Interpolating missing land cover data using stochastic spatial random forests for improved change detection

Abstract

Forest cover requires large scale and frequent monitoring as an indicator of biodiversity and progress towards United Nations and World Bank Sustainable Development Goal 15. Measuring change in forest cover over time is an essential task in order to track and preserve quality habitats for species around the world. Due to the prohibitive expense and impracticality of mass field data collection to monitor forest cover at regular intervals, satellite images are a key data source for monitoring forest cover globally. A challenge of working with satellite images is missing data due to clouds. Existing methods for interpolating the missing data based on past images, such as compositing, are effective for stable land cover but can be inaccurate for dynamic and substantially changing landscapes. Here we present an adaptation of our recent stochastic spatial random forest (SS-RF) method, which combines observed data from a prior image and modelled estimates of the current image to produce interpolated land cover values and associated probabilities of those values. Results show our SS-RF method accurately detected simulated land cover change under both clear felling (0.83 average overall accuracy) and tree thinning (0.85 average overall accuracy). Our method detected forest cover change substantially more accurately than compositing, offering 39% and 12% increases in average overall accuracy for clear felling and tree thinning simulations respectively. However, when natural fluctuation occurs and there is minimal change in land cover, compositing has equivalent or more accurate performance than our method. Overall we find that our SS-RF method produces accurate estimates under a range of simulated forest clearing scenarios and has a more accurate and robust performance than compositing when modelling noticeably changing landscapes.

Mixture of multivariate Gaussian processes for classification of irregularly sampled satellite image time-series

“…This strategy was used to produce the first land use and land cover maps in France [12] and is now commonly used at large scale. Yet, for other applications, e.g., changes detection, the interpolation algorithm choice may be more critical than for pixel-wise classification [72].…”

Spatiotemporal Pattern of Invasive Pedicularis in the Bayinbuluke Land, China, during 2019–2021: An Analysis Based on PlanetScope and Sentinel-2 Data

“…Therefore, it is necessary to use pixel-based classification to extract Pedicularis, and feature engineering is a particularly useful method. Some work in the literature has indicated that spectral features and combined vegetation indices are more essential than textural features and principal components in target identification and classification [41,42].…”

Reconstructing missing data by comparing interpolation techniques: Applications for long‐term water quality data

“…Random forests can identify and rank the variables of importance for making predictions (Tyralis et al 2019), which can aid ecologists in recognizing which measured variables to retain in long‐term monitoring, recognize if unmeasured (latent) variables exist and caused low prediction accuracy, and determine ecological relationships. Another neat application can use random forest prediction errors in space–time, such as particular aquatic habitats or periods in time, which may reveal habitat heterogeneity, spatiotemporal variability and dynamics (Chiao et al 2012; Louvet et al 2016; Vizcaino et al 2016) and detect change under a range of simulated scenarios (Holloway‐Brown et al 2021). Random forest outputs can also link conceptually to places with high ecological resiliency or risk of ecological state transitions that could inform aquatic management and restoration priorities (Delaney and Larson, in review).…”