Satellite-based forest monitoring: spatial and temporal forecast of growing index and short-wave infrared band

Abstract

For detecting anomalies or interventions in the field of forest monitoring we propose an approach based on the spatial and temporal forecast of satellite time series data. For each pixel of the satellite image three different types of forecasts are provided, namely spatial, temporal and combined spatio-temporal forecast. Spatial forecast means that a clustering algorithm is used to group the time series data based on the features normalised difference vegetation index (NDVI) and the short-wave infrared band (SWIR). For estimation of the typical temporal trajectory of the NDVI and SWIR during the vegetation period of each spatial cluster, we apply several methods of functional data analysis including functional principal component analysis, and a novel form of random regression forests with online learning (streaming) capability. The temporal forecast is carried out by means of functional time series analysis and an autoregressive integrated moving average model. The combination of the temporal forecasts, which is based on the past of the considered pixel, and spatial forecasts, which is based on highly correlated pixels within one cluster and their past, is performed by functional data analysis, and a variant of random regression forests adapted to online learning capabilities. For evaluation of the methods, the approaches are applied to a study area in Germany for monitoring forest damages caused by wind-storm, and to a study area in Spain for monitoring forest fires.

Methods for Mapping Forest Disturbance and Degradation from Optical Earth Observation Data: a Review

“…Also the mapping examples (MEs) given in Section 4 are listed in the table. [26], [27], [28] [29], [30] [31], [32], [33], ME2 [34] [35] [36,37], [38] Defoliation, leaf discoloration ME1 [39] [40], [31], ME1 [39] [35], [41] [40], [42] Logging roads, skid trails, gaps [43], [28], [44], [45] [46]…”

“…Study area 2 is located in northern Spain where forest fires frequently occur. For the spatio-temporal approach, overall accuracies of 76% in the Spanish and 94% in the German study site were achieved [34].…”

“…A known shortcoming of single curve fitting, as described above, is that standard underlying statistical assumptions, such as data normality and equal variance, are generally difficult to meet [70] and violation of these assumptions results in an inadequate representation of the data by the fitted function. Functional time series analysis as a statistical approach for curve fitting has been applied for monitoring storm damages and forest fires based on dense MODIS time series [34]. In addition to deriving the trend component, curve fitting also captures the underlying dynamic of seasonality.…”

Forest Stand Species Mapping Using the Sentinel-2 Time Series

“…However, combining individual images that achieve the highest accuracies does not necessarily lead to a high accuracy of the classification with combined images, and the timing of image acquisition is more important than the image quantity [35]. One of the newest optical satellite missions, Sentinel-2, can significantly improve forest mapping [36]. The Sentinel-2 sensor, MultiSpectral Instrument (MSI), acquires data in, among others, three red-edge bands, which are very useful in providing information about vegetation, e.g., chlorophyll content [37,38].…”

Spatio-temporal statistics: applications in epidemiology, veterinary medicine and ecology

“…Spatial and temporal forecast of satellite time series data in the field of forest monitoring was addressed by Bayr et al (2016). They employed several methods of functional data analysis, including functional principal component analysis, and a novel form of random regression forests with online learning capability.…”