Landslides are increasing in mountainous regions of the world, causing significant threats to human lives and the economy. Landslide Susceptibility Mapping (LSM) is vital for delineating areas prone to future landslides. However, the LSM preparation often encounters limited data availability, causing a major obstacle in understanding the occurrence of landslides and the environmental factors responsible. This article attempts to bridge the gap of limited data availability using a transfer learning approach wherein a random forest approach is applied. The knowledge obtained from the data-rich region is applied in the scarce region. The results depicted that factors such as distance to road, distance to river and lineament were among major contributing factors. Furthermore, the value of statistical measures such as AUC-ROC, precision, and F-score increased significantly for the LSM when trained from source data. Overall, the study successfully addresses the role of transfer learning approach in LSM preparation, especially in data-scarce regions.

The Indo-Gangetic plains (IGP) suffers from poor air quality and high atmospheric pollution over the years. Various natural and anthropogenic sources are the causes of the poor air quality, affecting the population in the IGP and surrounding areas. The local and long-range transport of dust affects the air quality of the IGP region and sometimes depending upon the favorable meteorological conditions dust reaches the Himalayan foothills. region. We have carried out the analysis of aerosol properties and air quality using a Microtop Sun photometer and low-cost air quality sensors at the IIT Mandi campus (high altitude site) and Dhampur (rural site). The observed$\text{PM}_{2.5}$and aerosol optical depth (AOD) at IIT Mandi show a strong influence of dust, however, the$\text{PM}_{2.5}$and AOD are found to be lower compared to the values observed at Dhampur (located 200 km northeast of Delhi). Local industries, dust, and crop residue burning are the main sources of pollution in the rural areas of the northern parts of India. The low-cost sensors provide good quality data from IIT Mandi and Dhampur, our results show the need for a dense network in rural places to understand the dynamics of pollutants, especially in the IGP, one of the polluted regions in India.

In this study we compared Land Surface temperature (LST) computed from the high spatial resolution Landsat 8 and the high temporal resolution MODIS Aqua and Terra satellites to in situ near-surface air temperature (Tair) provided by IMD from 2013 to 2020. This study was carried out on Bhuntar station located in data sparse Kullu valley of Higher Himalayas. After spatiotemporal analysis with the IMD station, a total of 117 clear sky Landsat 8 scenes and 1771 clear sky MODIS scenes were analyzed using Google Earth Engine (GEE). LST was computed on these scenes on GEE using the Statistical Mono-Window (SMW) technique. The satellite derived LST values were extracted at the Bhuntar station and compared with IMD in-situ data. The result revealed strong correlation (R2around 0.84) between Landsat-8 LSTs and Tair. The averaged Night time LSTs obtained from MODIS Aqua and Terra satellites performed better than averaged Day time LSTs as the RMSE is lower during night time as compared to day time. The correlation between Landsat 8 derived LST is more in case of maximum Tairthan mean Tair. Based on these correlation, we proposed empirical equations for calculating air temperature from remotely observed LSTs. We expect that our findings will be valuable in bridging data gaps in in situ monitoring with daily resolution.

Due to significant increase in landslide activity in all over the world, landslide susceptibility mapping has proven to be an effective tool for mitigation and management of this problem. Present study aims to prepare landslide susceptibility map of Mandi district using frequency ratio method. Total 981 landslide points were identified of past occurrence and were randomly divided into 697 points as training and 294 points as testing. Eight causative factors (slope, aspect, distance to road, distance to streams, distance to lineament, lithology, geomorphology, distance to road and elevation) were identified for landslide. Based on the methods lithology, geomorphology, and aspect were identified important to generate landslides. LSI was prepared using the FR values and were classified into five zones using quantile, natural break, and equal interval classification. For validation the Area under Curve (AUC) for all the classification types were calculated out of which equal interval classification found to be most accurate.

Recently, some spectral–spatial band selection (BS) strategies have become hugely popular as they fuse the spectral information of the pixels and the spatial relationship with the neighboring pixels to enhance the performance of classification methods. However, being unsupervised in nature, these methods do not utilize the class information of the training samples which could substantially empower the capabilities of such spectral–spatial BS methods. To circumvent this limitation, a supervised spectral–spatial BS method based on component loadings obtained from the principal components of spectral–spatial principal component analysis (PCA) and using a novel super-pixel based graph Laplacian embedding is proposed. The methodology attempts to unify the two strategies of dimensionality reduction, i.e., BS and feature extraction (FE), so that the benefits from both of them can be combined. The importance of each band is estimated in terms of its component loadings along the principal components which are estimated from a unified objective function consisting of three terms: data fidelity, classification error term, and spatial prior. Additionally, the spatial relationship among the neighboring samples is characterized using a novel superpixel-based graph model. An objective comparison of the proposed approach with several widely used, state-of-the-art BS methods demonstrates a significant improvement in the classification accuracy.

Machine learning methods require a large amount of training data, however, the landslides do not occur everywhere and the number of landslide occurrences are limited in an area. This leads to a small number of landslide samples and a higher number of non-landslide samples. This underrepresented data and severe skewness in class distribution create an imbalance for learning algorithms, which becomes biased towards the majority class and have a low performance on the minority class. We have used two algorithms namely EasyEnsemble and BalanceCascade for reducing the imbalance in data. The balanced data is used with SVM to generate landslide susceptibility zonation maps. The results of the study show that SVM with balanced data has major improvements in the preparation of susceptibility maps over imbalanced data.

In the last three last decades air pollution level has been increased significantly especially in the northern parts of India with Delhi being the most polluted city. Population growth, energy demand, industrial growth, increased crop burning, increased traffic density and coal burning have affected the air quality of Delhi and surrounding areas upto a great extent. In the present paper, we have used MODIS Terra and Aqua data to study the annual and seasonal variations of Aerosol Optical Depth (AOD) and Angstrom Exponent (AE) and also tried to identify the sources of pollution based on AOD and AE. A wide variation in AOD, (range 0.07 to 3.49) and AE, (range 0-1.8) is observed during 2000-2019 over Delhi and surrounding regions. This shows very poor air quality and highly polluted atmospheric. Different aerosol types are found to be season dependent based on the detailed analysis of AOD and AE retrieved from satellite data during the period 2000-2019. The present analysis shows dominance of anthropogenic aerosols (AA) during winter whereas biomass burning (BB) during post monsoon seasons. During pre monsoon season pollutant continental (PC) and mostly dust (MD) were observed for many days as air mass travel from Thar Desert, Middle East and Arabia with the westerly winds. In the recent decade, clean days in Delhi are reduced significantly.

Data acquired from remote sensing of snow provides us information about its physical and non-physical properties such as grain size, albedo, depth, snow water equivalent (SWE). The estimation of snow grain size is quite essential apart from the regular snow class categorization. The ice absorption feature centered at λ=1.03 μm is typically known to be sensitive to the optically equivalent snow grain size. This work explores the applicability of Airborne Visible Infrared Imaging Spectrometer (AVIRIS-NG) hyperspectral data with electromagnetic spectrum spread from 380 nm - 2510 nm, at 5 nm band interval for estimation of grain size of near surface snow layer at the study site located near Patsio, Himachal Pradesh, India. Continuum absorption area, being inherently insensitive to the topography and known to deliver high signal to noise ratio as compared to absorption depth forms the basis of this study.