Geodaysit 2023

Authors: Claudio Ladisa, Manuel A. Aguilar, Alessandra Capolupo, Eufemia Tarantino, Fernando J. Aguilar.
The use of renewable energy sources in power generation is increasing due to environmental awareness and technological advancements. Solar energy, with its extensive availability and minimal greenhouse gas emissions, is a promising source. However, large photovoltaic (PV) plants require constant monitoring to ensure efficiency and reliability. Remote sensing technology can be beneficial in providing accurate information on the plant's size, shape, and location, reducing costs and increasing monitoring efficiency. The detection of large PV plants can be carried out using various technologies, including the use of satellite imagery, drones imagery or observation from aircraft. However, the use of satellite imagery is advantageous for the detection of large PV plants because it allows to acquire data on a large area without having to move around the site and to monitor the plant over time without interfering with its activity. Open-source imagery from satellites like Sentinel-2 (S2) and Landsat 9 has led to a significant increase in remote sensing research related to extracting (PV) systems. This is because the free and public availability of high-quality images with extensive spatial coverage has eliminated the need to buy costly private satellite images. Additionally, the frequency of image acquisition, which can occur every few days, has allowed for quick and accurate monitoring of areas of interest. Several research have recently merged remote sensing with machine learning (ML) methods to develop automatic classification algorithms for PV systems. Most of these algorithms employ different spectral indices, such as the Normalized Difference Water Index (NDWI), the Normalized Difference Vegetation Index (NDVI), and Normalized Difference Bare Index (NDBI), as input. These spectral indices provide useful information on the presence of water, vegetation and bare soil, respectively, which can be used to identify PV systems more accurately, thus improving classification accuracy. However, there is no specific spectral index that has been tested exclusively for the extraction of PV. This is partially because PV arrays may be constructed on many kinds of surfaces, in various environmental and climatic circumstances, and with different solar panel sizes and types. In this regard, the goal of this work was to suggest a Photovoltaic Systems Extraction Index (PVSEI) for the detection of PV installations from S2 images in two distinct study areas characterized by the persistent presence of large PV installations: The province of Viterbo (Italy) and the province of Seville (Spain). The development of the PVSEI was based on the combination of different bands provided by S2, in order to maximise the spectral difference between the solar panels and their surroundings. For each study area, two S2 images, one taken in February and the other one in August, were used to analyse the seasonal variation of the solar panels' spectral signature and test the PVSEI's accuracy in each of the four scenarios. The image analysis was carried out using an Object-Based Image Analysis (OBIA) method since it allowed for a more accurate identification of PV systems than the pixel-based method, which analyzes individual elements without taking their spatial arrangement and semantic significance into account. Multi-resolution segmentation was used to create segments with different dimensions based on scale, shape and compactness parameters. The Decision Tree (DT) classifier was used to evaluate the effectiveness of the PVSEI and its importance in comparison to the other indices used in the literature in both locations and for both periods after the objects had been labelled as "PV" and "No-PV”. The effectiveness of the new index was demonstrated through the results obtained from the DT analysis. In three out of four scenarios, the PVSEI was selected as the first cut in the DT analysis. In the remaining scenario where it was not ranked first, it still maintained a high level of significance, being the second index in importance. The accuracy was assessed using an error matrix calculated on both the entire segmentation dataset (i.e. using all the objects) and with TTA mask with 2 m pixel size. Four metrics were used to evaluate accuracy of the PVSEI, including Overall Accuracy (OA), Kappa Index of Agreement (KIA), Producer Accuracy (PA), and User Accuracy (UA) for both classes. OA exceeded 98% in all scenarios, both for the segmentation dataset and the TTA mask. KIA values for the TTA mask ranged from 0.81 to 0.86, while values for the segmentation objects ranged from 0.74 to 0.82. In conclusion, the new index has demonstrated favourable outcomes in both study areas, with only a limited number of misclassifications involving bare soil objects that have a spectral signature resembling certain photovoltaic systems.

Autores: Abderrahim Nemmaoui, Fernando J. Aguilar, Manuel A. Aguilar
Forests act as important carbon sinks, therefore being key components of the global carbon cycle. The carbon dioxide emissions account is essential for climate regulation policies and the evaluation of the effects of these policies, as well as for understanding the services they provide to societies.
Traditionally, forest inventories are completed by ground-based expert crews. These field surveys are uneconomical, time consuming and not adequate for studies dealing with periodic data collection. Consequently, one of the key topic in forest applications is to find an effective method to produce effective and accurate inventories.
In recent years, Remote Sensing (RS) has proven to be capable of providing independent, timely and reliable forest information. RS data are used to estimate several forest variables of silvicultural interest such as crown diameter (CD), tree height (H), diameter at breast height (DBH) and aboveground biomass (AGB). In this sense, and due to its ability to estimate attributes at tree level, LiDAR derive point cloud data has become a valuable data source in the field of efficient and accurate detection and segmentation of individual trees (IT).
State-of-the-art approaches use different algorithms for individual tree segmentation (ITS). For each algorithm, a specific methodology to create the input Canopy Height Model (CHM) and/or many parameters should be tuned to somehow adapt the segmentation algorithm to each particular forest stand. This approach makes the results highly dependent on the applied local fitting parameters, which implies difficulties when applied for large-scale mapping. In addition, the parameter setting process is quite time consuming and requires learning and understanding the meaning and role of each parameter.
The main goal of this work aims at developing a pipeline that requires minimal user interaction when working on large areas of Mediterranean forests. The expected results should facilitate the production of broad-extend IT maps and extract the corresponding dendrometric parameters from low-density airborne laser scanning (ALS) data without spending time tuning algorithm parameters.
The study area was located in Sierra de María-Los Vélez Natural Park (Almeria, Spain). Up to 38 reference square plots of 25 m side containing reforested stands of Aleppo pine (Pinus halepensis Mill.) with variable density, tree height and presence of shrubs and low vegetation mainly represented by little holm oak trees (Quercus ilex L.). This forest composition and structure make up a forest typology that is very representative of the Mediterranean forests.
Three open source raster-based (i.e., CHM-based) were tested to extract tree location and some dendrometric parameters such as tree H and CD. The first algorithm is the method proposed by Dalponte & Coomes(2016) adapted and introduced in the package lidR (Roussel et al.2020). The second one is the algorithm developed by Silva et al.(2016), which is focused on the way to better approximating the intersecting canopy of multiple trees after locating treetops by local maxima. The last algorithm tested is included in the library Digital Forestry Toolbox (DFT). In addition, the point cloud-based algorithm proposed by Li et al.(2012) was also tested.
For every algorithm tested, we tried different parameters to find the best pipeline, finally obtaining up to 4024 combinations of all tested algorithms for each experimental plot. For each setting, tree detection accuracy was assessed by computing the detection rate, and the commission and omission errors. Some statistics, such as median, RMSE and relative RMSE, were also used to quantitatively assess the accuracy of tree H and CD estimates over each reference plot.
The IT detection accuracy rates, in terms of precision, recall, and F1-score, showed the successful performance of the pipeline proposed in this study. The algorithm proposed by Li et al.(2012) showed detection F1-score average values of 82.65% (using the same parameter combination for the 38 experimental plots). However, it failed in delimiting the crown diameter (relative RMSE 57.06% and Pearson r of 0.55). The method developed by Silva et al.(2016), when applied on a CHM generated with the point-to-raster algorithm and using a LM based on a variable Tree Window Size (TWS), presented a similar F1-score for ITS (i.e., 82.53%), but being most successful delimiting the crown (relative RMSE 22.21% and Pearson r of 0.68). Finally, Dalponte & Coomes(2016) and DFT methods showed slightly worse results, with average F1-scores of 80.41% and 75.66%, respectively.
The results obtained confirms the usefulness of low-density ALS data to both detect IT and estimate H and CD, also underlining some key aspects regarding the choice of the correct method and parameters to perform single tree detection for Aleppo pine in large areas of Mediterranean forests.