Satellite Image Analysis Conducted for a Project In Vietnam
“I was honored to be selected to represent GISCorps on this noble project. It was exciting to see how the change detection results validated what they were seeing in the field.”
Jason San Souci, GISCorps Volunteer
In December 2009, CartONG; a French based non profit organization requested the assistance of a GISCorps volunteer with expertise in remote sensing. The overall goal of the project was to support the Vietnamese government to find a mechanism to improve the livelihood of the poor by increasing the value of their forests through a pilot study in two regions in the Central Highlands of Vietnam.
Following a series of analysis and field work, CartONG and their team of collaborators had learned of inconsistencies in the carbon content levels that were measured in 2004 and 2009. In order to verify that information they decided to request a remote sensing specialist’s assistance to analyze Spot satellite imageries from those two years.
Shortly after the request came in an email broadcast was sent to the volunteer database and 15 individuals from 10 countries applied for the project. The review of the resumes and interviews resulted in selecting Jason San Souci, GISP from AFE Advisors LLC. in Colorado and he began working with the CartONG team shortly there after. He recently completed the analysis and contributed in a report which can be accessed in its entirety from HERE.
The following is a summary of that report.
When comparing fieldwork results from 2004 and 2009 on determining carbon content of one village in the Vietnamese Highlands, the carbon ranges differed enormously, suggesting that either the methods of fieldwork have changed or that the carbon stock was depleted through deforestation and degradation. However, analyzing SPOT satellite images from 2004 and 2009 on carbon content, verified the findings of the fieldwork and demonstrated that the carbon stock had declined through both, deforestation and degradation.
The Need for Satellite Imagery Analysis
After the fieldwork started and was completed in one of the villages; it was found that whereas the newly collected data would give a good overview and provide a good base for the Carbon stocks model projecting future carbon contents under different forest management scenarios; the data on carbon content which could be obtained through past forest inventories was fragmented and verification was needed. Especially the range on carbon content was reported to be much higher in 2004 than 2009. Whereas this was to be expected, the differences were striking and raised questions.
The decision to verify these inconsistencies was taken at a late state of the project and there was no budget available to pay a specialist for the analysis. Hence, CartONG asked if GISCorps could provide assistance. Fortunately, GISCorps identified Jason San Souci a registered volunteer who not only had many years experience of working with satellite images, but also had worked on carbon content estimation before.
• Converting the Digital Number values to at-satellite reflectance; some input was given by the SPOT technical team
• Image registration for change detection
• Radiometric correction and Normalized Difference Vegetation Index (NDVI) layer generation
• Overlaying the plot coordinates of 2009; extracting the NDVI values.
• Computing the average Carbon content based on the extracted NDVI values
• Applying the average Carbon content on Thon 4 and 6 as well as on the full images
• Accuracy Assessment of Carbon content results with field data
When comparing the overall computed results it becomes evident that the trend identified during the fieldwork with a strong deforestation and degradation since 2004 can be confirmed through the Satellite Image Analysis.
The Carbon range in the satellite images is slightly below the carbon range determined through fieldwork in 2004 and 2009. This can be explained through the fact that trees with a bigger diameter will store more Carbon, which the mixed pixel value of the image can’t reflect to the same extent (Figure 1).
On the left hand side graph Amount of Carbon shows the Carbon range prevailing in 2004 in Thon (village) 6; as can be seen, none of the forested areas are below 751 Tones/Ha. On the right hand side the situation in 2009 is depicted. Here none of the forested areas contains more than 750 Tones/Ha. This trend is reflected in the results of the satellite image analysis as well. In 2004, the lowest carbon value is ranging between 501 – 750 Tones/Ha; in 2009 the highest value is not exceeding 500 Tones/Ha. The boundaries are not as defined which is linked to the fact that the boundaries shown in the plot analysis have been obtained by interpolating (kriging) – Figure 2.
In order to get an overview and make the user of the data and the data model understand how important Sustainable Forest Management is, degradation of the forest was categorized and displayed. Degradation is defined as Percentage Change (excluding deforested areas) from 2004 to 2009. As can be seen in the graph Percentage Change of Thon 6; most areas experienced degradation of more than 50%; with the percentage over 81% being strikingly high (Figure 3).
When roughly estimating degradation for Thon 6 from the results of the satellite analysis; the following figures were established:
For questions regarding this project, contact Jason San Souci at: firstname.lastname@example.org and CartONG’s Sandra Sudhoff at: email@example.com.