Spatioformer: A Geo-encoded Transformer for Large-Scale Plant Species Richness Prediction

Yiqing Guo,Karel Mokany,Shaun R. Levick,Jinyan Yang,Peyman Moghadam

2024-10-25

Abstract:Earth observation data have shown promise in predicting species richness of vascular plants ($\alpha$-diversity), but extending this approach to large spatial scales is challenging because geographically distant regions may exhibit different compositions of plant species ($\beta$-diversity), resulting in a location-dependent relationship between richness and spectral measurements. In order to handle such geolocation dependency, we propose Spatioformer, where a novel geolocation encoder is coupled with the transformer model to encode geolocation context into remote sensing imagery. The Spatioformer model compares favourably to state-of-the-art models in richness predictions on a large-scale ground-truth richness dataset (HAVPlot) that consists of 68,170 in-situ richness samples covering diverse landscapes across Australia. The results demonstrate that geolocational information is advantageous in predicting species richness from satellite observations over large spatial scales. With Spatioformer, plant species richness maps over Australia are compiled from Landsat archive for the years from 2015 to 2023. The richness maps produced in this study reveal the spatiotemporal dynamics of plant species richness in Australia, providing supporting evidence to inform effective planning and policy development for plant diversity conservation. Regions of high richness prediction uncertainties are identified, highlighting the need for future in-situ surveys to be conducted in these areas to enhance the prediction accuracy.

Machine Learning

What problem does this paper attempt to address?

### What problem does this paper attempt to solve?

This paper aims to solve the challenges encountered in predicting plant species richness (i.e., α - diversity) on a large spatial scale. Specifically, Earth observation data shows potential in predicting plant species richness, but it is difficult to extend it to a large spatial scale because geographically distant regions may exhibit different plant species compositions (i.e., β - diversity), resulting in a location - dependent relationship between richness and spectral measurements.

To solve this problem, the authors proposed the **Spatioformer** model, which encodes geographical location information into remote - sensing images by combining a new - type geocoder and the Transformer model. This can better handle differences in plant species compositions at different geographical locations, thereby improving the accuracy of large - scale species - richness prediction.

### Main research objectives

1. **Improve large - scale plant species - richness prediction**: By introducing a geographical location encoder, enhance the prediction performance of existing models on a large scale.
2. **Reveal the spatial distribution pattern of plant species richness in Australia**: Use the richness maps generated by Spatioformer to analyze the spatio - temporal dynamic changes of plant species richness in Australia.
3. **Guide future field investigations**: Identify areas of high uncertainty and recommend more field investigations in these areas in the future to improve prediction accuracy.

### Method overview

- **Spatioformer model**: Combines the Transformer and a geocoder and is able to handle location - dependent features.
- **Geocoder**: Encodes geographical coordinates using multi - scale sine functions (such as sine and cosine functions) to ensure that each pixel has a unique geographical location identifier.
- **Experimental setup**: Use Landsat satellite images and ground - measured samples for training and evaluate the model performance through cross - validation.

Through this method, the authors hope to more accurately predict plant species richness on a large scale and provide strong support for ecological protection and policy - making.