Conservation impacts of a near real‐time forest monitoring and alert system for the tropics

Abstract

Near real‐time (NRT) forest monitoring and alert systems based on remotely sensed data are among the most recently developed tools to help manage and protect forest resources. The use of satellite and airborne remote sensing to quickly and accurately detect activities associated with deforestation has great potential for catalyzing local response teams responsible for assessing and interdicting threats to tropical forest ecosystems. To better understand the utility of NRT monitoring technologies in improving environmental protection and management, from 2008 to 2016 Conservation International conducted a series of surveys and interviews with users of these systems in four countries where wildfires threaten tropical forest ecosystems: Madagascar, Indonesia, Bolivia and Peru. Users reported that NRT forest monitoring systems made significant contributions to improving the ability of conservation and forest management organizations to respond to and reduce the impacts of fire, deforestation and other illegal or undesirable forest activity. Understanding the types of applications for which users successfully employed forest monitoring data, the challenges they faced in accessing, analyzing or disseminating these data, and the lessons learned from our attempts to develop and deploy NRT forest monitoring systems is useful to institutions interested in successfully incorporating NRT monitoring into their work.

Introduction

The loss of the world's natural habitat through timber extraction, wildland fires and agricultural expansion is causing wide‐ranging environmental and economic impacts, including biodiversity loss and the degradation of ecosystem services vital to human well‐being (Cochrane ; Tscharntke et al. ; Foley et al. ). Anticipating and responding to illegal, unsustainable human activity has long been a challenge for forest managers. Fire or small‐scale encroachment that occurs in remote locations is often difficult to track from the ground. The use of ground patrols to monitor large forested landscapes in rugged terrain is difficult, particularly during the rainy season or when road or trail access is limited.

Operational near real‐time (NRT) monitoring and alert systems utilizing remotely sensed data to detect forest threats from fire, road development, logging and agriculture provide a tremendous opportunity to help manage and protect natural resources (Rose et al. ; Pratihast et al. ). NRT forest monitoring involves repeated or ongoing tracking of forest threats or disturbances with a minimum time‐lag between the observation, analysis and distribution of information on threats or disturbances detected (Hansen et al. ). The main purpose of NRT forest monitoring is to provide the information necessary for rapidly responding and intervening to reduce threats to ecologically sensitive, economically valuable or culturally important forest resources, usually from deforestation, logging or fire.

Various satellite‐based monitoring systems now exist that channel NRT data on fire and deforestation to both technical and non‐technical users in tropical forest countries (Davies et al. ; MZ ; Wheeler et al. ; Diniz et al. ; Hansen et al. ). In 2002, Conservation International (CI) created one of the first NRT email alert systems; the initial prototype enabled delivery of manually generated fire alerts to users from seven countries, compiled by overlaying NASA's MODIS active‐fires (Giglio et al. ) on protected area boundaries in ArcView. In 2007, CI introduced an automated version of the system based on ArcGIS Server in which online user registration, spatial querying and email generation were fully integrated to produce fire alerts for a range of user‐defined areas of interest, including protected areas, key biodiversity areas, indigenous areas, forest concessions, political units and vegetation types in Madagascar. In 2008, CI expanded the system to include Indonesia, Bolivia and Peru (Butler ). That same year CI created a complementary system providing daily flammability risk forecasts for the SW Amazon derived from NRT satellite estimates of temperature, precipitation and relative humidity (Steininger et al. ). Combined, these systems supported more than 1300 subscribers from 43 countries. In addition, from 2008 to 2012, CI analysts produced NRT email alerts of suspected forest encroachment for a subset of national parks in Indonesia and Madagascar based on manual interpretation of multi‐date pan‐sharpened Landsat imagery. In 2013, with support from NASA's Wildfires program, CI merged the fire monitoring components into Firecast (firecast.conservation.org), an enhanced monitoring and forecasting system incorporating email alerts with MODIS and VIIRS active fire data for Bolivia, Colombia, Indonesia, Madagascar, Peru and Suriname; daily flammability risks forecasts for the entire Amazon, and fire season severity forecasts for South America (Chen et al. ). In 2014, CI expanded the daily forest flammability risk forecasts to include Indonesia, with support from NASA and World Resource's Institute's Global Forest Watch. Firecast features an online interactive web map for fire data, a dashboard for analyzing and displaying statistics, and a downloadable mobile phone app [Firecast Onsight (http://firecast.conservation.org/OnSight)] for field‐based data collection and NRT visualization of active fires (Conservation International ).

With the exception of Brazil's forest monitoring systems (Diniz et al. ), there are few published assessments of operational NRT monitoring and alert systems for conservation effectiveness, particularly those examining their utility across multiple countries with unique environmental stressors and sociopolitical contexts. In this paper, we describe how CI's systems have been used by decision makers in four tropical forest countries to combat threats to forest resources and meet ambitious sustainable development goals, and biodiversity and CO₂ emissions targets. We highlight challenges to the application of these systems for forest management, and we make recommendations on how NRT forest monitoring and alert systems can be improved to facilitate conservation decision making.

Materials and Methods

To better understand the benefits and the challenges users experience integrating NRT fire and forest monitoring information into their work, as well as the contributions that continual enhancements to Firecast and its predecessors have been making to forest conservation, from 2008 to 2016 we conducted a series of an online surveys and in‐person interviews with subscribers of CI's NRT forest monitoring systems. For each online survey and in‐person interview we drafted a series of questions for users to: (1) provide feedback on how forest monitoring system technologies and data were employed; (2) describe the accomplishments and success stories these systems had facilitated; (3) recommend additional training and technical support needed to further build local capacity in their use; (4) recommend outreach activities needed to broaden the user base; and (5) suggest improvements, such as demand for other NRT products (e.g. burned area or aerosol data). The user feedback helped CI adapt to the evolving needs of the field‐based conservation and management. However, the specific questions varied among surveys and interviews, depending on the types of information we attempted to solicit from users at that moment in time.

Online surveys

In 2008, we conducted the first online survey involving more than 350 subscribers of the original fire alert system to evaluate its utility in Madagascar, eliciting responses from 43 data users. In 2011 and 2012, we circulated online surveys among 1108 subscribers in Madagascar, Indonesia, Bolivia and Peru, resulting in feedback from 118 users. In 2016, 3 years after Firecast's launch, we distributed a final online survey to which 125 individuals responded.

Interviews

From 2008 to 2013 we conducted a series of in‐person interviews with existing users of CI's forest monitoring systems to obtain a more in‐depth understanding of the real‐world uses and impacts of NRT satellite data, to identify priorities for future system improvements, and to assess the commitment of government institutions to use these data as part of their official business practices. In certain cases, we observed users applying the data to inform routine decisions.

In 2008, we interviewed representatives from five national governmental agencies and local conservation organizations in Madagascar as part of a USAID project supporting development of the fire alert system. In 2009, as part of a U.S. Department of State project combating illegal logging in Indonesia, we conducted interviews with subscribers from 10 national and local government agencies, international and national environmental organizations, and development and educational institutions. In 2013, under the auspices of a NASA Wildfires project we held interviews with Firecast users from 23 national and local government agencies, international and national environmental organizations, research institutions, commercial firms, and community groups in Bolivia, Peru, Indonesia and Madagascar.

The online surveys and in‐person interviews had significant limitations: they were not designed following social research methodologies; similar, but not identical questions were used in each survey; the sample sizes were small, and the target audiences varied. Nevertheless, they did provide important anecdotal information about how the fire alerts, flammability forecasting and suspected forest encroachment data were being used.

Results

Online survey results

In the 2008, 2011/2012 and 2016 surveys, users consistently reported that CI's NRT monitoring systems were of very high to moderately high utility for their work (Fig. ). The increase in the percentage of users reporting high utility from 2008 to 2011/2012 may have been due to increased acceptance and popularity of the systems within that time period; the subsequent decrease in the percentage of those reporting high utility and increase in those indicating moderate utility from 2011/2012 to 2016 may have resulted from the appearance of other NRT monitoring systems such as Global Forest Watch.

Overall utility of Firecast, and its predecessor the Fire Alert System (FAS), to users’ line of work based on 286 responses to a series of online surveys conducted in 2008, 2011/2012 and 2016. Firecast is a monitoring and forecasting system providing fire alerts and flammability risk forecasts for Bolivia, Colombia, Indonesia, Madagascar, Peru and Suriname.

CI's monitoring systems were also perceived as important for helping to deter undesirable fire or deforestation activity: In 2011/2012, 31% users found them to be very effective, 37% moderately effective, 16% marginally effective, 2% not at all effective, and 13% did not know. In 2016, users reported that Firecast was very useful or moderately useful for a wide range of conservation and management activities (Fig. ).

Usefulness of Firecast near real‐time monitoring tools (based on 286 survey responses), including fire alerts, daily forest flammability risk forecasts, fire season severity forecasts and suspected forest encroachment alerts, to the specific forest conservation and management activities of users representing international and national development and environmental organizations (44% of subscribers), national and local government agencies (28%), research and educational institutions (16%), commercial firms (11%) and the media (1%).

Country‐specific results

Madagascar

Feedback from surveys and interviews in Madagascar suggest that a diverse group is using fire monitoring data for a broad range of applications. Local forestry officials reported using daily fire alerts to notify ground verification teams of the locations of fires as part of their assessment of the extent of wildfire damage. Officials also reported using fire alerts to provide warnings to local fire brigades of larger fires underway, and said these data enabled them to focus their interventions on areas designated as ‘highly sensitive’ to fire, while compiling and reporting fire statistics to their superiors. Fire data were also incorporated into education and outreach activities with local communities by local fire brigades as part of their fire deterrence activities.

A national NGO described using fire data to evaluate an increase in areas susceptible to erosion by integrating active fire data into the Wischmeier and Smith's Empirical Soil Loss Model (Wischmeier and Smith ); the results were then presented to communities as part of awareness building on the impacts of fire and deforestation on their lands. Other NGOs reported using Firecast data to quantify impacts of fire on species health, to help delineate core protected areas, to calibrate deforestation models predicting forest loss as a function of fire occurrence for a forest carbon project in east‐central Madagascar, and as part of research into the links between climate change and fire intensity.

Members of the local offices of international NGOs described how they routinely forwarded fire alerts to field‐based colleagues working in intervention zones/landscapes, helping them adjust their field conservation strategies. Representatives from commercial entities revealed that the previous year's fire maps were being used to identify vulnerable areas after conservation actions had taken place, to raise awareness about fires, and to inform patrols when fires were occurring within their areas of operation in the event these were associated with illegal forest removal.

One unique application of active fire alerts was shared by a national conservation NGO engaging communities in Baly Bay, home to one of the world's most endangered tortoises, the ploughshare tortoise, whose habitat is under serious threat from fire. The NGO created a year‐long friendly competition between villages to see how effectively they could control fires threatening adjacent protected forests. Using Global Positioning System (GPS) coordinates of detected fires distributed by CI's alert system, the NGO monitored each village's success at managing wildfire, and at the end of the year convened a festival where the winners received financial aid for development projects of their choice, including school improvements, constructing wells for clean water, and purchasing solar panels for electricity (J. Durbin, pers. comm., 23 March 2013).

Indonesia

Interviews with Indonesian government agencies and NGOs revealed numerous applications of fire alerts and forest encroachment alerts. At the national level, government staff used active fire data from Firecast to cross‐validate data from other government sources (e.g. Indofire, NOAA, SiPongi). At the provincial level, authorities from a national park in southern Sumatra utilized Firecast in conjunction with NOAA fire data to determine whether intentional burning was occurring inside the protected area, and to take action against violators. NGOs from the same province used both fire and suspected encroachment alerts to optimize their forest protection activities related to Sumatran tiger conservation.

One international NGO reported that in August 2007 a fire alert led to expulsion of illegal loggers and squatters from Kerinci Seblat National Park (KSNP). Then, in August 2008, KSNP patrols responded to an alert and arrested 81 people for illegally clearing forests within the boundaries of the park. In May 2009, a major enforcement campaign was launched focusing on KSNP, where encroachment alerts showed evidence of new, suspected forest invasions and illegal logging.

Staff from a national park in central Sumatra used the encroachment data as evidence to confront encroachers. According to one interviewee, ‘some encroachers argue that they have been there a long time, and they demand land benefits’. The encroachment alerts allowed officials to present evidence of the precise timing of encroachment.

Bolivia

Fire risk data were extensively used by an environmental NGO working throughout the Department of Santa Cruz in development of its own fire risk prediction model, combining CI's flammability risk forecast with density of active fire data and wind speed and direction from NOAA's Global Summary of the Day product. This system enabled the NGO to engage with rural and indigenous communities, educating farmers on the risks of prescribed burns during peak fire weather conditions. The NGO has since developed their own fire danger forecasting model and operates an independent system for national fire monitoring and forecasting.

A national government institution responsible for forest management, enforcement and governance reported overlaying active fire data from NASA's Fire Information for Resource Management System (FIRMS) on to MODIS burn scar maps. They disseminated these maps to municipal directors who investigate active fires, fining landowners who violated forest management laws.

Peru

In Peru, an NGO dedicated to environmental and cultural conservation and sustainable development in the Amazon reported using Firecast to receive daily alerts and monthly statistics on fire activity in each municipality and within the focus areas of REDD+ forest carbon projects, enabling them to monitor trends in fire activity and the expansion of deforestation along the forest frontier.

A Peruvian forestry company confirmed that alerts contributed to their efforts to develop and sustainably manage forest plantations. Each day their Geographic Information System (GIS) technician received fire alerts, and when a fire appeared to threaten a plantation they immediately notified the field team responsible for control and vigilance; armed with this information the team would deploy personnel to confirm the severity of the event and its impacts. Fire information, ‘formed a spatial analysis package that is very valuable in the execution of our projects’.

A Peruvian non‐profit research and conservation organization reported downloading monthly statistical reports on fire activity in Peru, integrating these data into their internal monitoring system and channeling information from Firecast to field patrols involved in protecting and managing forests of importance. Once they received the alerts, they created maps for park patrols who verified the accuracy of the alerts. When they discovered illegal activity, they intervened in accordance with the country's protected areas law.

Challenges

In 2016, we asked Firecast subscribers about the barriers or challenges they encountered using NRT forest monitoring data in their work. Thirty percent of respondents stated that limited Internet access or slow access speeds were a major barrier; 12% noted difficulties in communicating the information to other stakeholders with whom they worked; 8% reported that Firecast downtimes and system errors prevented them from receiving monitoring data reliably; and 4% indicated that a lack of interest from their organization in utilizing these data was a barrier to maximizing their impact. Survey respondents also reported challenges related to limited budgets and insufficient technical capacity to fully integrate NRT data into their work; a lack of sufficient resources to support field staff such as forest patrols and trainers; and insufficient or intermittent political support from national and local governments. Some government agency subscribers disclosed that limited Internet access in remote regions was a particular problem, forcing them to use their personal email accounts via cellphone connections and thus restricting their access to timely fire information. Many institutions reported a persistent lack of computer resources and technical staff with GIS and remote sensing skills, limiting their ability to produce statistical reports demonstrating the need for increased resources to combat fire and deforestation. Even when NGOs and government institutions were able to disseminate data, limited budgets often led to insufficient patrol staff needed to respond to forest threats. Finally, survey respondents described a lack of continuity between governments. In Madagascar, under one administration a country‐wide system was established to track and build capacity for the management of fire. After a change in government, the entire fire monitoring system was dismantled by the new administration.

Discussion

After 15 years of developing and operating Firecast and its predecessors, several important lessons have emerged related to implementation of NRT monitoring systems to improve forest management in the tropics.

Bandwidth issues

Broadband Internet access in many countries is limited, particularly outside of major urban centers. It is therefore critical that NRT data distribution systems be optimized so that users are able to access notifications (e.g. email alerts) in a timely manner. Web‐based, content‐rich data portals with robust functionality, including interactive mapping, may operate too slowly for institutions constrained by low bandwidth Internet connections. Many users of NRT forest monitoring data spend extensive time in remote field locations without access to computers, where communications are limited to cell phones. Distribution of alerts and reports to these users is best performed via text messages, email alerts, or low‐bandwidth web‐based notifications and simple mobile applications [e.g. Firecast OnSight or Advanced Fire Information System (https://southernafrica.afis.co.za)].

Official buy‐in and ownership

One of the most significant barriers to widespread adoption of NRT forest monitoring systems by government agencies involves control and ownership of data. Governments often prefer to control forest monitoring data, particularly if public dissemination has legal or financial implications for influential constituents. Governments may resist adopting or endorsing data from sources managed by foreign entities, and may prohibit or suppress internal use of forest monitoring data from unofficial sources. It is imperative to establish collaborative relationships with government counterparts to understand their concerns and learn how to address them. Developers of forest monitoring systems may find it useful to provide government counterparts with ‘co‐ownership’, for example, by rebranding the website and associated emails with government logos and acknowledgments. Assisting governments with development of their own forest monitoring systems may also be appropriate.

Outreach and engagement

To maximize impact, forest monitoring systems require comprehensive community outreach and engagement, particularly in early stages when basic system functionality is under development. Outreach enables enhanced appreciation of the needs, technical limitations and barriers to entry of the potential user base. Outreach is also key to building a critical mass of users. For example, in Madagascar the fire alerts project team led a series of training workshops together with government officials in different regions of the country to increase the user base. They also ran advertisements in national newspapers with instructions on how to subscribe to the system, reaching a much larger pool of users.

Capacity building

Training practitioners to use high‐level products, such as those delivered in NRT monitoring and alert systems, facilitates the use of remote sensing data for conservation applications (Palumbo et al. ). Providing hands‐on capacity building in the manipulation and analysis of GIS and remote sensing data may significantly enhance an end‐user's ability to effectively incorporate these data into their business practices. For example, training targeted to conservation practitioners in Africa has successfully increased the application of remote sensing and GIS for conservation (de Klerk and Buchanan ). Focusing training on building the capacity of institutions to generate their own NRT data can be especially beneficial since it empowers them to produce information they control, and their enhanced technical capacity provides them with a better understanding of the limitations and potential uncertainties associated with forest observations from satellite.

User‐customization

Forest monitoring and data delivery systems that tailor analyses and alerts to a user's area‐of‐interest may significantly increase the relevance of these data for decision making. Filtering out non‐essential information and targeting only those areas relevant to a user's needs may lead to a reduction in data volumes transferred (important for users accessing data over low‐bandwidth connections or via cell phones), a decrease in the frequency in which users receive alerts and an increased likelihood that the alerts will be given proper attention once they are received.

Automated push‐based data delivery systems versus manual data access

Automated ‘push’‐based data delivery systems like Firecast allow data access with minimum effort, latency and bandwidth requirements. Manual data access systems such as interactive mapping web sites require query‐and‐display of monitoring data over a background image or map. Since manual data access systems require a user to repeatedly navigate to the website to obtain the latest information, push‐based alert systems increase data accessibility for rapid response.

Role of multiple forest monitoring systems

While NRT forest monitoring may eventually be consolidated into a limited number of powerful, multi‐functional monitoring systems serving data at global scales (Musinsky ), there continues to be a role for specialized forest monitoring systems providing access to data in unique ways. In our surveys we discovered that many users simultaneously access data through different systems (e.g. Firecast and FIRMS). The existence of multiple data delivery systems is not inherently bad: it creates redundancy, which may increase the likelihood that users continue to receive forest monitoring information if one system is retired; it provides users with the possibility of cross‐checking data for accuracy; and it enables opportunities for ‘re‐branding’, allowing governments to use officially sanctioned sources of data, even if the original data sources are the same.

Conclusion

Dissemination of NRT satellite‐based evidence of fire and illegal forest activity are used by governments, industry, non‐government organizations, and community‐based groups for many different purposes. Public access to NRT information helps increase transparency and serves as a metric to measure the effectiveness of institutions protecting and managing public forest resources; this in‐turn may lead to public pressure for improved governance and reform. A continuous stream of NRT alerts that track fire activity and other threats over time helps institutions design and implement management plans that lead to improved protection of forested lands. NRT forest monitoring systems are increasingly used by project managers as adaptive management tools supporting REDD+ forest carbon projects, and may contribute to sustainable commodity value chains by providing timely information to buyers showing where commercial crops and biofuels are planted, and whether they are cultivated using sustainable land use practices compliant with national laws and private sector sourcing policies. Finally, the success of NRT forest monitoring systems has led to unexpected results, such as decreases in the average size of deforestation patches in certain regions as loggers have learned to adapt (Faleiros ).

An ever‐growing number of NRT forest monitoring systems are introduced each year focused on fire, deforestation, illegal logging/encroachment and smoke/air quality. Some existing systems are evolving to integrate additional data streams and offer an expanding array of services, including social networking and mobile data exchange. With more satellite data options becoming available, greater opportunities for development of NRT forest monitoring and alert systems exist. As our study shows, NRT forest monitoring systems, while not the only technologies required to achieve ambitious global biodiversity, climate mitigation, and sustainable development targets, are arguably among the most essential.

Acknowledgments

We would like to thank Kelly Maynard, Zoe Cullen, Debby Martyr, Mahefa Rijasoa Rakotondranoa and Andy Keck for their assistance in building institutional partnerships; Marc Steininger and Daniel Juhn for their input into the design of our NRT forest monitoring systems; Diane Davies and Minnie Wong for providing access to MODIS active fire data; Rob Waller and Leanne Miller for their help in early implementation of email alerts; Keith Green, Howard Kistler, and Alex Dinnouti for software development of the Fire Alert System and Firecast; and Kellee Koenig, Bonie Dewantara and Hendi Sumantri for their assistance with training and technical support.

Conflicts of Interest

The authors declare that they have no conflicts of interest, financial or otherwise, that might be perceived as influencing the objectivity of the findings reported in this paper.