{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Introduction\n", "\n", "In recent years we hear a lot of news about wildfires occurring in areas where they did not occur often in past. In spring and summer 2018 we have witnessed the boreal forests in Siberia burning at extraordinary rates (https://www.sciencealert.com/nasa-images-capture-worst-siberian-wildfires-in-10-000-years). Other areas are known to be prone to wildfires but the wildfires tends to be bigger and more severe. Wildfires, which erupted in Portugal in summer 2017 took 66 lives and caused an economic loss of more than 565 millions USD (http://thoughtleadership.aonbenfield.com/Documents/20180124-ab-if-annual-companion-volume.pdf). In September 2017 California, US was in flames, firefighters were fighting with more than 9000 fires, which burned more than 500 ha (https://www.fire.ca.gov/incidents/2017/).\n", "\n", "Wildfires have many negative consequences:\n", "- They can **damage** human properties.\n", "- They present a **threat for human health and lives**. In addition to the direct threat from burning, wildfires also release pollutants detrimental for human health and ecosystems. Close to the fires, smoke is a health risk because it contains a mixture of hazardous gases and small particles that can irritate the eyes and respiratory system (http://www.xinhuanet.com/english/2018-07/27/c_137352746.htm).\n", "- Vegetation fires release large amounts of **particulate matter** and toxic gases including **carbon monoxide**, **nitrogen oxides**, and **non-methane organic compounds** into the atmosphere (https://public.wmo.int/en/media/news/drought-and-heat-exacerbate-wildfires). This contribute significantly to global warming.\n", "- Extinguishing of the fires is **risky and costly**.\n", "\n", "However, wildfires can also be beneficial. High-severity wildfire create complex early seral forest habitat (also called “snag forest habitat”), which often has higher species richness and diversity than in unburned old forest. E.g., giant sequoias, found in the U.S. Sierra Nevada, require heat from fire to regenerate (https://video.nationalgeographic.com/video/yosemite-sequoias-fire).\n", "\n", "While we cannot - and do no aim to - fully prevent wildfires it is important that we try to understand under what circumstances wildfires are most likely to appear, how they spread and what impact they have so that we can minimize their negative effects. Usually we try to minimize the amount of flammable material either by igniting smaller fires (controlled burning) or by logging.\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Wildfires and satellite imagery\n", "\n", "Satellite images are images acquired with sensors (\"cameras\") carried on board of satellites. Images are taken from different satellites and in different parts of electromagnetic spectrum. Such images give us various information about the Earth. They are a good source of different type of information about wildfires which can help us build our understanding of wildfires and plan (re)action to their occurrence accordingly. Advances in satellite technology has made it possible to monitor wildfire activity better than in the past (https://public.wmo.int/en/media/news/drought-and-heat-exacerbate-wildfires). Collecting the information about the fire on a field can be dangerous, while satellites enable us to get such information from (more than) safe distance. It also has the advantages of covering larger areas, gathering data on less accessible areas (Leblon et al., 2012), it is time and cost effective.\n", "\n", "Satellite imagery is, in connection to wildfires, commonly used to:\n", "\n", "1. **Detect areas with high potential for wildfire occurrence.** \n", "This is estimated based on vegetation maps, material (potential fuel) moisture maps both derived from satellite images. These maps are combined with other combination such as weather information and topography and distance from roads and settlements to produce fire risk maps (http://www.isprs.org/proceedings/XXXV/congress/yf/papers/927.pdf).\n", "2. **Map areas with potential fires:**\n", "Check the world map with potential fires from the latest satellite images:\n", "https://fires.globalforestwatch.org/map/#activeLayers=viirsFires%2CactiveFires%2CfireStories%2Ctwitter&activeBasemap=topo&activeImagery=&planetCategory=PLANET-MONTHLY&planetPeriod=Aug%202018&x=-12&y=18&z=3\n", "3. **Observe extent and severity of burned scar** (as we will demonstrate below)\n", "4. **Observe impact on vegetation and its recovery** (as we will demonstrate below)\n", "5. **Observe impact on build areas.** \n", "As an example, check how satellite images can be used to observe the burned villages in Africa: https://www.bellingcat.com/resources/how-tos/2018/09/04/identify-burnt-villages-satellite-imagery%e2%80%8a-case-studies-california-nigeria-myanmar/\n", "6. **Observe spread of smoke and gases.**\n", "Check out how spread of smoke in case of wildfire sin California was captured from MODIS satellite https://www.nbcsandiego.com/news/national-international/NASA-Satellite-Images-California-Wildfire-Brush-Fires-Images-Photos-433867203.html\n", "\n", "We will now check how some of the wildfires and their consequences were seen from space. We will use images acquired from satellite Sentinel 2 to observe the consequences of wildfire at Madeira in summer 2016 and images acquired from satellites Sentinel 2 and Sentinel S5p to observe wildfires in Siberia.\n" ] }, { "cell_type": "markdown", "metadata": { "collapsed": true }, "source": [ "# Madeira, August 2016\n", "\n", "Madeira is a Portuguese island in Atlantic Ocean well known for its vivid vegetation and beautiful nature. In August 2016 flames of deadly fire spread throughout the region of Southern Madeira and to its capital Funchal. More than 200 houses were destroyed, vegetation - including botanical garden near the capital - was severly damaged, 4 people died (https://www.madeiraislandnews.com/2016/08/fire-damage.html, 6.9.2018).\n", "\n", "Let’s check how the consequences of the fire were seen from Sentinel 2 satellite:\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\" \"Drawing\"
Before wildfire, true color image acquired from Sentinel 2 on 7.8.2016 (EO Browser link). After wildfire, true color image acquired from Sentinel 2 on 17.8.2016 (EO Browser link).
\n", "\n", "Two bigger burn scars can be observed on southern part of the island. It is difficult to distinguish them from surrounding un-burned areas, though. To make these areas easier to detect we will visualize images of the same area on the same dates acquired in near-infrared and short-wave infrared part of the spectrum. Healthy vegetation has a high reflectance in the near-infrared portion of the spectrum (NIR), while offering low short-wave infrared reflectance (SWIR). On the other hand, burned areas have a high shortwave infrared reflectance but low reflectance in the near infrared (https://www.skywatch.co/blog/assessing-impact-wildfire-normalized-burn-ratio-satellite, 30.8.2018). To emphasize these differentce we will calculate Normalized Burn Ration (NBR)(http://gsp.humboldt.edu/OLM/Courses/GSP_216_Online/lesson5-1/NBR.html):\n", "\n", "$$NBR = \\frac{(NIR-SWIR)}{(NIR+SWIR)}$$\n", "\n", "The formula above is applied for each pixel in the image and the result is as folows:\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\" \"Drawing\"
NBR before wildfire, on 7.8.2016 (EO Browser link) NBR after wildfire, on 17.8.2016 (EO Browser link)
\n", "\n", "Analyzing the right image we notice that burned areas are visualized in darker, almost black color and it is now easier to distinguish them from the rest of areas. By digitizing a polygon around the scar we can roughly estimate the size of damaged area, which is approx. $41 km^2$. Note also that clouds and some water areas (sea) appear darker on the image but they shall not be be mistaken for burned areas.\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\"
Polygons digitized around burned areas.
\n", "\n", "In order to estimate the severity of damage that wildfire left on vegetation we will calculate differences between NBR before and NBR after the wildfire. The result is than classified into 4 classes to produce a wildfire severity map.\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\"
Map of wildfire severity based on dNBR, where the most damaged area are colored dark red, severly damaged ones are orange and less damaged areas are yellow (EO Browser link).
\n", "\n", "Right after a wildfire is successfully extinguished the vegetation starts with recovery. The peace of recovery depends on the severity of damage, weather conditions, etc. A wildfire usually turns organic material to ashes so that nutrients return to the soil. Wildfire also clears thick growth so sunlight can reach the forest floor and encourage the growth of native species. Fire frees these plants from the competition delivered by invasive weeds and eliminates diseases or droves of insects that may have been causing damage to old growth (https://science.howstuffworks.com/environmental/green-science/how-forest-fire-benefit-living-things-2.htm). Recovery of vegetation can also be monitored using time series of satellite images acquired after the event. \n", "\n", "For our example of Madeira wildfire we will calculate Normalized difference vegetation index (NDVI, https://en.wikipedia.org/wiki/Normalized_difference_vegetation_index) for burned area and compare this with NDVI calculated for healthy (unburned) vegetation. In the figure below we plot NDVI values for period of 2 years. Value of NDVI is correlated to amount of chlorophyl in vegetation. Higher the value of NDVI, more chlorophyl and healthier vegetation.\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\" \"Drawing\"
NDVI values of vegetation on unburend area. NDVI values of vegetation on burned area.
\n", "\n", "In the left figure we can see the normal yearly cycle of vegetation with lower values in winter time (November - March) higher values in summer time (April - October). In the right figure, the decrease in NDVI value as a consequence of wildfire in August 2016 is obvious. Even two years after the wildfire the values of NDVI are still lower comparing to the values in unburned areas (note a different scale of y axis in both graphs)." ] }, { "cell_type": "markdown", "metadata": { "collapsed": true }, "source": [ "# Siberia, July 2018\n", "\n", "Not only fires and burned areas, satellite images can also be used to observe the direction in which smoke spreads from a fire or to estimate a concentration of released gasses. We will use wildfires in Siberia, Russia as an example to show how to vizualize this information. \n", "\n", "Dry, warm conditions in the spring set the stage for fires in Siberia. The wildfires burning boreal forests in in mid of July can be nicely observed from satellite images. \n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\" \"Drawing\"
Sibiria fires on 21st of July 2018. Scene is vizualized using a custom script by Pierre Markuse. Important to note, yellow and red areas colored area shall be interpreted as the hottest areas but they do not necessarily present the active fires (EO Browser link).Zoomed in to the area marked with blue in the left figure. Scene is vizualized using a custom script by Pierre Markuse. Important to note, yellow and red areas shall be interpreted as the hottest areas but they do not necessarily present the active fires (EO Browser link).
\n", "\n", "\n", "Vegetation fires release large amounts of particulate matter and toxic gases including carbon monoxide (CO), nitrogen oxides (NO), and non-methane organic compounds into the atmosphere. Measurements of these exhaustions are essential for forecasts, research on atmospheric composition and to develop warning systems (https://public.wmo.int/en/media/news/drought-and-heat-exacerbate-wildfires, 7.9.2018). Special sensors are needed to observe these gases and they are on board of Satellite Sentinel 5p. For the area a bit larger than the one shown in figures above, the Sentinel 5p measured increased concentration of CO as a consequence of the wildfire.\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "\n", "
\"Drawing\" \"Drawing\"
Map of CO concentration before the wildfire (get image from Sentinel-Hub). White pixels in the image are pixels with no data probably because of thick clouds. Map of CO concentration after the wildfire (get image from Sentinel-Hub). White pixels in hte image are pixels with no data probably because of thick clouds.
\n", "\n", "Besides CO, wildfires also release carbon dioxide (CO2) into the atmosphere, contributing to global warming. For instance, fires burned around 3 million hectares of land in Indonesia during the 2015 dry season released about 11.3 teragramms of CO2, which is roughly 120% of daily release of CO2 from fossil fuel burning in the European Union https://public.wmo.int/en/media/news/drought-and-heat-exacerbate-wildfires, 7.9.2018).\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Conclusions\n", "\n", "We have learned that several aspects of wildfires and their influence on the environment can be observed from satellite images. We visualized and measured the extent of burned area and severity of burn scar for wildfire in Madeira. The same approach can be used for any other wildfire. We checked what was the influence on vegetation and how long it took it to recover by inspecting NDVI. We also checked how the smoke and CO spread form the fires in Siberia earlier this year.\n", "\n", "However, whenever satellite data is used for we need to keep in mind the limitation of the data. Images are acquired from space and light must travel through different layers of atmosphere before it reaches the sensors, which can influence the accuracy of the images. Spatial and temporal resolutions of images are limited. Clouds or shades can obscure scenes in which we are interested as we experienced in the example of Madeira. We need to take these factors into consideration when interpreting satellite images or when making decisions based on such analysis.\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Other Resources to Check\n", "\n", "If you find this topic interesting, you might also want to check out:\n", "\n", "-\tEO Browser: https://apps.sentinel-hub.com/eo-browser/\n", "\n", "-\tSentinel Playground: https://apps.sentinel-hub.com/sentinel-playground/\n", "\n", "-\tGlobal for est watch: https://fires.globalforestwatch.org/map/#activeLayers=viirsFires%2CactiveFires%2CfireStories%2Ctwitter&activeBasemap=topo&activeImagery=&planetCategory=PLANET-MONTHLY&planetPeriod=Aug%202018&x=-12&y=18&z=3\n", "\n", "-\tEuropean, Rapid Damage Assessment: http://effis.jrc.ec.europa.eu/about-effis/technical-background/rapid-damage-assessment/ \n", "\n", "-\tCopernicus, Emergency Management Service http://effis.jrc.ec.europa.eu/static/effis_current_situation/public/index.html\n", "\n", "-\tDrought and heat exacerbate wildfires: https://public.wmo.int/en/media/news/drought-and-heat-exacerbate-wildfires \n", "\n", "-\tFrom California to Siberia: satellite images of wildfires around the world: https://unearthed.greenpeace.org/2018/08/10/california-wildfires-nasa-satellite-map/\n", "\n", "-\tLandsat Image Maps Aid Fire Recovery Efforts: https://landsat.gsfc.nasa.gov/landsat-image-maps-aid-fire-recovery-efforts/ \n", "\n", "-\tDetection of burned areas with Machine Learning: https://webthesis.biblio.polito.it/8197/1/tesi.pdf\n", "\n", "-\tAssessing the impact of a wildfire with satellites: https://www.skywatch.co/blog/assessing-impact-wildfire-normalized-burn-ratio-satellite\n", "\n", "- How does a forest fire benefit living things?: https://science.howstuffworks.com/environmental/green-science/how-forest-fire-benefit-living-things-2.htm\n", "\n", "- Brigitte Leblon, Laura Bourgeau-Chavez and Jesús San-Miguel-Ayanz (August 1st 2012). Use of Remote Sensing in Wildfire Management, Sustainable Development Sime Curkovic, IntechOpen, DOI: 10.5772/45829. Available from: https://www.intechopen.com/books/sustainable-development-authoritative-and-leading-edge-content-for-environmental-management/use-of-remote-sensing-in-wildfire-management" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.3" } }, "nbformat": 4, "nbformat_minor": 2 }