Land surface temperature (LST) is a critical parameter for many applications. A default aerosol type and a fixed aerosol loading were commonly used in the development of LST retrieval algorithms. However, numerical simulations showed that a significant bias up to -3 K was found at nadir view if dust aerosol was not considered in the LST retrieval algorithm. To reduce the influence of dust aerosol on LST estimation, a three-channel algorithm was proposed based on a widely used two channel algorithm for clear-sky conditions. With simulated data of MODIS channels 29, 31, and 32, the LST could be estimated with a root mean squire error (RMSE) of 1.4 K when viewing zenith angle (VZA) was 0° and aerosol optical depth (AOD, at 0.55 μm) less than 1.0; the maximum RMSE was 2.3 K and 2.2 K for the day-time and night-time algorithm, respectively, when VZA less than 60°. In case of clear-sky conditions, namely AOD=0, the RMSE was less than 1.2 K and 1.8 K when VZA=0° and 60°, respectively. It indicates that the proposed three-channel algorithm can be used to estimate LST not only for dust aerosol skies but also for clear-sky conditions. In addition, the influences of adjency effect induced by the scatter of dust aerosol are also studied, it shows that with considering the adjency affect the RMSE of LST retrieval algorithm increased about 0.5 K.

Considering the significant influence of clouds on thermal infrared (TIR) data sea surface temperature (SST) retrieval, this study focuses on the reduction of cloud influence on Moderate Resolution Imaging Spectroradiometer (MODIS) longwave SST retrieval. First, the quality level (QL) 3 MODIS SSTs are classified into cloudy or clear-sky pixels based on MODIS cloud mask products. The cloudy SST pixels, flagged as confident cloudy or probably cloudy in the cloud masks, are further classified into three groups according to their associated daytime cloud-top and optical properties. Taking the SSTs measured by 11 buoys over one year as the reference data, those three groups of cloudy SSTs are significantly underestimated, with biases of −33.45 °C, −6.35 °C, and −4.72 °C. The QL 3 clear-sky SSTs are identified as probably and confident clear in the cloud masks and are not influenced by clouds, with a bias of nearly 0 °C. Then, three support vector regression (SVR) models are individually proposed for the three groups of cloudy SSTs. Taking the cloud-top and optical parameters as inputs of the proposed SVR models, we can obtain SSTs with a bias of 0 °C and a root-mean-square error (RMSE) of less than 1.6 °C for the three groups of cloudy SSTs. The method proposed in this study shows the potential for TIR SST estimation under some cloudy conditions using satellite remote sensing cloud products. Considering the RMSE of 0.4 °C in operational sea surface temperature and sea ice analysis, further study is needed before actual applying the method proposed in this study.

The high resolution visible images from space borne or airborne platforms are widely used for earthquake building damage degree or type identification. But the visible images cannot be used for damage detection during night time. Considering the availability of thermal infrared (TIR) images to get land surface information all day long, this study try to study the possibility of using UAV based TIR image to detect earthquake damaged buildings. The study area is Beichuan county earthquake ruins protection site of the ‘5.12’ Wenchuan earthquake in 2008, which located in Qushan Town, Beichuan Qiang Autonomous County, Mianyang city, Sichuan Province, China. And the TIR image are acquired using UAV of DJI M200 version 2 with the payload of ZENMUSE XT2. The study shows that the texture features in TIR images can be used for earthquake building damage degree detection with the accuracy of 73.4% and 71.6%, respectively, for the training and test dataset.

Landslide detection from high-resolution satellite imagery plays a significant role in disaster management. Recently, deep learning has emerged as one of the most powerful tools for landslide detection. However, the existing deep learning models for landslide detection still have room for improvement. In this paper, we propose a novel deep learning model named DCA-Net for the automatic detection of landslides. This model adopts an asymmetric encoder-decoder structure. The core module designed in the DCA-Net is the dual-channel depthwise block, integrating two parallel channels of asymmetric depthwise separable convolutions and residual connections to enlarge the receptive field and improve the performance. Experiments are conducted on the open-source Bijie landslide dataset. Several state-of-the-art networks are also employed for quantitative and qualitative comparisons. The results indicate that our proposed DCA-Net is superior to other deep learning models, which can be well utilized in landslide detection from aerial images.

It is well know that for the same intensity areas, the buildings with different structure types can show different vulnerabilities. Thus, building structure type is one the key parameters for rapid estimation of casualties and injuries after earthquake, which is vital for emergency response and rescue. To estimate building structure types, the buildings are firstly extracted based on the spectrum, texture, and height information of UAV visible images. Then, the structure type of individual extracted buildings is classified using convolution neural network. To evaluate the accuracy of the proposed method, the images of Xuyi county, Huai'an City, Jiangsu Province are acquired using a small rotorcraft UAV. The results show that the user accuracy and cartography accuracy are 80.69% and 78.42%, respectively.

Because of the thin optical depth, the cirrus cloud is one the clouds that commonly overlooked as clear sky by the cloud mask products. To study the influence of cirrus clouds on the satellite thermal infrared (TIR) data estimated sea surface temperature (SST), the MODIS Terra SST products are used in this study. Six buoy stations measured sea temperatures are acquired from National Oceanic and Atmospheric Administration's National Data Buoy Center (NDBC). In addition, the cirrus optical depth are calculated based on the cirrus reflectance of MODIS cloud mask products. The SST pixels with MODIS cloud products labeled as confident clear-sky and cirrus reflectance flag is cirrus or contrail pixel are taken as cirrus pixels. Taking buoy sea temperatures as the true values, the MODIS SST errors are between 1.2 K and -0.6 K. In addition, five stations' MODIS SST are tend to underestimated as cirrus optical depth increase.

The quantitative study of active faults is highly dependent on high-precision and high-resolution topographic data. Though Light Detection and Ranging (LiDAR) technology can provide such data, its high cost greatly limits its use in many geoscience applications. Recently, the Structure from Motion (SfM) photogrammetry shows a great potential to provide topographic information with high precision, but at significantly lower costs than the laser scanning survey. In this study, the applicability of SfM photogrammetry method in modeling the topography of fault zone was investigated by using images acquired with a low-cost digital camera mounted on an UAV. The resolution and accuracy of the SfM-derived topographic data was evaluated in detail using existing airborne LiDAR data as a benchmark. The results show that the SfM photogrammetry method can produce a point cloud with the density seventy times higher than the airborne LiDAR. Furthermore, considering the errors in LiDAR data itself, and the precision of the SfM-derived point cloud is comparable to that of the LiDAR point cloud, demonstrating that the SfM photogrammetry method is an inexpensive and effective alternative to airborne LiDAR for the topography modeling of fault zone.

The accuracy of traditional spectral based land surface classification method can be decreased for areas with similarity spectral characters for different land cover types. To extract building distributions in those areas, a method based on the height information using unmanned aerial vehicle (UAV) optical remote sensing data were proposed in this study. With Digital Surface Model (DSM) and Digital Elevation Model (DEM) data produced by EasyUAV software using UAV acquired remote sensing images, the differences between DSM and DEM larger than 2.6 m was taken as a threshold for building extraction. It showed that the user accuracy and mapping accuracy of the building extraction method proposed in this study were 88.69% and 97.42%, respectively. To evaluate the performance of the proposed method, the buildings were also extracted based on traditional supervised classification method using Digital Orthophoto Map (DOM) produced by EasyUAV. The results showed that the new method was more accurate with the user accuracy and mapping accuracy of the supervised classification method of 43.23% and 85.30%, respectively.

Soil moisture is a key state variable which plays a significant role in many hydrological processes. The Soil Moisture Active Passive (SMAP) mission was launched on 31 January 2015 which can provide global information of soil moisture. Among the released SMAP data sets, the Level 4 Surface and Root Zone Soil Moisture Analysis Product (L4_SM) can not only provide information on surface soil moisture (top 5 cm of the soil column), but also provide estimates of root zone soil moisture (top 1 m of the soil column) which is very important for several key applications targeted by SMAP. However, since this product has been released only for a short time, its accuracy and reliability has not been validated so far. In this study, we evaluated the L4_SM soil moisture analysis product against in-situ soil moisture measurements collected from the Little Washita Watershed network located in southwest Oklahoma in the Great Plains region of the United States. The results show that both the surface and root zone soil moisture estimates in the L4_SM product are in good agreement with the in-situ measurements, and the RMSE is 0.027 m3/m3 and 0.032 m3/m3 for the surface and root zone soil moisture respectively which both have exceeded the RMSE requirement of 0.04 m3/m3 for this product.

The atmospheric aerosols are one of the uncertainties that significantly influence the study of energy balance at regional or global scale. However, few studies focus on the aerosols caused by earthquake induced damage of hazard-affected bodies. To study the influences of earthquakes on the loadings of atmospheric aerosols, the AErosol RObotic NETwork (AERONET) sites retrieved aerosol optical depth (AOD) are used in this study. It showed that the AOD became larger after earthquakes for six cases in this study. After approximately one day, the AOD of the six cases tend to close to the normal range. In addition, the other four cases show that the there is no significant increasement of AOD or the AOD tend to become smaller after earthquake.

Because of the complex influences of cirrus clouds on the estimation of Land Surface Temperature (LST), the traditional LST retrieval algorithms can only be used for clear-sky conditions and there is no LST when the pixel is identified as clouds by cloud mask algorithm. To retrieve LST under cirrus clouds, a three-channel algorithm what is dependent on cirrus optical depth (COD) and effective radius was proposed. The simulated data showed that the daytime LST could be retrieved using the three-channel algorithm with a root mean square error of less than 3.0 K when COD (at 12 μm) was less than 0.7 and viewing zenith angle was less than 60°. Compared with the results of the traditional clear-sky two-channel LST retrieval algorithm, where the maximum RMSE was 17.8 K, the algorithm proposed in this study could significantly improve the accuracy of the daytime LST retrieved using satellite thermal-infrared data.

Land surface temperature (LST) is a critical parameter for numerical weather forecasting, drought monitoring, water resources management and global climate change studies. Because of the supercooled temperature, the cirrus cloud can significantly reduce the LST retrieved from thermal infrared data. This paper focused on analyzing and reducing the influence of thin cirrus cloud on the accuracy of LST retrieved using the generalized split-window (GSW) algorithm. A correction method was proposed with the LST retrieval error expressed as linear functions of cirrus optical depth (COD). The slopes of the linear functions were further written as the combination of the difference and mean of two used channels emissivities and cirrus cloud top height (CTH). The results showed that the LST retrieval accuracy could be significantly improved with root mean square error (RMSE) of LST changing from 14.4 K before LST error correction to 1.8 K after LST error correction for COD equivalent to 0.3.