Cloud and snow identification in remote sensing images is critical for snow mapping and snow hydrology research. Aimed at the problem that the semantic segmentation model is prone to producing blurred boundaries, slicing traces and isolated small patches for cloud and snow identification in high-resolution remote sensing images, the feasibility of combining DeepLab v3+ and conditional random field (CRF) models for cloud and snow identification based on GF-1 WFV images is studied. For GF-1 WFV images, the model training and testing experiments under the conditions of different sample numbers, sample sizes and loss functions are compared. The results show that, firstly, when the number of samples is 10,000, the sample size is 256 × 256, and the loss function is the Focal function, the model accuracy is the optimal and the Mean Intersection over Union (MIoU) and the Mean Pixel Accuracy (MPA) reach 0.816 and 0.918, respectively. Secondly, after post-processing with the CRF model, the MIoU and the MPA are improved to 0.836 and 0.941, respectively, compared with those without post-processing. Moreover, the misclassifications such as blurred boundaries, slicing traces and isolated small patches are significantly reduced, which indicates that the combination of the DeepLab v3+ and CRF models has high accuracy and strong feasibility for cloud and snow identification in high-resolution remote sensing images. The conclusions can provide a reference for high-resolution snow mapping and hydrology applications using deep learning models.

Based on the daily precipitation data during 1960–2016 at 72 stations and the daily streamflow data during 1956–2016 at 7 hydrological stations in the Huai River Basin (HRB), China, eco-surplus and eco-deficit under influences of abrupt streamflow behaviors were analyzed using Flow Duration Curve (FDC). The relations between indicators of hydrological alteration (IHA) and ecological indicators (Shannon Index, SI) were quantified, investigating impacts of altered hydrological processes on the evaluations of the ecological instream flow. Besides, we also quantified fractional contributions of climatic indices to nonstationary ecological instream flow using the Generalized Additive Models for Location Scale and Shape (GAMLSS) framework. While the possible impact of human activities on ecological instream flow will be revealed based on land use changes data. The results indicated that: (1) FDC is subject to general decrease due to hydrological alterations, and most streamflow components are lower than 25% FDC. We found increased eco-deficit and decreased eco-surplus due to altered hydrological processes. The FDC of the streamflow in the main stream of the HRB is lower than that along the tributaries of the HRB. Eco-surplus (eco-deficit) changes are in good line with precipitation anomaly changes during the Spring, Autumn and Winter periods. However, the hydrological alterations due to hydrological regulations by the reservoirs are the primary cause behind the mismatch between ecological instream flow and precipitation anomalies during summer; (2) Annual and seasonal eco-surplus (eco-deficit) is decreasing (increasing) and that during winter season is an exception. Although higher eco-surplus in winter than in other seasons, the eco-surplus is decreasing persistently and the 21st century witnessed the lowest eco-surplus along the main stream of the HRB. Meanwhile, the Shannon index indicated decreased ecological diversity across the HRB; (3) The ecological instream flow is highly sensitive to The Pacific Decadal Oscillation (PDO), North Atlantic Oscillation (NAO) and Niño 3.4 Sea Surface Temperature Index (Nino3.4). Meanwhile, the ecological instream flow along the mainstream of the HRB is highly sensitive to climate indices. While the ecological instream flow by GAMLSS model has better fitting performance in describing the extreme values and local trends.