Search

Timely and accurate crop yield information can ensure regional food security. In the field of predicting crop yields, deep learning techniques such as long short-term memory (LSTM) and convolutional neural networks (CNN) are frequently employed. Many studies have shown that the predictions of models combining the two are better than those of single models. Crop growth can be reflected by the vegetation index calculated using data from remote sensing. However, the use of pure remote sensing data alone ignores the spatial heterogeneity of different regions. In this paper, we tested a total of three models, CNN-LSTM, CNN and convolutional LSTM (ConvLSTM), for predicting the annual rice yield at the county level in Hubei Province, China. The model was trained by ERA5 temperature (AT) data, MODIS remote sensing data including the Enhanced Vegetation Index (EVI), Gross Primary Productivity (GPP) and Soil-Adapted Vegetation Index (SAVI), and a dummy variable representing spatial heterogeneity; rice yield data from 2000–2019 were employed as labels. Data download and processing were based on Google Earth Engine (GEE). The downloaded remote sensing images were processed into normalized histograms for the training and prediction of deep learning models. According to the experimental findings, the model that included a dummy variable to represent spatial heterogeneity had a stronger predictive ability than the model trained using just remote sensing data. The prediction performance of the CNN-LSTM model outperformed the CNN or ConvLSTM model. [ABSTRACT FROM AUTHOR]

Total phosphorus (TP) concentration is high in countless small inland waterbodies in Hubei province, middle China, which is threating the water environment. However, there are almost no ground-based water quality monitoring points in small inland waterbodies, because the cost of time, labor, and money is high and it does not meet the needs of spatiotemporal dynamic monitoring. Remote sensing provides an effective tool for TP concentration monitoring spatiotemporally. However, monitoring the TP concentration of small inland waterbodies is challenging for satellite remote sensing due to the inadequate spatial resolution. Recently, unmanned aerial vehicles (UAV) have been applied to quantitatively retrieve the spatiotemporal distribution of TP concentration without the challenges of cloud cover and atmospheric effects. Although state-of-the-art algorithms to retrieve TP concentration have been improved, specific models are only used for specific water quality parameters or regions, and there are no robust and reliable TP retrieval models for small inland waterbodies at this time. To address this issue, six machine learning methods optimized by intelligent optimization algorithms (IOA-ML models) have been developed to quantitatively retrieve TP concentration combined with the reflectance of original bands and selected band combinations of UAV multispectral images. We evaluated the performances of models in terms of coefficient of determination (R2), root mean squared error (RMSE), and residual prediction deviation (RPD). The results showed that the R2 of the six IOA-ML models for training, validation, and test sets were 0.8856–0.984, 0.8054–0.8929, and 0.7462–0.9045, respectively, indicating the methods had high precision and transferability. The extreme gradient boosting optimized by genetic algorithm (GA-XGB) performed best, with the highest precision for the validation and test sets. The spatial distribution of TP concentration of each flight derived from different models had similar distribution characteristics. This paper provides a reference for promoting the intelligent and automatic level of water environment monitoring in small inland waterbodies. [ABSTRACT FROM AUTHOR]

Forest logging detection is important for sustainable forest management. The traditional optical satellite images with visible and near-infrared bands showed the ability to identify intensive timber logging. However, less intensive logging is still difficult to detect with coarse spatial resolution such as Landsat or high spatial resolution in fewer spectral bands. Although more high-resolution remote sensing images containing richer spectral bands can be easily obtained nowadays, the questions of whether they facilitate the detection of logging patterns and which spectral bands are more effective in detecting logging patterns, especially in selective logging, remain unresolved. Therefore, this paper aims to evaluate the combinations of visible, near-infrared, red-edge, and short-wave infrared bands in detecting three different logging intensity patterns, including unlogged (control check, CK), selective logging (SL), and clear-cutting (CC), in north subtropical plantation forests with the random forest algorithm using Sentinel-2 multispectral imagery. This study aims to explore the recognition performance of different combinations of spectral bands (visual (VIS) and near-infrared bands (NIR), VIS, NIR combined with red-edge, VIS, NIR combined with short-wave infrared bands (SWIR), and full-spectrum bands combined with VIS, NIR, red edge and SWIR) and to determine the best spectral variables to be used for identifying logging patterns, especially in SL. The study was conducted in Taizishan in Hubei province, China. A total of 213 subcompartments of different logging patterns were collected and the random forest algorithm was used to classify logging patterns. The results showed that full-spectrum bands which contain the red-edge and short-wave infrared bands improve the ability of conventional optical satellites to monitor forest logging patterns and can achieve an overall accuracy of 85%, especially for SL which can achieve 79% and 64% for precision and recall accuracy, respectively. The red-edge band (698–713 nm, B5 in Sentinel-2), short-wave infrared band (2100–2280 nm, B12 in Sentinel-2), and associated vegetation indices (NBR, NDre2, and NDre1) enhance the sensitivity of the spectral information to logging patterns, especially for the SL pattern, and the precision and recall accuracy can improve by 10% and 6%, respectively. Meanwhile, both clear-cutting and unlogged patterns could be well-classified whether adding a red-edge or SWIR band or both in VIS and NIR bands; the best precision and recall accuracies for clear-cutting were enhanced to 97%, 95% and 81%, 91% for unlogged, respectively. Our results demonstrate that the optical images have the potential ability to detect logging patterns especially for the clear-cutting and unlogged patterns, and the selective logging detection accuracy can be improved by adding red-edge and short-wave infrared spectral bands. [ABSTRACT FROM AUTHOR]