Wetlands are extremely dynamical systems and their behavior depends on the characteristics of the surroundings (topography, geology and vegetation, among others) as well as on meteorological and hydrological processes. Wetlands receive groundwater through diffuse upwelling and through springs. Studying upwelling is of great importance to e.g. evaluate the overall ecology or capacity to remove nitrate of the wetland system. One problem is that diffuse upwelling is difficult locate and measure. We analyze the temporal dynamics of a groundwater-fed wetland in central Jutland (Denmark) by the use of a range of thermal methods across a lowland stream valley. A monitoring system consisting of Distributed Temperature Sensing (DTS), wells with temperature depth profiles and thermal infrared (TIR) imaging on an unmanned aerial vehicle, in conjunction with hydrological and atmospheric data, provide a quasi 3D time-lapse characterization of the thermal behavior of the system, both on the ground and in the subsurface, over a period of two years. We infer potential locations of groundwater upwelling to the land surface by studying the temperature in both the wetland surface and the groundwater. Each thermal method provides different, partially overlapping estimates of the upwelling location and magnitude, highlighting the need to incorporate classic hydrological metrics to constrain the results obtained using heat as a tracer. The integration of these data indicates that temperature measurements can be used to study groundwater upwelling in stream valleys.

Prior information regarding subsurface patterns may be used in geophysical inversion to obtain realistic subsurface models. Field experiments require prior information with sufficiently diverse patterns to accurately estimate the spatial distribution of geophysical properties in the sensed subsurface domain. A variational autoencoder (VAE) provides a way to assemble all patterns deemed possible in a single prior distribution. Such patterns may include those defined by different base training images and also their perturbed versions, e.g. those resulting from geologically consistent operations such as erosion/dilation, local deformation and intrafacies variability. Once the VAE is trained, inversion may be done in the latent space which ensures that inverted models have the patterns defined by the assembled prior. Inversion with both a synthetic and a field case of cross-borehole GPR traveltime data shows that using the VAE assembled prior performs as good as using the VAE trained on the pattern with the best fit, but it has the advantage of lower computation cost and more realistic prior uncertainty. Moreover, the synthetic case shows an adequate estimation of most small scale structures. Estimation of absolute values of wave velocity is also possible by assuming a linear mixing model and including two additional parameters in the inversion.