Skip to content

A shrewder glance from space

What is the extent of real loss to the Russian forest during fire season? In recent years, the issue has been causing permanent conflicts between environmentalists and representatives of public authorities at both federal and regional levels. ScanEx RDC company believes that it is possible to accurately count our losses already today.

The main problem is that the data of environmentalists and government officials would not match.  Moreover, this data differs a lot – in some region severalfold. Environmentalists keep telling that forest fires cause ten times more damage to nature than it can be judged according to official statements. The increasing serious distrust to official information, which is referred to by forest agencies, gave birth to space images. Thanks to the global Internet, and technical achievements of Americans in the expensive exploration of space, images of our planet became publicly available today. Environmentalists affirm that satellite images taken by U.S. satellites indicate that there is a catastrophic understatement of forest fires areas in Russia. In response, the authorities allude to the lack of professionalism of environmentalists, far from the art of satellite images interpretation.

To understand the problem, we decided to address to the ScanEx Research & Development Center, where Earth remote sensing data is processed in a professional manner . It is difficult to accuse of bias the independent commercial entity – ScanEx company, which announced democratization of remote sensing data as its primary mission.
Head of the Operational Monitoring Department Artem Nikitsky patiently explained me all the nuances, which are constantly faced by the experts involved in the analysis of data coming from space.

How long ScanEx deals with the problem of forest fires?

Since 2002, approximately. We were partially working with this even earlier. Most information about fires comes from two U.S. satellites – Terra and Aqua, which take low spatial resolution images. This information is available worldwide. But if we have an interest in some specific area, the error in determining the fire area can be significant because of the low resolution of images. Images are acquired  frequently – twice a day by each of the satellites. In other words, any spot on the planet can be analyzed in terms of fire four times per day. Of course, if there are no clouds. Prior to the launch of the first Terra satellite in 1999, the images of the Earth were taken by weather satellites with a lower sensitivity and a poor ability to determine the size of forest fires. A new page in the forest fire monitoring was opened thanks to the active application of Terra and Aqua satellites. It is important that the survey and data from satellites are transmitted  in direct broadcasting mode. If you have a ground station, you can receive the data free of charge. Another way – to download them from the Internet, where they are also available for free public access.

Do all thermal spots on satellite images indicate fires? In other words, are there problems with decoding data from satellites?

Of course, there are problems. We must remember that the satellite provides information covering a huge area. The detection is carried out by the images, one pixel corresponds to one kilometer. When a thermal spot is detected we can confirm that there is a fire site within this area. But we cannot determine neither its size nor location with a 100% percent accuracy. If, for example, we have a fire within 10 pixels, it is clear that, most of all, this whole area is on fire. When we face a small fire, then we cannot determine its size and strength for sure. In other words, we can determine with a reasonable degree of accuracy the fires that cover a really large area. But the error may reach about 30 percent due to the ragged outline of a fire. Illustration 1: Thermal spots in the “Kosmosnimki – fires” interface

Another thing, affecting the accuracy of the area detection: we can have only one edge burning, instead of the whole area. Consequently, we will see a set of thermal spots. If we ignore the burnt area, which is within this edge, the area of the fire will dramatically reduce. This is a question of who counts fires and how. There is a method, according to which only the burning edge is taken into account, but what has already burned out is not considered. Accordingly, long-term fire that has spread on a large area, we will get a multifold understatement of burnt areas.

Can a campfire be interpreted as fire?

No, we do not see a campfire. A satellite can detect small ground fire under the forest canopy, by rough estimates, ten to one hundred meters. If a fire is on the areas one hundred by one hundred meters, then the satellite will surely recognizes it. Crown fire is determined, even if it is 20 to 20 meters. Theoretically, there is a danger to confuse the fire with burning flares , which are mounted on the oil rigs. But we have data on most of  the stationary flare stacks, so we ignore such thermal spots. A strongly heated Earth surface can be misleading in our interpretation. For example, the highly-heated 1×1 km black arable land spot can be taken for fire. Similar errors can be with the peats. Glare on the shallow ponds or overheated sandy surface can also be misleading for the automatic equipment. Clouds may also become a major obstacle to obtaining reliable information.

Imagine small ragged clouds within a few kilometers. The fact is that the edges of the cloud can produce the effect of burning surface, but actually there is no fire down there. The fire detection algorithms provide a mechanism for elimination of error sites, but, unfortunately, it is not 100 percent accurate as well.

It’s all part of the 30-percent error?

Good question. The problem is that we have a weak feedback. Ideally, the data obtained from space, should be verified by area ground survey. In this case we would have an impartial picture of how much we can trust the satellite images. But we are not able to organize such a large-scale data verification, whereas the regions do not hurry to report the results of their surveys. We are working very closely with the Emergency Ministry, for example, in such issues as peat fires. Technology is as follows: we are sending information about a potential fire, it is being forwarded to the region, but we very rarely get any feedback on whether it was a real fire or it was a mistake.

In other words, technically you cannot double-check the information received from the satellites?

That’s right. When huge forested areas are burning in Yakutia, it is obvious that it is a fire without any field surveys. The error may be in identification of a specific area, but the fact that it is a fire indeed leaves no doubts. Another thing, when we face with a stand-along small fire, especially in the European part of the country.

In recent years the debates about real areas of forest fires intensified. Ecologists blame the authorities for a substantial, sometimes multi-fold, understatement of the area of fire. What do you feel about this kind of accusations? What is the probability that they really conceal the actual scales of the disaster from us?

I am inclined to think that, of course, much more is actually burning than it is reflected in official statistics. It’s difficult to say what is the realistic extent of underreporting. To do so, we need to count in details the actually burnt area with respect to each and every fire. We have been doing the studying of burned area in the Republic of Komi. We found that as a result of the analysis of higher resolution images the area covered by fires is by 35-40% more than what was initially detected. As to how local foresters were reporting, I can only guess. There is another nuance. The official statistics of forest fires include only data on forest fund lands. But the fire can spread to forest areas, which are not officially regarded as forest fund lands. It can be, for example, overgrown agricultural land. To get a more impartial picture one needs to have an accurate map of forest fund lands and overlay it with the map, which shows the burnt area. There is another fundamental question of how to account for the huge fires in forest areas within the area of satellite monitoring. We must understand that those are difficult-to-reach forests, where there is no infrastructure, no roads, no towns, no social facilities. Naturally, no one extinguishes fires over there. I think that it is the fires within the area of space monitoring that give main differences between the data of environmentalists and the official statistics.

How does the space, from your point of view, can help in obtaining an impartial, independent on the desire or unwillingness of authorities, information? Is there a mechanism that allows you to bring our “lame” on both legs statistics to reality?

These satellite images should certainly be verified. Most operators launch satellites to take images in the visible and near infrared bands. Such images give no unbiased information about the fires. A special shortwave middle infrared band is required. Those images that are now acquired by satellites are best suited for the determination of the burned areas. But they are no good for high-quality real-time monitoring of the fire situation.

Illustration 2: UK-DMC2 image of fires burning in California © DMCii, 2013. Smoke is clearly visible rising from the dark burnscar caused by the fires.

It looks like we have a great opportunity to obtain reliable information about the real damage done to our forests? Is anyone using this chance?

Periodically we receive requests of this kind from the regions. However, they do not always reach the implementation stage. Sometimes in the regions come to the conclusion that it is cheaper to calculate the actual damage applying  ground survey. But we have executed some of such orders. Science and space technology is developing rapidly. I am convinced that it will become increasingly difficult with every new year to withhold information on the areas of forest fires. This is an objective process that cannot be stopped. By the way, a lot of issues of satellite images application, including fire situation monitoring will be discussed at the International Conference Earth from Space:the Most Effective Solutions to be held in October 1-3 in Moscow Region. We are pleased to invite the editors and readers of the “Russian forest news” to take part in it.


Illustration 1: Thermal spots in the “Kosmosnimki – fires” interface
Illustration 2: UK-DMC2 image of fires burning in California © DMCii, 2013.
Smoke is clearly visible rising from the dark burnscar caused by the fires.