How can images and big data help protect them?

With more than 65,000 hectares of vegetation gone up in smoke, the summer of 2022 will have been a summer of fires and alongside 2003 and 2019 is one of the worst in decades.

These fires, which are no longer limited to the Mediterranean region, cause destruction, economic and ecological losses with their share of casualties among the population and firefighters. Repeated climatic influences, which also lead to a progressive deterioration in the health of the trees.

Diagnostic tools have been developed over the past century to closely monitor the condition of forests. Today, advances in imaging offer immense potential.

Fragile forests, the slow realization

From the beginning of the 20th^e In the 19th century, foresters noticed the emergence of new pathogens in trees – oak powdery mildew, chestnut canker, beetles… But it was not until 1958 that the National Forest Inventory was created, an institute designed to collect statistics on forests.

It was not until the 1960s that the first observatories and then the first monitoring of the state of forests in large cities were established. The drought of 1976, the oak dieback in the Tronçais forest or even the decline of conifers in the East led to the creation of a Forest Health Department (DSF) within the Ministry of Agriculture in 1989. Implementation of monitoring of the condition of around 12,000 trees in 600 fixed plots, systematically installed every 16 km throughout metropolitan France.

Each of these trees is examined with a magnifying glass to identify, on the one hand, the mortality of the branches for all species and, on the other hand, the lack of branching for the Hardwoods and the needle deficit in softwoods. The observations made since the introduction of this system and on non-permanent plots paint a worrying picture of the health of our forests. If less than 1% of the monitored trees die each year, the Mortality has been increasing steadily since 2010.

[Près de 80 000 lecteurs font confiance à la newsletter de The Conversation pour mieux comprendre les grands enjeux du monde. Abonnez-vous aujourd’hui]

In the crown, i.e. the uppermost part of the trees, the observation is even clearer: leaf loss – known as leaf deficit – has increased steadily since measurements began in 1997. For mortality and leaf deficit, practically all tree species are affected.

the “forest dieback” is generated when this degradation phenomenon affects an entire stand or even massifs. In general, several environmental factors are involved, which may follow one another or act together.

Vicious circle and race against time

A distinction is made between predisposing factors (soil conditions, identity and genetic origin of the trees present, population density and/or considered), triggering factors (drought events and severe or repeated heat waves, consecutive). defoliation after phytophagous attacks) leading to a loss of stand vitality, and finally aggravating factors (pests of aerial parts or roots) contributing to the death of weakened trees.

These long-known phenomena are now manifesting themselves with unprecedented intensity and magnitude. However, the models do not inspire optimism: the climate will continue to deteriorate, undermining the ability of many tree species to sustain themselves and adjust their distribution naturally.

This threatens the diverse services of the forest: A dying forest stores less and less carbon and becomes more sensitive to fire.

The carbon is released in the form of CO₂ in the atmosphere during fires, this increases the intensity, duration and frequency of droughts and heat waves and thus the risk of fire. A vicious circle, to which is added an increased risk of storms and pest infestations, favored by climate change.

Dying trees lose their economic value very quickly, recognizing the phenomenon quickly is crucial in order to save what can be saved and harvest the timber.

The forest seen from the sky

It is also essential understand better the relative role of predisposing, aggravating, and triggering factors to suggest other forest management practices. On-site observations, such as DSF monitoring, are necessary in this regard to alert and understand the phenomenon of dying. But they don’t cover the whole area.

The development of satellites opens up new horizons. The first dedicated to Earth observation, Landsat-1, was launched on July 23, 1972 by the United States. France acquired SPOT (Earth Observation Probatory System) satellites beginning in 1985. Although they provide high-resolution imagery, the prohibitive cost of image processing limits their use.

The images are not only in the visible range, but also in the infrared, because this spectral range provides precise information about the inner structure of leaves (near infrared) or about the moisture in soil and plants (mid-infrared) and enables hardwoods from softwoods to be healthy Correctly distinguish foliage from old foliage.

Today there are a number of metrics that take into account different spectral bands for a given pixel in their calculation. The best known of these is undoubtedly the NDVI (normalized difference vegetative index) based on the reflectance of the red (R) and near infrared (NIR) channels.

Dying off in the age of big data

Today, with great strides, the SPOT satellites have given way to the European Sentinel satellites of the European Copernicus program. The resolution of the images on the ground is at least 10m×10m and the devices carrying optical and/or radar sensors return to the same locations every five days. As the icing on the cake, the images are free.

Research into the exploitation of these images began a long time ago. And the dramatic increase in computing power to process them offers the possibility of large-scale, high-resolution surveys that were unimaginable a decade ago.

In the field of forest remote sensing, research stakeholders (ONF, CNPF, DSF, IGN, CNES, INRAE, University of Orléans, among others) have joined forces at the Center for Scientific Expertise, Change and Health of Temperate Forests.

spruce colors

So they developed a computer routine (ForDead, Dutrieux et al., 2021) to quantify mortality in spruce forests using sentinel images.

In fact, France has experienced massive attacks on spruce trees since 2018, first in the Grand Est, then in almost the entire northern half of France in the following years. The Blame of the Typographer, a half-centimeter beetle capable of killing spruce trees during their pullulations by attacking the tree’s conductive vessels. It is estimated that this crisiswhich is just beginning to falter has affected some 55,000 ha, requiring the urgent harvest of 19 million m³ of spruces in France.

A healthy spruce is dark green, a freshly dead spruce takes on a diagnostic orange color with a tendency towards gray when the needles fall. It is this radical change in crown color as measured by various vegetation metrics that ForDead exploits.

This processing chain, which has proven itself with spruce, is currently being evaluated to monitor the health status of other species. This is especially true for oak groves.

Hecatomb of Pedunculate Oaks

Massive diebacks of English oaks and, to a lesser extent, Sessile oaks are observed in several large lowland forests such as Chantilly and Vierzon. The first studies showed that remote sensing made it possible to clearly identify the healthy and most affected stocks, but struggled to correctly predict mean infestation rates. The exercise is more delicate than with Fichte.

In fact, oak dieback is a slow process that can take several decades before resulting in tree death, with oak trees even recovering in good years. The color change is also less spectacular than with spruce. Finally, they are often mixed with other species, with an undergrowth and an undergrowth that, seen from the air, can mask the deterioration in the health of the oaks.

Studies are now geared towards identifying the most relevant metrics or the use of artificial intelligence, whose interest in the exploitation of time series of images no longer needs to be proven. This is what the Sycamore Research Program. The foresters’ expectations are high and the scientific and technical venture is on the verge of victory. So that our forests don’t go up in dust or smoke.

---

This article is published as part of the Fête de la Science (to be held October 7-17, 2022 in mainland France and November 10-27, 2022 overseas and internationally) in which The Conversation France participates. This new edition will have the theme “Climate Change”. Find all events in your region on the website Fetedelascience.fr.