 1) Victoria-Anepahan mountain range, Palawan; 2) Southern Sierra Madre mountains; and 3) Southern Leyte province. This is actually and conveniently aligned with the REDD+ pilot projects at Fauna & Flora International (where I am presently affiliated) and our partners; and the work in Southern Leyte where Kix is currently engaged in with Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH.

These three sites exhibit distinct physical geographies (e.g., elevation, rainfall, climate, geology), and could represent different forest formations that are the focus of this study. These sites also correspond to REDD+ pilot areas designated by Philippine government where on-going demonstration activities are being conducted.

My next task is to start image pre-processing work on the Landsat data and the ALOS/PALSAR mosaics. I'm also planning to apply for a grant fellowship through the International Tropical Timber Organization (ITTO).

Wish me luck.

today's RG meeting

I was not able to present my revised thesis at the RG meeting today since there were several members who discussed their progress reports, and also a few presenting their prospective topics as well. The discussions were nevertheless engaging. I was, in fact, really interested in the ideas of fellow members, particularly: Cristy's work on using MODIS data for understanding ENSO events at a watershed level; Melanie's application of a lightweight LiDAR instrument (or so I think) for assessing water inundation in Marikina's river systems; and Erika's idea on developing the first-ever landslide inventory for the Philippines. I think I'm also starting to see the formation of the forest conservation subgroup (which includes me, Kix, and Anji), in addition to the already existing energy engineering subgroup. My one-on-one discussion with Eric is rescheduled to Saturday morning after my trip to Polillo.

texture, and then OBIA

I presented the first draft of my prospective thesis topic on mapping the Philippine forest formations by relating forest stand parameters to synthetic aperture radar (SAR) data at our research group meeting at the AGST lab earlier this evening. I think Eric was only somewhat pleased with the idea while I was explaining (like there was nothing really novel with the concept), but later on he gave very constructive inputs and comments to really improve the idea. The most useful take-home messages, in addition to conveying my thoughts more objectively and better expressing the significance of my study, were the concepts on texture analysis of SAR data (which I strongly believe can show me the way to mapping limestone forests, among others), and object-based information analysis (OBIA, which complements the modeling of radar backscatter and can even further improve the accuracy of my proposed method). It seems I will be crafting an improved draft and methodology very soon, thanks to Eric. So far, so good.