MAAP #125: Detecting Illegal Logging with Very High Resolution Satellites

Very high resolution satellite image showing illegal logging in the southern Peruvian Amazon. Data: Maxar. Analysis: MAAP/ACCA.

Illegal logging in the Peruvian Amazon is mainly selective and, until now, difficult to detect through satellite information.

In this report, we present the enormous potential of very high resolution satellite imagery (<70 cm) to identify illegal logging.

The leading entities that offer this type of data are Planet (Skysat) and Maxar (Worldview).

We emphasize that this technique has the potential to detect the illegal activity in real time, when preventive action is still possible.

This is an important advance because when an intervention normally occurs, such as detaining a boat or truck with illegal timber, the damage is done.

Below, we show a specific case of using very high resolution satellite imagery to detect and confirm probable illegal logging in the southern Peruvian Amazon (Madre de Dios region).

 

 

 

 

Case: Turbina SAC

The Base Map below shows the intensity of probable illegal logging activity* in the Turbina SAC forestry concession, from 2016 to the present. Specifically, it shows the exact points of illegal logging events (felled trees) and logging camps, as identified through our analysis of very high-resolution satellite images. Note that this forestry concession is adjacent to the Los Amigos Conservation Concession, an important long-term (20 years) biodiversity conservation area.

Base Map. Illegal logging activities in the Turbina SAC forestry concession. The size of the points is for reference only. Data: MAAP/Amazon Conservation.

Very High Resolution Satellite Imagery

Below, we show a series of very high-resolution satellite images, courtesy of the innovative satellite companies Planet and Maxar.

The first image shows the identification of probable illegal logging between June 2019 (left panel) and August 2020 (right panel). The red circle indicates the exact area (canopy) of the illegally logged tree.

The identification of illegal logging between June 2019 (left panel) and August 2020 (right panel). Click to enlarge. Data: Maxar, Planet, MAAP.

The following image shows the identification of illegal logging in March 2020. The red circle indicates the exact area of the illegally logged trees.

Identification of illegal logging. Data: Maxar, MAAP.

The following image shows the identification of a logging camp in March 2o20. The red circle indicates the area of the camp.

Satellite image of an illegal logging camp. Data: Maxar, MAAP.

*Statement on Legality

We determined that this logging activity is illegal from a detailed analysis of official information from the Peruvian Government (specifically, the Peruvian Forestry Service, SERFOR, and forestry oversight agency, OSINFOR). This information indicates that, although the concession is in force (Vigente), its status is classified as Inactive (Inactiva). In addition, 2013 was the last year that this concession had an approved logging plan (Plan Operativo de Aprovechamiento, or POA), and it was for a different sector of the concession from the newly detected logging activity.

To confirm our assumption of illegal activity, we requested the technical opinion from the corresponding regional forestry and wildlife authority, however, as of the date of publication of this report, we have not yet received a response.

Thus, with the information we had at the time of publication, we concluded the logging was illegal as it was not conducted within a current management plan.

Methodology

We carried out the analysis in two main steps:

The first step was the visual interpretation and digitization of new logging events and associated logging camps within the Turbina forestry concession. This analysis was based on the evaluation of submetric images obtained from the satellite companies Planet and Maxar, for the period 2019-20. It is worth noting that for Planet, we had the new ability to “task” new images for a specific area, rather than waiting for an image to appear by other means. Logging in the Peruvian Amazon is usually highly selective for high-value species, thus its detection requires a comparative analysis of images (before and after), in such a way that the trees cut during the study period (2019-20 in this case) can be identified.

The second step focused on an analysis of the legality of the identified logging events. The locations of the logged trees and camps were cross-referenced with spatial information on the state and status of forestry concessions provided by the GeoSERFOR (SERFOR) portal, as well as the areas delimited in the annual operational plans of the concessions, verified by OSINFOR and distributed through the SISFOR portal (WMS). We considered both spatial and temporal aspects to the forestry concession data.

Citation

Novoa S, Villa L, Finer M (2020) Detecting Illegal Logging with Very High Resolution Satellites. MAAP: 125.

Acknowledgments

We thank A. Felix (USAID Prevent), M.E. Gutierrez (ACCA), and G. Palacios for their helpful comments on this report.

This report was conducted with technical assistance from USAID, via the Prevent project. Prevent is an initiative that, over the next 5 years, will work with the Government of Peru, civil society, and the private sector to prevent and combat environmental crimes in Loreto, Ucayali and Madre de Dios, in order to conserve the Peruvian Amazon.

This publication is made possible with the support of the American people through USAID. Its content is the sole responsibility of the authors and does not necessarily reflect the views of USAID or the US government.

MAAP #58: Link between Peru’s Flooding and Warm Coastal Waters

In previous articles MAAP #56 and MAAP #57, we presented a series of striking satellite images of the recent deadly floods in northern Peru. Satellites provide additional types of data critical to better understanding events such as extreme flooding. Here, we present two more types of satellite data related to the flooding: ocean water temperature and precipitation.


Warming Coastal Waters

Image 58a. Data: NOAA

Satellite data from NOAA (the U.S. National Oceanic and Atmospheric Administration) clearly shows the warming of the northern Peruvian coastal waters immediately before and during the heavy rains and flooding (1, 2). Specifically, Image 58a shows the sudden warming in January, followed by intensifying warming in February and March (white inset box indicates primary flooding zone). Peruvian experts have referred to this phenomenon as “coastal El Niño”.

Heavy Rains

Image 58b. Data: Senamhi, GPM/NASA

Image 58b shows the resulting accumulated monthly precipitation totals (white inset box indicates primary flooding zone). In January, as expected, the dry northern coast had much lower precipitation than the Amazon region to the east. In February and March, however, the northern coast experienced abnormally intense rainfall, even more than many parts of the Amazon.

Floods linked to Climate Change?

Questions have emerged regarding the link between the deadly Peruvian floods and climate change (3). As seen in the images above, the sudden appearance of warm coastal waters coincides with intense rains in the primary flooding zone. Additional analysis is needed to better understand the link between the Peruvian floods and climate change, but such events are consistent with predictions related to heavy rains fueled by ocean warming due to climate change (3). Climate change could also increase the frequency or intensity of El Niño events (4).

References

  1. Villa, L. (27 de marzo 2017). Radar Sentinel-1: Evaluación Preliminar del Impacto del Niño Costero en Perú (Parte II). [Mensaje en un blog]. Recuperado de: http://luciovilla.blogspot.com/2017/03/radar-sentinel-1-evaluacion-preliminar_27.html
  2. Villa, L. (17 de marzo 2017). Radar Sentinel-1: Evaluación Preliminar del Impacto del Niño Costero en Perú (Parte I). [Mensaje en un blog]. Recuperado de: http://luciovilla.blogspot.com/2017/03/radar-sentinel-1-evaluacion-preliminar.html
  3. Berwyn B (2017) Peru’s Floods Follow Climate Change’s Deadly Extreme Weather Trend. Inside Climate News. Link: https://insideclimatenews.org/news/24032017/peru-floods-extreme-weather-climate-global-warming-el-nino
  4. Fraser B (2017) Coastal El Niño catches Peru by surprise. EcoAmericas March 2017.

Citation

Finer M, Novoa S, Gacke S (2017) Link between Peru’s Flooding and Warm Coastal Waters. MAAP: 58.

MAAP: What satellites show us about Peru’s flooding

Image 57. Data: ESRI, INEI, MINAM. Click to enlarge.

Satellites provide unique information that is critical to understanding events on Earth, including the recent deadly flooding in northern Peru.

In the previous MAAP #56, we showed a series of satellite images of the deadly floods that recently hit northern Peru.

Here, we highlight how satellites can show us the extent, indicators, impacts, and causes of the flooding.

Image A (see left) shows the general extent of the flooding in northern Peru. Analyzing satellite imagery, we identified 13 major rivers that flooded, indicated in blue.

 

 

 

 

 

Indicators of Flooding

An indicator of intense rains and flooding in northern Peru is the formation of the temporary lagoons La Niña and La Niña Sur, in the region of Piura. Image B shows the rapid formation of the lagoons between late January (left panel) and March 2017 (right panel).

Image B. Data: ESA

Impact of Flooding

The centerpiece of our analysis is a series of high resolution satellite images of the flooding. Images C and D show, in striking detail, some of the local impacts to the Panamerican Highway and croplands between January (left panel) and March (right panel) 2017.

Image C. Data: DigitalGlobe (Nextview)
Inset C1. Data: DigitalGlobe (Nextview)
Image D. Data: DigitalGlobe (Nextview)
Inset D1. Data: DigitalGlobe (Nextview)

Causes of Flooding

Satellites also provide data about the link between ocean water temperature and the heavy rains causing the floods. Image E shows the warming of the northern Peruvian coastal waters immediately before and during the heavy rains and flooding. Peruvian experts have referred to this phenomenon as “coastal El Niño”.

Image E. Data: NOAA


Image F shows  the resulting accumulated monthly precipitation totals (white inset box indicates primary flooding zone). In January, as expected, the dry northern coast had much lower precipitation than the Amazon region to the east. In February and March, however, the northern coast experienced abnormally intense rainfall, even more than many parts of the Amazon.

Image F. Data: Senamhi, GPM/NASA

Citation

Novoa S, Finer M (2017) What satellites show us about Peru’s flooding. MAAP.

MAAP #57: High Resolution Satellite Images of the Flooding in Peru

Image 57. Data: ESRI, INEI, MINAM. Click to enlarge.

In the previous MAAP #56, we showed a series of satellite images of the deadly floods that recently hit northern Peru.

In this report, we show a series of new, very high resolution satellite images (50 cm) of the flooding. They show, in striking detail, some of the local impacts, including to croplands and the Panamerican Highway.

Image 57 shows the 13 rivers that recently overflowed in northern Peru.

Below, we show images of the flooding around four of the rivers, labelled A-D.

 

 

 

 

 

 

 

 

Tumbes River

Image 57a shows the flooding along a stretch of the Tumbes River between October 2016 (left panel) and March 2017 (right panel). The yellow inset boxes indicate the areas of the follow-up zooms.

Image 57a. Data: Digital Globe (Nextview)
Inset A1. Data: Digital Globe (Nextview)
Inset A2. Data: Digital Globe (Nextview)

Chira River

Image 57b shows the flooding along a stretch of the Tumbes River between January (left panel) and March 2017 (right panel). The yellow inset boxes indicate the areas of the follow-up zooms.

Image 57b. Data: Digital Globe (Nextview)
Inset B1. Data: Digital Globe (Nextview)
Inset B2. Data: Digital Globe (Nextview)

La Leche River

Image 57c shows the flooding along a stretch of the La Leche River between January (left panel) and March 2017 (right panel). The yellow inset boxes indicate the areas of the follow-up zooms. Note the flooding of the PanAmerican Highway.

Image 57c. Data: Digital Globe (Nextview)
Inset C1. Data: Digital Globe (Nextview)

Jequetepeque River

Image 57d shows the flooding along a stretch of the Jequetepeque River between January (left panel) and March 2017 (right panel). The yellow inset boxes indicate the areas of the follow-up zooms.

Image 57d. Data: Digital Globe (Nextview)
Inset D1. Data: Digital Globe (Nextview)
Inset D2. Data: Digital Globe (Nextview)

References

UNOSAT, 2017. Efectos del Niño Costero: Inundaciones en Perú, Departamentos de La Libertad & Ancash. _Marzo_20170321

UNOSAT, 2017. Efectos del Niño Costero: Inundaciones en Perú, Departamentos de La Libertad & Ancash. _Marzo_20170321

UNOSAT, 2017. Efectos del Niño Costero: Inundaciones en Perú, Departamentos de Piura. Marzo_20170320

Citation

Novoa S, Finer M (2017) High Resolution Images of the Flooding in Peru. MAAP: 57

MAAP #55: New 2017 “Hurricane Winds” in Peruvian Amazon

In the previous MAAP #54, we described the phenomenon of natural forest loss due to “hurricane winds,” showing several examples from 2016 in the Peruvian Amazon. Strong winds from these localized storms can knock down hundreds of acres of forest at a time.

In January 2017, GLAD tree loss alerts indicated two new hurricane wind events in the southern Peruvian Amazon (Madre de Dios region). Below, we show high-resolution images of these cases. The first is a large hurricane wind event that knocked down 780 acres (Image 55a). The second is an event of 185 acres that took place within a forestry concession (Image 55b).

Image 55a: Data: Planet
Image 55b: Data: Planet

Very High Resolution View

We also show a new very high resolution image (0.5 meters) of one of the hurricane wind events in 2016 in the Loreto region (example B of MAAP #54). Image 55c shows the following pattern: fan-shaped pattern of forest loss with a defined orientation following the direction of the storm winds. It is worth mentioning that this event occurred within a protected area, Maijuna-Kichwa Regional Conservation Area.

Image 55c. Data: Digital Globe (Nextview)

Reference

Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com

Citation

Finer M, Novoa S (2017) New 2017 “Hurricane Winds” in Peruvian Amazon. MAAP: 55.

MAAP #54: Natural forest loss due to “hurricane winds” in the Peruvian Amazon

Image 54. Base Map

A little-known, but not uncommon, type of natural forest loss in the Peruvian Amazon is blowdown due to strong winds from localized storms (locally known as “hurricane winds”).

The intense winds cause a chain reaction of fallen trees, resulting in a fan-shaped pattern of forest loss with a defined orientation following the direction of the storm winds.

This phenomenon has previously been reported in Brazil and Colombia (see References below).

The base image (Image 54) shows the location of some recent (during 2016) examples of forest loss due to blowdowns in the Peruvian Amazon.

These examples were initially detected from analysis of GLAD alerts, early warning tree loss data produced by the University of Maryland (see Annex).

Below, we detail the 7 blowdown examples indicated on the base map. They are located in both northern (Loreto region) and southern (Madre de Dios region) Peru, and include 4 Protected Areas. The forest loss in these examples ranged from 24 to 900 hectares.

 

 

 

Loreto Examples

This section highlights 3 examples of blowdowns in Loreto. In each example, we show an image of before (left panel) and after (right panel) the forest loss due to the winds. The documented forest loss in these areas includes: 912 hectares in Example A, 124 hectares in Example B (Ampiyacu Apayacu Regional Conservation Area), and 357 hectares in Example C.

Image 54a. Data: Planet.
Image 54b. Data: Planet. Note: Blowdown in Ampiyacu Apayacu RCA, not Maijuna.

 

 

Image 54c. Data: Planet.

Madre de Dios Examples

This section highlights 4 examples of blowdowns in Madre de Dios. In each example, we show an image of before (left panel) and after (right panel) the forest loss due to the winds. The documented forest loss in these areas includes: 73 hectares in Example D (Manu National Park), 77 hectares in Example E, 93 hectares in Example F (Bahuaja Sonene National Park), and 24 hectares in Example G (Tambopata National Reserve).

Image 54d. Data: Planet.
Image 54e. Data: Planet.

 

Image 54f. Data: Planet.

 

Image 54g. Data: Planet.

Annex

This last image shows how the tree loss patterns from blowdowns appear in the GLAD alerts.

Coordinates

A.      -1.386944, -73.679444
B.      -3.029722, -72.786666
C.      -3.456111, -76.713333
F.       -13.294722, -69.295833

References

Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com

Espırito-Santo, F. D. B. et al. Storm intensity and old-growth forest disturbances in the Amazon region. Geophys. Res. Lett. 37, L11403 (2010).

Nelson, B. W. et al. Forest disturbance by large blowdowns in the Brazilian Amazon. Ecology 75, 853–858 (1994).

Garstang, M., White, S., Shugart, H. H. & Halverson, J. Convective cloud downdrafts as the cause of large blowdowns in the Amazon rainforest. Meteorol. Atmos. Phys. 67, 199–212 (1998).

Etter y Botero (1990) Efectos de los procesos climáticos y geomorfológicos en la dinámica del Bosque Húmedo Tropical de la Amazonía Colombiana. Colombia Amazonica 4:7.

Citation

Novoa S, Finer M (2017) Natural forest loss due to “hurricane winds” in the Peruvian Amazon. MAAP: 54.

MAAP SYNTHESIS #2: PATTERNS AND DRIVERS OF DEFORESTATION IN THE PERUVIAN AMAZON

We present our second synthesis report, building off our first report published in September 2015. This synthesis is largely based on the 50 MAAP reports published between April 2015 and November 2016. The objective is to synthesize all the information to date regarding deforestation trends, patterns and drivers in the Peruvian Amazon.

MAAP methodology includes 4 major components: Forest loss detection, Prioritize big data, Identify deforestation drivers, and Publish user-friendly reports. See Methodology section below for more details.

Our major findings include:

  • Trends. During the 15 years between 2001 and 2015, around 4,448,000 acres (1,800,000 hectares) of Peruvian Amazon forest has been cleared, with a steadily increasing trend. 2014 had the highest annual forest loss on record (438,775 acres), followed by a slight decrease  in 2015. The preliminary estimate for 2016 indicates that forest loss remains relatively high. The vast majority (80%) of forest loss events in the Peruvian Amazon are small-scale (<5 hectares), while large-scale events (> 50 hectares) pose a latent threat due to new agro-industrial projects.
  • Hotspots. We have identified at least 8 major deforestation hotspots. The most intense hotspots are located in the central Amazon (Huánuco and Ucayali). Other important hotspots are located in Madre de Dios and San Martin. Two protected areas (Tambopata National Reserve and El Sira Communal Reserve) are threatened by these hotspots.
  • Drivers. We present an initial deforestation drivers map for the Peruvian Amazon. Analyzing high-resolution satellite imagery, we have documented six major drivers of deforestation and degradation: small/medium-scale agriculture, large-scale agriculture, cattle pasture, gold mining, illegal coca cultivation, and roads. Small-scale agriculture and cattle pasture are likely the most dominant drivers overall. Gold mining is a major driver in southern Peru. Large-scale agriculture and major new roads are latent threats. Logging roads are likely a major source of forest degradation in central Peru.

Deforestation Trends

Image 1 shows forest loss trends in the Peruvian Amazon from 2001 to 2015, including a breakdown of the size of the forest loss events. This includes the official data from the Peruvian Environment Ministry, except for 2016, which is a preliminary estimate based on GLAD forest loss alerts.

Image 1. Data: PNCB/MINAM, UMD/GLAD. *Estimate based on GLAD alerts.

During the 15 years between 2001 and 2015, around 4,448,000 acres (1,800,000 hectares) of Peruvian Amazon forest has been cleared (see green line). This represents a loss of approximately 2.5% of the existing forest as of 2001.There have been peaks in 2005, 2009, and 2014, with an overall increasing trend. In fact, 2014 had the highest annual forest loss on record (386,626 acres). Forest loss decreased in 2015 (386,732 acres), but is still the second highest recorded. The preliminary estimate for 2016 indicates that forest loss continues to be relatively high.

It is important to note that the data include natural forest loss events (such as storms, landslides, and river meanders), but overall serves as our best proxy for anthropogenic deforestation. The non-anthropogenic forest loss is estimated to be approximately 3.5% of the total.1

The vast majority (81%) of forest loss events in the Peruvian Amazon are small-scale (<5 hectares, equivalent of 12 acres), see the yellow line. Around 16% of the forest loss events are medium-scale (5-50 hectares, equivalent of 12-124 acres), see the orange line. Large-scale (>50 hectares, equivalent of 124 acres) forest loss events, often associated with industrial agriculture, pose a latent threat. Although the average is only 2%, large-scale forest loss rapidly spiked to 8% in 2013 due to activities linked with a pair of new oil palm and cacao plantations. See MAAP #32 for more details on the patterns of sizes of deforestation events.

Deforestation Patterns

Image 2 shows the major deforestation hotspots in 2012-14 (left panel) relative to 2015-16 (right panel), based on a kernel density analysis.We have identified at least 8 major deforestation hotspots, labeled as Hotspots A-H.

Image 2. Data: PNCB/MINAM, GLAD/UMD. Click to enlarge.

The most intense hotspots, A and B, are located in the central Amazon. Hotspot A, in northwest Ucayali, was dominated by two large-scale oil palm projects in 2012-14, but then shifted a bit to the west in 2015-16, where it was dominated by cattle pasture and small-scale oil palm. Hotspot B, in eastern Huánuco, is dominated by cattle pasture (MAAP #26).

Hotspots C and D are in the Madre de Dios region in the southern Amazon. Hotspot C indicates the primary illegal gold mining front in recent years (MAAP #50). Hotspot D highlights the emerging deforestation zone along the Interoceanic Highway, particularly around the town of Iberia (MAAP #28).

Hotspots E-H are agriculture related. Hotspot E indicates the rapid deforestation for a large-scale cacao plantation in 2013-14, with a sharp decrease in forest loss 2015-16 (MAAP #35). Hotspot F indicates the expanding deforestation around two large-scale oil palm plantation (MAAP #41). Hotspot G indicates the intensifying deforestation for small-scale oil palm plantations (MAAP #48).

Hotspot H indicates an area impacted by intense wildfires in 2016.

Protected Areas, in general, are effective barriers against deforestation (MAAP #11). However, several protected areas are currently threatened, most notably Tambopata National Reserve (Hotspot C; MAAP #46). and El Sira Communal Reserve (Hotspot B; MAAP #45).

Deforestation Drivers

Image 3. Data: MAAP, SERNANP. Click to enlarge.

Surprisingly, there is a striking lack of precise information about the actual drivers of deforestation in the Peruvian Amazon. According to an important paper published in 2016, much of the existing information is vague and outdated, and is based solely on a general analysis of the size of deforestation events.3  

As noted above, one of the major advances of MAAP has been using high-resolution imagery to better identify deforestation drivers.

Image 3 shows the major deforestation drivers identified thus far by our analysis. As far as we know, it represents the first spatially explicit deforestation drivers map for the Peruvian Amazon.

To date, we have documented six major direct drivers of deforestation and degradation in the Peruvian Amazon: small/medium-scale agriculture, large-scale agriculture, cattle pasture, gold mining, illegal coca cultivation, and roads.

At the moment, we do not consider the hydrocarbon (oil and gas) and hydroelectric dam sectors as major drivers in Peru, but this could change in the future if proposed projects move forward.

We describe these major drivers of deforestation and degradation in greater detail below.

Small/Medium-scale Agriculture

The literature emphasizes that small-scale agriculture is the leading cause of deforestation in the Peruvian Amazon.However, there is little actual empirical evidence demonstrating that this is true.3 The raw deforestation data is dominated by small-scale clearings that are most likely for agriculture or cattle pasture. Thus, it is likely that small-scale agriculture is a major driver, but a definitive study utilizing high-resolution imagery and/or extensive field work is still needed to verify the assumption.

In several key case studies, we have shown specific examples of small-scale agriculture being a deforestation driver. For example, using a combination of high-resolution imagery, photos from the field, and local sources, we have determined that:

  • Oil Palm, in the form of small and medium-scale plantations, is one of the main drivers within deforestation Hotspot B (Ucayali; MAAP #26), Hotspot G (northern Huánuco; MAAP #48), and Hotspot F (Loreto-San Martin;MAAP #16). This was also shown for Ucayali in a recent peer-reviewed study.4 See below for information about large-scale oil palm.
  • Cacao is causing rapid deforestation along the Las Piedras River in eastern Madre de Dios (MAAP #23, MAAP #40). See below for information about large-scale cacao.
  • Papaya is an important new driver in Hotspot D, along the Interoceanic Higway in eastern Madre de Dios (MAAP #42).
  • Corn and rice plantations may also be an important driver in Hotspot D in eastern Madre de Dios (MAAP #28).

Large-scale Agriculture

Large-scale, agro-industrial deforestation remains a latent threat in Peru, particularly in the central and northern Amazon regions. This issue was put on high alert in 2013, with two cases of large-scale deforestation for oil palm and cacao plantations, respectively.

In the oil palm case, two companies that are part of the Melka group,5 cleared nearly 29,650 acres in Hotspot A in Ucayali between 2012 and 2015 (MAAP #4, MAAP #41). In the cacao case, another company in the Melka group (United Cacao) cleared 5,880 acres in Hotspot E in Loreto between 2013 and 2015 (MAAP #9, MAAP #13, MAAP #27, MAAP #35). Dennis Melka has explicitly stated that his goal is to bring the agro-industrial production model common in Southeast Asia to the Peruvian Amazon.6

Prior to these cases, large-scale agricultural deforestation occurred between 2007 and 2011, when oil palm companies owned by Grupo Palmas7 cleared nearly 17,300 acres for plantations in Hotspot H along the Loreto-San Martin border (MAAP #16). Importantly, we documented the additional deforestation of 24,215 acres for oil palm plantations surrounding the Grupo Palmas projects (MAAP #16).

In contrast, large-scale agricultural deforestation was minimal in 2015 and 2016. However, as noted above, it remains a latent threat. Both United Cacao and Grupo Palmas have expansion plans that would clear over 49,420 acres of primary forest in Loreto.8

Cattle Pasture

Using an archive of satellite imagery, we documented that deforestation for cattle pasture is a major issue in the central Peruvian Amazon. Immediately following a deforestation event, the scene of hundreds or thousands of recently cut trees often looks the same whether the cause is agriculture or cattle pasture. However, by using an archive of imagery and studying deforestation events from previous years, one can more easily determine the drivers of the forest loss. For example, after a year or two, agriculture and cattle pasture appear very differently in the imagery and thus it is possible to distinguish these two drivers.

Using this technique, we determined that cattle pasture is a major driver in Hotspots A and B, in the central Peruvian Amazon (MAAP #26, MAAP #37).

We also used this technique to determine that much of the deforestation in the northern section of El Sira Communal Reserve is due to cattle pasture (MAAP #45).

Maintenance of cattle pasture, and small-scale agriculture, are likely important factors behind the escaped fires that degrade the Amazon during intense dry seasons (MAAP #45, MAAP #47).

Gold Mining

Gold mining is one of the major drivers of deforestation in the southern Peruvian Amazon (Hotspot C). An important study found that gold mining cleared around 123,550 acres up through 2012.9 We built off this work, and by analyzing hundreds of high resolution imageres, found that gold mining caused the loss of an additional 30,890 acres between 2013 and 2016 (MAAP #50). Thus, gold mining is thus far responsible for the total loss of around 154,440 acres in southern Peru. Much of the most recent deforestation is illegal due to its occurrence in protected areas and buffer zones strictly off-limits to mining activities.

Most notably, we have closely tracked the illegal gold mining invasion of Tambopata National Reserve, an important protected area in the Madre de Dios region with renowned biodiversity and ecotourism. The initial invasion occurred in November 2015 (MAAP #21), and has steadily expanded to over 1,110 acres (MAAP #24, MAAP #30, MAAP #46). As part of this invasion, miners have modified the natural course of the Malinowski River, which forms the natural northern border of the reserve (MAAP #33). In addition, illegal gold mining deforestation continues to expand within the reserve’s buffer zone, particularly in an area known as La Pampa (MAAP #12, MAAP #31).

Further upstream, illegal gold mining is also expanding on the upper Malinowski River, within the buffer zone of Bahuaja Sonene National Park (MAAP #19, MAAP #43).

In contrast to the escalating situation in Tambopata, we also documented that gold mining deforestation has been contained in the nearby Amarakaeri Communal Reserve, an important protected area that is co-managed by indigenous communities and Peru’s national protected areas agency. Following an initial invasion of 27 acres in 2014 and early 2015, satellite imagery shows that management efforts have prevented any subsequent expansion within the protected area (MAAP #6, MAAP #44).

In addition to the above cases in Madre de Dios, gold mining deforestation is also increasingly an issue in the adjacent regions of Cusco and Puno (MAAP #14).

There are several small, but potentially emerging, gold mining frontiers in the central and northern Peruvian Amazon (MAAP #49). The Peruvian government has been working to contain the illegal gold mining in the El Sira Communal Reserve (MAAP #45). Further north in Amazonas region, there is gold mining deforestation along the Rio Santiago (MAAP #36, MAAP #49), and in the remote Condor mountain range along the border with Ecuador (MAAP #49).

Roads

Roads are a well-documented driver of deforestation in the Amazon, particularly due to their ability to facilitate human access to previously remote areas.10 Roads often serve as an indirect driver, as most of the deforestation directly associated with agriculture, cattle pasture, and gold mining is likely greatly facilitated by proximity to roads. We documented the start of a controversial road construction project that would cut through the buffer zones of two important protected areas, Amarakaeri Communal Reserve and Manu National Park (MAAP #29).

Logging Roads

In relation to general roads described above, we distinguish access roads that are constructed to gain entry to a particular project. The most notable type of access roads in Peru are logging roads, which are likely a leading cause of forest degradation as they facilitate selective logging of valuable timber species in remote areas.

One of the major recent advances in forest monitoring is the ability to quickly identify the construction of new logging roads. The unique linear pattern of these roads appears quite clearly in Landsat-based tree cover loss alerts such as GLAD and CLASlite. This advance is important because it is difficult to detect illegal logging in satellite imagery because loggers in the Amazon often selectively cut high value species and do not produce large clearings. But now, although it remains difficult to detect the actual selective logging, we can detect the roads that indicate that selective logging is taking place in that area.

In a series of articles, we highlighted the recent expansion of logging roads, including the construction of 1,134 km between 2013 and 2015 in the central Peruvian Amazon (MAAP #3, MAAP #18). Approximately one-third of these roads were within the buffer zones of Cordillera Azul and Sierra del Divisor National Parks (MAAP #15).

We documented the construction of an additional 83 km of logging roads during 2016,  (MAAP #40, MAAP #43) including deeper into the buffer zone of Cordillera Azul National Park.

Another major finding is the rapid construction of the logging roads. In several cases, we documented the construction rate of nearly five kilometers per week (MAAP #18, MAAP #40, MAAP #43).

Determining the legality of these logging roads is complex, partly because of the numerous national and local government agencies involved in the authorization process. Many of these roads are near logging concessions and native communities, whom may have obtained the rights for logging from the relevant forestry authority (in many cases, the regional government).

Coca

According to a recent United Nations report, the Peruvian land area under coca cultivation in 2015 (99,580 acres) was the lowest on record (since 2001) and part of a declining trend since 2011 (154,440 acres).11 There are 13 major coca growing zones in Peru, but it appears that only a few of them are actively causing new deforestation. Most important are two coca zonas in the region of Puno that are causing deforestation within and around Bahuaja Sonene National Park (MAAP #10, MAAP #14). Several coca zones in the regions of Cusco and Loreto may also be causing some new deforestation.

Hydroelectric Dams

Although there is a large portfolio of potential new hydroelectric dam projects in the Peruvian Amazon,12 many of not advanced to implementation phase. Thus, forest loss due to hydroelectric dams is not currently a major issue, but this could quickly change in the future if these projects are revived. For example, in adjacent western Brazil, we documented the forest loss of 89,205 acres associated with the flooding caused by two dams on the upper Madeira River (MAAP #34).

Hydrocarbon (Oil & Gas)

During the course of our monitoring, we have not yet detected major deforestation events linked to hydrocarbon-related activities. As with dams, this could change in the future if oil and gas prices rise and numerous projects in remote corners of the Amazon move forward.

Methodology

MAAP methodology has 4 major components:

  1. Forest Loss Detection. MAAP reports rely heavily on early-warning tree cover loss alerts to help us identify where new deforestation is happening. Currently, our primary tool is GLAD alerts, which are developed by the University of Maryland and Google,13 and presented by WRI’s Global Forest Watch and Peru’s GeoBosques. These alerts, launched in Peru in early 2016, are based on 30-meter resolution Landsat satellite images and updated weekly. We also occasionally incorporate CLASlite, forest loss detection software based on Landsat (and now Sentinel-2) developed by the Carnegie Institution for Science, and the moderate resolution (250 meters) Terra-i alerts. We are also experimenting with Sentinel-1 radar data (freely available from the European Space Agency), which has the advantage of piercing through cloud cover in order to continue monitoring despite persistent cloudy conditions
  2. Prioritize Big Data. The early warning systems noted above yield thousands of alerts, thus a procedure to prioritize the raw data is needed. We employ numerous prioritization methods, such as creation of hotspot maps (see below), focus on key areas (such as protected areas, indigenous territories, and forestry concessions), and identification of striking patterns (such as linear features or large-scale clearings).
  1. Identify Deforestation Drivers. Once priority areas are identified, the next challenge is to understand the cause of the forest loss. Indeed, one of the major advances of MAAP over the past year has been using high-resolution satellite imagery to identify key deforestation drivers. Our ability to identify these deforestation drivers has been greatly enhanced thanks to access to high-resolution satellite imagery provided by Planet 14
    (via their Ambassador Program) and Digital Globe (via the NextView Program, courtesy of an agreement with USAID). We also occasionally purchase imagery from Airbus(viaApollo Mapping).
  2. Publish User-Friendly Reports. The final step is to publish technical, but accessible, articles highlighting novel and important findings on the MAAP web portal. These articles feature concise text and easy-to-understand graphics aimed at a wide audience, including policy makers, civil society, researchers, students, journalists, and the public at large. During preparation of these articles, we consult with Peruvian civil society and relevant government agencies in order to improve the quality of the information.

Endnotes

MINAM-Peru (2016) Estrategia Nacional sobre Bosques y Cambio Climático.

Methodology: Kernel Density tool from Spatial Analyst Tool Box of ArcGis. The 2016 data is based on GLAD alerts, while the 2012-15 data is based on official annual forest loss data

Ravikumar et al (2016) Is small-scale agriculture really the main driver of deforestation in the Peruvian Amazon? Moving beyond the prevailing narrative. Conserv. Lett. doi:10.1111/conl.12264

4 Gutiérrez-Vélez VH et al (2011). High-yield oil palm expansion spares land at the expense of forests in the Peruvian Amazon. Environ. Res. Lett., 6, 044029.

Environmental Investigation Agency EIA (2015) Deforestation by Definition.

NG J (2015) United Cacao replicates Southeast Asia’splantation model in Peru, says CEO Melka. The Edge Singapore, July 13, 2015.

Palmas del Shanusi & Palmas del Oriente; http://www.palmas.com.pe/palmas/el-grupo/empresas

Hill D (2015) Palm oil firms in Peru plan to clear 23,000 hectares of primary forest. The Guardian, March 7, 2015.

Asner GP, Llactayo W, Tupayachi R,  Ráez Luna E (2013) Elevated rates of gold mining in the Amazon revealed through high-resolution monitoring. PNAS 46: 18454. They reported 46,417 hectares confirmed and 3,268 hectares suspected (49,865 ha total).

10 Laurance et al (2014) A global strategy for road building. Nature 513:229; Barber et al (2014) Roads, deforestation, and the mitigating effect of protected areas in the Amazon.  Biol Cons 177:203.

11 UNODC/DEVIDA (2016) Perú – Monitoreo de Cultivos de Coca 2015.

12 Finer M, Jenkins CN (2012) Proliferation of Hydroelectric Dams in the Andean Amazon and Implications for Andes-Amazon Connectivity. PLoS ONE 7(4): e35126.

13 Hansen MC et al (2016) Humid tropical forest disturbance alerts using Landsat data. Environ Res Lett 11: 034008.

14 Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com

Citation

Finer M, Novoa S (2017) Patterns and Drivers of Deforestation in the Peruvian Amazon. MAAP: Synthesis #2.

MAAP #53: Wildfire Hotspots in the Peruvian Amazon in 2016

Imagen 53. VIIRS/NASA, SERNANP.

During 2016, Peru experienced an intense wildfire season, exacerbated by widespread drought conditions across the country.

The base map (Image 53, to the left) shows the 2016 fire alert hotspots.

These alerts are generated from a moderate-resolution (375 meters) satellite sensor (VIIRS) that detects significant new heat sources.

Although there has not yet been a comprehensive evaluation of the causes of these fires, evidence indicates that many are linked to agricultural practices that allow fires to escape to surrounding natural habitats.

In the image, we highlight 5 significant fire hotspots in the Amazon basin, labeled A-E (A. Northern Peru; B. Lower Huallaga; C. Huánuco/Ucayali, D. Ene River, E. Southern Manu, F. Interoceanic Highway).

These areas are described in more detail below.

 

 

 

 

 

 

A. Northern Peru

Image 53a. Data: VIIRS/NASA, SERNANP, MODIS

Hotspot A indicates the area in northern Peru that experienced a wave of intense fires in late 2016. Most of the fires occurred in the headwaters of the Amazon, in the Cajamarca and Lambayeque regions.

As previously reported, we estimate that 6,594 acres were burned within 11 Protected Areas (see MAAP #51 and MAAP #52).

Image 53a shows where the concentrations of heat sources were recorded.

B. Lower Huallaga

Hotspot B corresponds to the area along the lower Huallaga river basin, between the regions of Loreto and San Martín. Although most of the fires were in established agricultural areas, some impacted forest and secondary vegetation for the opening of new agricultural areas (Image 53b).

Image 53b. VIIRS/NASA, Planet

C. Huánuco/Ucayali

Hotspot C overlaps with one of the primary deforestation hotspots in the country. As previously reported, one of the primary drivers of deforestation in this area is cattle pasture (see MAAP #37). Therefore, there may be a relationship between the use of fire in agricultural activities and the high deforestation rates in this area.

D. Ene River

Hotspot D highlights an area that generated national and international attention in 2016, when fires along the Ene River threatened two national protected areas (Asháninka Communal Reserve and Otishi National Park) in the Junin region. Image 53d shows a comparison of before (left panel) and during (right panel) the fires. We did not document any fires entering the protected areas.

Image 53d. VIIRS/NASA, SERNANP, Planet

E. South of Manu

Hotspot E corresponds to an area of grassland, inter-Andean valley, and cloud forest in the buffer zone of Manu National Park and surrounding the Wayqecha Private Conservation Area. According to estimates of local officials, around 3,000 hectares burned.

Image 53e. VIIRS/NASA, SERNANP, Planet

F. Interoceanic Highway

Hotspot F indicates an area in southern Peru experiencing increasing deforestation along the Interoceanic Highway in the Madre de Dios region. We previously documented a correlation between the areas with high concentrations of fires and areas of elevated deforestation (see MAAP #47).

References

Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com

Citation

Novoa S, Finer M, Samochuallpa E (2017) Wildfire Hotspots in Peruvian Amazon in 2016. MAAP: 53.

MAAP #52: Update – Fires Degrade 11 Protected Areas in northern Peru

Image 52a. Data: MODIS/NASA, SERNANP, NCI.

In the previous MAAP #51, we gave an initial impact assesment regarding the recent wave of fires in protected areas in northern Peru. Here, we provide a more comprehensive update.

Our revised estimate is 6,594 acres (2,668 hectares) burned in 11 Protected Areas (see Image 52a) in late 2016. Note that the image is from November and smoke from the fires is clearly seen.

The majority (4,165 acres) occured in 7 national protected areas under national administration (Cutervo National Park, Pagaibamba Protected Forest, Laquipampa Wildlife Refuge, Tumbes National Reserve, Cerros de Amotape National Park, Tabaconas-Namballe National Sanctuary, Udima Wildlife Refuge).*

The estimates refer to areas directly affected by fires (i.e. burned) and come from two sources: our analysis of satellite images and field information from SERNANP, the Peruvian protected areas agency.

It appears that the primary cause of these fires is poor agricultural burning practices during a time of intense drought. These conditions allowed fires to escape into protected areas.

Below, we show a series of new satellite images of some of the burn areas (for images of other areas, see MAAP #51). We also publish a statement from SERNANP.

 

*The rest occured in 3 national protected areas under private administration (Chicuate-Chinguelas, Huaricancha, and Bosques de Dotor Private Conservation Areas; 1,927 acres) and 1 municipal protected area (ACA Cachiaco-San Pablo; 502 acres).


Cutervo National Park

The following image shows a comparison of the northern sector of Cutervo National Park before (left panel) and after (right panel) the fires. The estimated burn area within the park is 731 acres. The red dots indicate the fire alerts (heat sources) detected by the VIIRS satellite sensor (note the high correlation between the distribution of the alerts and confirmed burn areas).
Image 52b. Data: Planet, VIIRS/NASA, SERNANP. Click to enlarge.

Pagaibamba Protected Forest

The following image shows a comparison of the southern sector of Pagaibamba Protected Forest before (left panel) and after (right panel) the fires. The red dots indicate the fire alerts. SERNANP estimates the burn area within the protected forest at 1,013 acres (see SERNANP statement below).

Image 52c. Data: Planet, Digital Globe (Nextview), VIIRS/NASA, SERNANP. Click to enlarge.

Tumbes National Reserve

The following image shows a comparison of the western sector of Tumbes National Reserve before (left panel) and after (right panel) the fires. It also shows the smaller burn area within Cerros de Amotape National Park. The estimated burn area within the two adjacent protected areas is 1,285 acres. The red dots indicate the fire alerts.

Image 52d. Data: Planet, SERNANP, VIIRS/NASA. Click to enlarge.

Tabaconas-Namballe National Sanctuary

The following image shows a comparison of the western sector of Tabaconas-Namballe National Sanctuary before (left panel) and after (right panel) the fires. The estimated burn area within the national sanctuary is 35 acres. The red dots indicate the fire alerts.

Image 52e. Data: Planet, USGS/NASA, SERNANP, VIIRS/NASA. Click to enlarge.

Dotor Private Conservation Area

The following image shows a comparison of the northern sector of the private conservation area before (left panel) and after (right panel) the fires. The estimated burn area within the national sanctuary is 395 acres. The red dots indicate the fire alerts.

Image 52f. Data: Planet, VIIRS/NASA, SERNANP. Click to enlarge.

 

Statement from SERNANP

Note: This statement refers to the data in MAAP #51. In the current MAAP #52 report we have made the necessary corrections.

In regards to the effect of forest fires in 6 natural protected areas (Refugio de Vida Silvestre Laquipampa, Refugio de Vida Silvestre Bosques Nublados de Udima, Parque Nacional de Cutervo, Parque Nacional Cerros de Amotape, Reserva Nacional de Tumbes y Bosque de Protección Pagaibamba), located in the departments of Lambayeque and Cajamarca, we clarify that although the ACA and ACCA report refers to 1,400 hectares of heat sources in the particular case of the Pagaibamba Protected Forest, it should be noted that according to the verification carried out in-situ by the SERNANP personnel, the burned habitat amounts to only 410 hectares. The remaining 990 hectares were affected, but indirectly, by presence of smoke and ash.

In addition, SERNANP led a multisectoral action along with our park guards who hare specialized in forest fires, as part of immediate attention to the emergency regarding the forest fires in the affected protected areas, obtaining positive results in a short time.

Finally, SERNANP personnel are assessing the ecological damage and developing a recovery plan.

Citation

Novoa S, Finer M (2017) Update – Fires Degrade 11 Protected Areas in northern Peru. MAAP: 52.