Active fires are devastating natural disasters that cause socio-economical damage across the globe. The detection and mapping of these disasters require efficient tools, scientific methods, and reliable observations. Satellite images have been widely used for active fire detection (AFD) during the past years due to their nearly global coverage. However, accurate AFD and mapping in satellite imagery is still a challenging task in the remote sensing community, which mainly uses traditional methods. Deep learning (DL) methods have recently yielded outstanding results in remote sensing applications. Nevertheless, less attention has been given to them for AFD in satellite imagery. This study presented a deep convolutional neural network (CNN) “MultiScale-Net” for AFD in Landsat-8 datasets at the pixel level. The proposed network had two main characteristics: (1) several convolution kernels with multiple sizes, and (2) dilated convolution layers (DCLs) with various dilation rates. Moreover, this paper suggested an innovative Active Fire Index (AFI) for AFD. AFI was added to the network inputs consisting of the SWIR2, SWIR1, and Blue bands to improve the performance of the MultiScale-Net. In an ablation analysis, three different scenarios were designed for multi-size kernels, dilation rates, and input variables individually, resulting in 27 distinct models. The quantitative results indicated that the model with AFI-SWIR2-SWIR1-Blue as the input variables, using multiple kernels of sizes 3 × 3, 5 × 5, and 7 × 7 simultaneously, and a dilation rate of 2, achieved the highest F1-score and IoU of 91.62% and 84.54%, respectively. Stacking AFI with the three Landsat-8 bands led to fewer false negative (FN) pixels. Furthermore, our qualitative assessment revealed that these models could detect single fire pixels detached from the large fire zones by taking advantage of multi-size kernels. Overall, the MultiScale-Net met expectations in detecting fires of varying sizes and shapes over challenging test samples.

Accurate crop yield Mapping is paramount in agricultural monitoring and food security. In this study, we present a comprehensive investigation into estimating winter wheat yield in the Qazvin plane of Iran, leveraging the synergy between machine learning algorithms and the fusion of remote sensing data from radar and optical satellite sensors. The research is based on the availability of high-quality in situ yield data gathered by the Ministry of Agriculture in collaboration with the Food and Agriculture Organization (FAO), collected during the 2019–2020 crop year. The study area encompasses the Qazvin plane, an agriculturally significant region renowned for winter wheat production in Iran. In-situ data from various agricultural fields and seed types as reference measurements enabled us to conduct rigorous validation of the performance of machine learning algorithms and the effectiveness of the fused remote sensing data. The primary objective of this study is to assess and compare the performance of seven prominent machine learning algorithms for accurate estimation of the annual winter wheat yields. Furthermore, we investigate the individual and synergistic capabilities of radar and optical satellite sensors in estimating winter wheat yield. Through rigorous analysis of the pixel-level confusion matrices, we identify the most effective model for yield estimation, evaluating the complementarity and information redundancy between the two types of remote sensing data. In this study, we conducted an extensive comparison of various machine learning algorithms for winter wheat crop yield estimation in the Qazvin plane of Iran. Among the four best-performing algorithms examined, namely polynomial regression (RMSE = 0.5657 $\mathrm{t}/{\mathrm{h}\mathrm{a}}^{-1}$), random forest (RMSE = 0.1632 $\mathrm{t}/{\mathrm{h}\mathrm{a}}^{-1}$), XGBoost (RMSE = 0.3153 $\mathrm{t}/{\mathrm{h}\mathrm{a}}^{-1}$), and the proposed Multi-Layer Perceptron (MLP) (RMSE = 0.1324 $\mathrm{t}/{\mathrm{h}\mathrm{a}}^{-1}$), the MLP demonstrated superior performance. The MLP’s yield estimation exceeded the total yearly agricultural statistics of Qazvin by 0.19 percent. However, this discrepancy can be attributed to various factors, including errors in wheat and barley field mapping, miscalculation in cumulative statistics, and the inherent limitations of yield estimation algorithms in capturing the dynamic nature of agricultural systems. The findings of this research provide valuable insights into the potential of machine learning algorithms and remote sensing data fusion for accurate crop yield estimation, paving the way for enhanced agricultural monitoring and decision-making processes in the region.