Depth determination in shallow water area is critical to model for instance, a detailed shoreline position, a change in beach topography, and the potency of beach erosion. Multispectral images can provide a complete map of areas that are difficult to map by conventional hydrographic surveys due to their logistics difficulties and limited spatial coverage. The aim of this study is to evaluate the effect of various training and testing set ratios to model bathymetric data from remotely sensed imagery. This research applied three methods to derive shallow water bathymetric data tested on two subsets located along Tanjung Kelayang coastal areas. The methods combined echo-sounding measurements and the reflectance of blue, green, red and near infrared of Sentinel 2A image with 10 m spatial resolution. In the experiment, the echo-sounding data set was split into training and testing set in three different ratios to see the effect of these various training and testing ratios to the accuracy of all algorithms. From the results, we can see that all models perform well in estimating bathymetric data for the shallow water depth, however, the accuracies were slightly changing by the variation of the training and testing data included in the model. In general, all methods provide a comparable performance for shallow water depth with RMSE less than 1 m and can be used effectively for deriving accurate and updated medium resolution bathymetric maps.

Abstract. Satellite-Derived Bathymetry (SDB) has been used in many applications related to coastal management. SDB can efficiently fill data gaps obtained from traditional measurements with echo sounding. However, it still requires numerous training data, which is not available in many areas. Furthermore, the accuracy problem still arises considering the linear model could not address the non-relationship between reflectance and depth due to bottom variations and noise. Convolutional Neural Networks (CNN) offers the ability to capture the connection between neighbouring pixels and the non-linear relationship. These CNN characteristics make it compelling to be used for shallow water depth extraction. We investigate the accuracy of different architectures using different window sizes and band combinations. We use Sentinel-2 Level 2A images to provide reflectance values, and Lidar and Multi Beam Echo Sounder (MBES) datasets are used as depth references to train and test the model. A set of Sentinel-2 and in-situ depth subimage pairs are extracted to perform CNN training. The model is compared to the linear transform and applied to two other study areas. Resulting accuracy ranges from 1.3 m to 1.94 m, and the coefficient of determination reaches 0.94. The SDB model generated using a window size of 9x9 indicates compatibility with the reference depths, especially at areas deeper than 15 m. The addition of both short wave infrared bands to the four visible bands in training improves the overall accuracy of SDB. The implementation of the pre-trained model to other study areas provides similar results depending on the water conditions.

ABSTRACT:Since its first inception, aerial photography has been used for topographic mapping. Large-scale aerial photography contributed to the creation of many of the topographic maps around the world. In Indonesia, a 2013 government directive on spatial management has re-stressed the need for topographic maps, with aerial photogrammetry providing the main method of acquisition. However, the large need to generate such maps is often limited by budgetary reasons. Today, SfM (Structure-from-Motion) offers quicker and less expensive solutions to this problem. However, considering the required precision for topographic missions, these solutions need to be assessed to see if they provide enough level of accuracy. In this paper, a popular SfM-based software Agisoft PhotoScan is used to perform bundle adjustment on a set of large-scale aerial images. The aim of the paper is to compare its bundle adjustment results with those generated by more classical photogrammetric software, namely Trimble Inpho and ERDAS IMAGINE. Furthermore, in order to provide more bundle adjustment statistics to be compared, the Damped Bundle Adjustment Toolbox (DBAT) was also used to reprocess the PhotoScan project. Results show that PhotoScan results are less stable than those generated by the two photogrammetric software programmes. This translates to lower accuracy, which may impact the final photogrammetric product.

Measuring the status of achievement of the SDGs is the task and concern of many countries in the world, including Indonesia. Indicators for achieving the SDGs enclose three main pillars, namely environmental, economic, and social. The updated land use/land cover information is needed for environmental pillars. One imperative land cover information is built-up land, which acts as a detector for expanding urban areas and measuring SDGs' target indicators. Indonesia's cultural diversity affects the distribution pattern of built-up land, especially settlements. This is a challenge in the up-to-date and rapid mapping of built-up land. This research aims to analyze the ability and transferability of the Random Forest model for built-up areas and settlements using Google Earth Engine (GEE) in Banyumas, Cilacap, and Tegal. Around 19 predictors from multi-sources satellites are integrated to identify four land cover classes. Discussion on predictor composition to improve model accuracy also carried on. The results showed that the algorithm separated four land cover classes, with the highest accuracy for separating water bodies and other classes (vegetation and open land), OA above 90%. Machine confusion regarding the separation between housing classes and other buildings was still found (F1 score 0.67 - 0.69). Applying the model to the other two areas resulted in a similar statistical trend to the trained model. However, the classification method developed in this paper can assist in the rapid description of land cover if up-to-date data from official sources are not available.