Role of Sampling Design When Predicting Spatially Dependent Ecological Data With Remote Sensing

Abstract

Remote sensing opens opportunities to assess spatial patterns on ecological data for a wide range of ecosystems. This information can be used to more effectively design sampling strategies for fieldwork, either to capture the maximum spatial dependence related to the ecological data or to completely avoid it. The sampling design and the autocorrelation observed in the field will determine whether there is a need to use a spatial model to predict ecological data accurately. In this article, we show the effects of different sampling designs on predictions of a plant trait, as an example of an ecological variable, using a set of simulated hyperspectral data with an increasing range of spatial autocorrelation. Our findings show that when the sample is designed to estimate population parameters such as mean and variance, a random design is appropriate even where there is strong spatial autocorrelation. However, in remote sensing applications, the aim is usually to predict characteristics of unsampled locations using spectral information. In this case, regular sampling is a more appropriated design. Sampling based on close pairs of points and clustered over a regular design may improve the accuracy of the training model, but this design generalizes poorly. The use of spatially explicit models improves the prediction accuracy significantly in landscapes with strong spatial dependence. However, such models have low generalization capacities to extrapolate to other landscapes with different spatial patterns. When the combination of design and size results in sample distances similar to the range of the spatial dependence in the field, it increases predictions uncertainty.

A map of global peatland extent created using machine learning (Peat-ML)

“…To provide estimates of the error associated with the LightGBM predictions we adopted a blocked-leave-one-out (BLOO) strategy, which is recommended for applications where the predictors could be expected to exhibit spatial autocorrelation (Roberts et al, 2017;Meyer et al, 2019;Ploton et al, 2020). BLOO tends to produce estimates of prediction error that are closer to the 'true' error (Roberts et al, 2017), particularly in cases where the sampling strategy is clustered (Rocha et al, 2021).…”

“…were not able to choose systematic (Rocha et al, 2021) or feature-based sampling strategies that could be more optimal for peatland prediction. Our approach would also benefit from greater availability of processed, global-scale products that should be sensitive to water status below the peat surface like L-band SAR (e.g.…”

Modelling hourly evapotranspiration in urban environments with SCOPE using open remote sensing and meteorological data

“…One reason is the necessity to train the model at representative locations and conditions, which is a challenge in urban areas due to the constantly changing land cover captured by the tower's footprint and lack of flux towers at different surfaces in a highly fragmented and heterogeneous environment (Feigenwinter et al, 2018). In addition, most of the widely used empirical models are unsuitable for variables with strong spatiotemporal dependency, such as ET (Rocha et al, 2018(Rocha et al, , 2020.…”

Modelling hourly evapotranspiration in urban environments with SCOPE using open remote sensing and meteorological data

“…One reason is the necessity to train the model at representative locations and conditions, which is a challenge in urban areas due the constantly changing land cover captured by the tower's footprint and lack of flux towers at different surfaces in a highly fragmented and heterogeneous environment (Feigenwinter et al, 2018). In addition, most of the widely used empirical models are unsuitable for variables with strong spatiotemporal dependency such as ET (Rocha et al, 2018(Rocha et al, , 2020.…”

“…Surface Energy Balance (SEB) model versions such as Surface Energy Balance Algorithm for Land (SEBAL), Surface energy balance index (SEBI) and Surface Energy Balance System (SEBS) include variables retrieved from remote sensing such as land surface temperature, albedo, and net radiation, but still require meteorological data Nouri et al, 2015;van der Tol and Norberto, 2012). SEBAL models which include RS derived parameters, such as LAI, are therefore semiempirical and spatially dependent, reducing the capacity to generalise to other locations (Rocha et al, 2020). Although many input variables for modelling ET are currently derived from optical and thermal sensors, ET cannot be empirically derived exclusively from remote sensing as it does not directly change the spectral reflectance (Timmermans et al, 2013;van der Tol and Norberto, 2012).…”