Country-wide high-resolution vegetation height mapping with Sentinel-2

Abstract

Sentinel-2 multi-spectral images collected over periods of several months were used to estimate vegetation height for Gabon and Switzerland. A deep convolutional neural network (CNN) was trained to extract suitable spectral and textural features from reflectance images and to regress per-pixel vegetation height. In Gabon, reference heights for training and validation were derived from airborne LiDAR measurements. In Switzerland, reference heights were taken from an existing canopy height model derived via photogrammetric surface reconstruction. The resulting maps have a mean absolute error (MAE) of 1.7 m in Switzerland and 4.3 m in Gabon (a root mean square error (RMSE) of 3.4 m and 5.6 m, respectively), and correctly estimate vegetation heights up to >50 m. They also show good qualitative agreement with existing vegetation height maps. Our work demonstrates that, given a moderate amount of reference data (i.e., 2000 km 2 in Gabon and ≈5800 km 2 in Switzerland), high-resolution vegetation height maps with 10 m ground sampling distance (GSD) can be derived at country scale from Sentinel-2 imagery.

Canopy Height Estimation from Single Multispectral 2D Airborne Imagery Using Texture Analysis and Machine Learning in Structurally Rich Temperate Forests

“…In addition to the high spatial resolution, Sentinel-2 provides a new image up to a frequency of 5 days. In a similar context is the representative example of [6], where Sentinel-2 multispectral images were used to regress per-pixel vegetation height for Gabon and Switzerland using a deep CNN. Suitable spectral and textural features that were applied resulted in a mean absolute error (MAE) of 1.7 m in Switzerland and 4.3 m in Gabon, correctly estimating heights up to 50 m.…”

“…Confusion matrices (see Tables 5 and 6) correspond to the highest accuracies acquired for both the six (6) and the four (4) height classes, respectively. High values of producer's (89.63% and 88.15% PA) and user's (77.73% and 76.13% UA) accuracy are observed for the class [5,40) or FPH, reflecting low values of omission and commission errors, respectively; the majority of [5,40) classes were classified correctly, as can be seen.…”

“…Remote sensing advancements emerging at an unprecedented rate seem to be well-performing, while providing accurate information on vegetation structure. In this context, numerous approaches have been proposed in the literature to estimate canopy height [5,6].…”

A framework for large-scale mapping of human settlement extent from Sentinel-2 images via fully convolutional neural networks

“…Additionally, there are two drop-out layers to avoid model overfitting to the training data, given that the goal is to map HSE globally. No additional pooling layers are used to avoid the information loss during downsampling process, which is also the design idea in [34] and [52]. As defined, the output prediction is with a 20-meter GSD, while the input data is with a 10-meter GSD; thus no upsampling layers are used.…”

“…The fundamental advantage of all these deep neural networks is their ability for enhanced feature representation and pixel-level recognition. Examples where convolutional neural networks (CNN) and, in particular, FCNs are used for remote sensing image classification or segmentation include [24,25,26,27,28,29,30,31,32,33,34,35]. Apart from the works focusing on very high resolution satellite or aerial imagery (i.e., with a ground sampling distance equal to or even less than 1 m), data of lower spatial resolution is also being studied, since the images of lower resolutions such as globally openly available Sentinel-2 imagery remain the key candidates for large-scale mapping [36,37].…”