Monitoring tropical forest dynamics using Landsat time series and community-based data

<p>Tropical forests cover a significant portion of the earth's surface and provide a range of</p> <p>ecosystem services, but are under increasing threat due to human activities. Deforestation</p> <p>and forest degradation in the tropics are responsible for a large share of global CO2</p> <p>emissions. As a result, there has been increased attention and effort invested in the</p> <p>reduction of emission from deforestation and degradation and the protection of remaining</p> <p>tropical forests in recent years. Methods for tropical forest monitoring are therefore vital</p> <p>to track progress on these goals. Two data streams in particular have the potential to</p> <p>play an important role in forest monitoring systems. First, satellite remote sensing is</p> <p>recognized as a vital technology in supporting the monitoring of tropical forests, of which</p> <p>the Landsat family of satellite sensors has emerged as one of the most important. Owing</p> <p>to its open data policy, a large range of methods using dense Landsat time series have</p> <p>been developed recently which have the potential to greatly enhance forest monitoring</p> <p>in the tropics. Second, community-based monitoring is supported in many developing</p> <p>countries as a way to engage forest communities and lower costs of monitoring activities.</p> <p>The development of operational monitoring systems will need to consider how these data</p> <p>streams can be integrated for the effective monitoring of forest dynamics.</p> <p>This thesis is concerned with the monitoring of tropical forest dynamics using a combi-</p> <p>nation of dense Landsat time series and community-based monitoring data. The added</p> <p>value conferred by these data streams in monitoring deforestation, degradation and re-</p> <p>growth in tropical forests is assessed. This goal is approached from two directions. First,</p> <p>the application of econometric structural change monitoring methods to Landsat time</p> <p>series is explored and the efficacy and accuracy of these methods over several tropical</p> <p>forest sites is tested. Second, the integration of community-based monitoring data with</p> <p>Landsat time series is explored in an operational setting. Using local expert monitoring</p> <p>data, the reliability and consistency of these data against very high resolution optical</p> <p>imagery are assessed. A bottom-up approach to characterize forest change in high the-</p> <p>matic detail using a priori community-based observations is then developed based on</p> <p>these findings.</p> <p>Chapter 2 presents a robust data-driven approach to detect small-scale forest disturbances</p> <p>driven by small-holder agriculture in a montane forest in southwestern Ethiopia. The</p> <p>Breaks For Additive Season and Trend Monitoring (BFAST Monitor) method is applied</p> <p>to Landsat NDVI time series using sequentially defined one-year monitoring periods. In</p> <p>addition to time series breakpoints, the median magnitude of residuals (expected versus</p> <p>observed observations) is used to characterize change. Overall disturbances are mapped</p> <p>with producer's and user's accuracies of 73%. Using ordinal logistic regression (OLR)</p> <p>models, the extent to which degradation and deforestation can be separately mapped is</p> <p>explored. The OLR models fail to distinguish between deforestation and degradation,</p> <p>however, owing to the subtle and diffuse nature of forest degradation processes.</p> <p>Chapter 3 expands upon the approach presented in Chapter 2 by tracking post-disturbance</p> <p>forest regrowth in a lowland tropical forest in southeastern Peru using Landsat Normalized</p> <p>Difference Moisture Index (NDMI) time series. Disturbance between 1999 and 2013 are</p> <p>mapped using the same sequential monitoring method as in Chapter 2. Pixels where</p> <p>disturbances are detected are then monitored for follow-up regrowth using the reverse of</p> <p>the method employed in Chapter 2. The time of regrowth onset is recorded based on a</p> <p>comparison to defined stable history period. Disturbances are mapped with 91% accuracy,</p> <p>while post-disturbance regrowth is mapped with a total accuracy of 61% for disturbances</p> <p>before 2006.</p> <p>Chapter 4 and 5 explore the integration of community-based forest monitoring data and</p> <p>remote sensing data streams. Major advantages conferred by community-based forest dis-</p> <p>turbance observations include the ability to report on drivers and other thematic details</p> <p>of forest change and the ability to detect low-level forest degradation before these changes</p> <p>are visible above the forest canopy. Chapter 5 builds on these findings and presents a</p> <p>novel bottom-up approach to characterize forest changes using local expert disturbance</p> <p>reports to calibrate and validate forest change models based on Landsat time series. Using</p> <p>random forests and a selection of Landsat spectral and temporal metrics, models describ-</p> <p>ing forest state variables (deforested, degraded or stable) at a given time are produced.</p> <p>As local expert data are continually acquired, the ability of these models to predict forest</p> <p>degradation are shown to improve.</p> <p>Chapter 6 summarizes the main findings of the thesis and provides a future outlook, given</p> <p>the prospect of increasing availability of satellite and in situ data for tropical forest mon-</p> <p>itoring. This chapter argues that forest change methods should strive to utilize satellite</p> <p>time series and ground data to their maximum potential. As “big data" emerges in the</p> <p>field of earth observation, new data streams need to be accommodated in monitoring</p> <p>methods. Operational forest monitoring systems that are able to integrate such diverse</p> <p>data streams can support broader forest monitoring goals such as quantitative monitoring</p> <p>of forest dynamics. Even with a wealth of time series based forest disturbance methods</p> <p>developed recently, forest monitoring systems require locally calibrated forest change esti-</p> <p>mates with higher spatial, temporal and thematic resolution to support a variety of forest</p> <p>monitoring objectives.</p> <p> </p> <p>