Estimating melt pond bathymetry from aerial images using photogrammetry

Abstract

<p>Melt ponds play a key role for the summery energy budget of the Arctic sea-ice surface. Observational data that enable an integrated understanding and improved formulation of the thermodynamic and hydrological pond system in global climate models are spatially and temporally limited.</p><p>Previous studies of shallow water bathymetry of riverbeds and lakes, experimental studies above sea ice and increasing availability of high-resolution aerial sea ice imagery motivated us to investigate the possibilities to derive pond bathymetry from photogrammetric multi-view reconstruction of the summery ice surface topography.</p><p>Based on dedicated flight grids and simple assumptions we were able to obtain pond depth with a mean deviation of 3.5 cm compared to manual in situ observations. The method is independent of pond color and sky conditions, which is an advantage over recently developed radiometric retrieval methods.</p><p>We present the retrieval algorithm, including requirements to the data recording and survey planning, and a correction method for refraction at the air— pond interface. In addition, we show how the retrieved elevation model synergize with the initial image data to retrieve the water level of each individual pond from the visually determined pond exterior.</p><p>The study points out the great potential to derive geometric and radiometric properties of the sea-ice surface emerging from the increasingly available image data recorded from UAVs or aircraft.</p>

Sea ice melt pond bathymetry reconstructed from aerial photographs using photogrammetry: A new method applied to MOSAiC data

“…After drainage, pond bottoms even of previously deep ponds surfaced, although pond level was close to sea level before. Both these observations are indications for a flexible ice cover (as shown in Fuchs, 2023c) and contradict the traditional assumption of a rigid ice cover on which ponds only reduce to their lowest points during the drainage event (e.g., Popović et al, 2020).…”

“…Orthomosaic and DEMs of the entire MOSAiC leg 4 floe are available from 30 June 2020, 17 July 2020 and 22 July 2020 (Neckel et al, 2022). They were compiled using the methods described in Neckel et al (2023) and Fuchs (2023c) and cropped here to the extent of 2 km 2 x 2 km 2 around the floe. The floe edge was retraced manually in QGIS (QGIS Development Team, 2020) to eventually retrieve statistical data only within the floe area.…”

“…The floe edge was retraced manually in QGIS (QGIS Development Team, 2020) to eventually retrieve statistical data only within the floe area. Classification into three main surface types (ice/snow, ponds and open water) was derived applying the sea ice image classification tool PASTA-ice (Fuchs, 2023c) on the brightness corrected orthomosaics (l2 data) on 17 July 2020 and 22 July 2020 and on the brightness corrected orthomosaic with cloud correction (l2b data) on 30 June 2020.…”

“…In the following, we shortly summarize the algorithm due its importance for the workflow automation. Details are found in Fuchs (2023c).…”

“…For spatial analysis, main class objects are converted to polygon geometries defining the outer and, if present, inner edges of the object. They also include an attribute table, that contains information on the sub-class proportions and a classification confidence proxy from the prediction probability output of the classifier (Fuchs, 2023c). Classification recall and precision are rather limited for the very specific sea-ice surface subclasses listed in Table 3 due to overlaps in appearance, but high for the combined main classes (Fig.…”