Improving Remote Species Identification through Efficient Training Data Collection

Abstract

Plant species identification and mapping based on remotely-sensed spectral signatures is a challenging task with the potential to contribute enormously to ecological studies. Success in this task rests upon the appropriate collection and use of costly field-based training data, and researchers are in need of ways to improve collection efficiency based on quantitative evidence. Using imaging spectrometer data collected by the Carnegie Airborne Observatory for hundreds of field-identified tree crowns in Kruger National Park, South Africa, we developed woody plant species classification models and evaluated how classification accuracy increases with increasing numbers of training crowns. First, we show that classification accuracy must be estimated while respecting the crown as the basic unit of data; otherwise, accuracy will be overestimated and the amount of training data needed to perform successful classification will be underestimated. We found that classification accuracy and the number of training crowns needed to perform successful classification varied depending on the number and spectral separability of species in the model. We also used a modified Michaelis-Menten function to describe the empirical relationship between training crowns and model accuracy, and show how this function may be useful for predicting accuracy. This framework can assist researchers in designing field campaigns to maximize the efficiency of field data collection, and thus the amount of biodiversity information gained from remote species identification models.

Tree Species Abundance Predictions in a Tropical Agricultural Landscape with a Supervised Classification Model and Imbalanced Data

“…Only the 20 species with 20 crowns or more were analyzed as single-species classes. A large increase in overall model accuracy was seen when including only species with more than 20 crowns ( Figure S2) and is similar to thresholds in crown numbers used in other studies [15]. All crowns of the 24 species with fewer than 20 crowns were grouped together into a mixed species class called "Others".…”

“…Many studies have explored the effect of the spectral uniqueness of species at multiple scales on the success of classification models [9,14,15]. These studies highlight the contributions of crown structure, phenology, and leaf chemistry of a species' unique spectral signature to spectral separability.…”

“…These studies highlight the contributions of crown structure, phenology, and leaf chemistry of a species' unique spectral signature to spectral separability. Other studies have explored the scope of the model, quantifying the decline in prediction accuracy with an increase in the number of classes [16] and smaller class sample sizes [17], in addition to providing guidelines for the optimal sample size needed to achieve maximum accuracy given the number of classes in the model [15].…”

“…The ideal labeled dataset for supervised species classification models should span the spatial extent of the image to capture the full variation in species' spectra [22], and include many crowns of all species, acknowledging that the exact number is highly dependent on the spectral uniqueness of all species [15]. Field datasets will often reflect the landscape distribution of species abundances, where the sample sizes may be small and uneven across all species.…”

“…To isolate the effects of changes in the number of species and sample sizes, studies have standardized the sample size for each species, giving species equal weight in the model [15][16][17]. In reality, equal sample size across all species is generated by either time-consuming field efforts to generate a large dataset of the rare species, or by subsampling the data from the large classes, which with random sample selection, may result in the removal of potentially useful data points for classifying the common species.…”