Uma Shankar Panday

Crop monitoring is vital for ensuring food security and supporting agricultural programs. In the context of Nepal where agricultural practices often showcase mixed crop systems within small land parcels, a detailed and high-resolution image is necessary. The objective of this study is to classify Paddy and Sugarcane crops in Terai districts of Nepal namely Dang, Banke, Bardiya, Kailali and Kanchanpur. This study explores the efficiency of machine learning algorithms Random Forest (RF), Classification and Regression Trees (CART), and Support Vector Machine (SVM)—for crop classification using Sentinel-2 10m satellite imagery within the Google Earth Engine (GEE) platform. Multi-temporal imagery proved crucial for distinguishing between different crop types, as the spectral properties of crops vary significantly across different growth stages and seasons. Employing a detailed and high resolution approach, median composite images were generated for key periods: October (during paddy ripening), June (prior to paddy plantation), and February (following sugarcane harvest). Post-classification analysis involved assessing accuracy through confusion matrices and calculating kappa coefficients. Findings indicate that utilizing GEE via a cloud platform enables efficient satellite data accessing, filtering, and pre-processing. In terms of overall classification accuracy and kappa coefficient, Random Forest (93.49%, 0.9061) and CART (91.32%, 0.8758) classifiers demonstrated higher classification accuracy compared to SVM (86.04%, 0.7956). Notably, from the results, it is clear that Random Forest is considered a particularly effective classifier.

Archaeological sites hold immense economic value for Nepal, contributing to income and overall economic growth. These sites are susceptible to damage from both human activities and natural factors, needing swift and cost-effective reconstruction efforts. In addressing this challenge, 3D scene reconstruction emerges as a crucial reverse engineering method. Various techniques, such as LiDAR and Photogrammetry, are used for this purpose. While LiDAR offers precise models, its high cost has limited its accessibility. A notable development in this context has been achieved due to the introduction of consumer-grade LiDAR sensors, particularly with the integration of such sensors into iPhones. The primary focus of this research revolves around assessing the efficacy of consumer-grade LiDAR sensors in archaeological scene reconstruction. Utilizing the SiteScape and 3D Scanner apps, several tests with various settings were conducted to generate point clouds of the studied archaeological structures. The obtained point clouds were analyzed to assess point density variations across various application settings. Meshes were generated from these point clouds and the measurements were then taken from digital meshes which are compared with in-situ measurements, allowing for the calculation of relative error. Measurements obtained through SiteScape displayed a maximum difference of 4.17% and a minimum difference of 0.05%. In contrast, the 3D Scanner exhibited a maximum difference of 6.22% and a minimum difference of 0.09%. The findings suggest that the SiteScape app provides a higher point density and a higher accuracy than the 3D Scanner.

After the political change in Nepal of 1951, leapfrog land policy improvements have been recorded, however, the land reform initiatives have been short of full success. Despite a land administration system based on cadaster and land registries in place, 25% of the arable land with an estimated 10 million spatial units on the ground are informally occupied and are off-register. Recently, a strong political will has emerged to ensure land rights for all. Providing tenure security to all these occupants using the conventional surveying and land administration approach demands a large amount of skilled human resources, a long timeframe and a huge budget. To assess the suitability of the fit-for-purpose land administration (FFPLA) approach for nationwide mapping and registration of informality in the Nepalese context, the identification, verification and recordation (IVR) of the people-to-land relationship was conducted through two pilot studies using a participatory approach covering around 1500 and 3400 parcels, respectively, in an urban and a rural setting. The pilot studies were based on the FFPLA National Strategy and utilized satellite imageries and smartphones for the identification and verification of land boundaries. Data collection to verification tasks were completed within seven months in the urban settlements and for an average cost of 7.5 USD per parcel; within the rural setting, the pilot study was also completed within 7 months and for an average cost of just over 3 USD per parcel. The studies also informed the discussions on building the legislative and institutional frameworks, which are now in place. With locally trained ‘grassroots surveyors’, the studies have provided a promising alternative to the conventional surveying technologies by providing a fast, inexpensive and acceptable solution. The tested approach may fulfill the commitment to resolve the countrywide mapping of informality. The use of consistent data models and mapping standards are recommended.

Food security is a longstanding global issue over the last few centuries. Eradicating hunger and all forms of malnutrition by 2030 is still a key challenge. The COVID-19 pandemic has placed additional stress on food production, demand, and supply chain systems; majorly impacting cereal crop producer and importer countries. Short food supply chain based on the production from local farms is less susceptible to travel and export bans and works as a smooth system in the face of these stresses. Local drone-based data solutions can provide an opportunity to address these challenges. This review aims to present a deeper understanding of how the drone-based data solutions can help to combat food insecurity caused due to the pandemic, zoonotic diseases, and other food shocks by enhancing cereal crop productivity of small-scale farming systems in low-income countries. More specifically, the review covers sensing capabilities, promising algorithms, and methods, and added-value of novel machine learning algorithms for local-scale monitoring, biomass and yield estimation, and mapping of them. Finally, we present the opportunities for linking information from citizen science, internet of things (IoT) based on low-cost sensors and drone-based information to satellite data for upscaling crop yield estimation to a larger geographical extent within the Earth Observation umbrella.

Food security is one of the burning issues in the 21st century, as a tremendous population growth over recent decades has increased demand for food production systems. However, agricultural production is constrained by the limited availability of arable land resources, whereas a significant part of these is already degraded due to overexploitation. In order to get optimum output from the available land resources, it is of prime importance that crops are monitored, analyzed, and mapped at various stages of growth so that the areas having underdeveloped/unhealthy plants can be treated appropriately as and when required. This type of monitoring can be performed using ultra-high-resolution earth observation data like the images captured through unmanned aerial vehicles (UAVs)/drones. The objective of this research is to estimate and analyze the above-ground biomass (AGB) of the wheat crop using a consumer-grade red-green-blue (RGB) camera mounted on a drone. AGB and yield of wheat were estimated from linear regression models involving plant height obtained from crop surface models (CSMs) derived from the images captured by the drone-mounted camera. This study estimated plant height in an integrated setting of UAV-derived images with a Mid-Western Terai topographic setting (67 to 300 m amsl) of Nepal. Plant height estimated from the drone images had an error of 5% to 11.9% with respect to direct field measurement. While R² of 0.66 was found for AGB, that of 0.73 and 0.70 were found for spike and grain weights respectively. This statistical quality assurance contributes to crop yield estimation, and hence to develop efficient food security strategies using earth observation and geo-information.

Forest biomass is the sum of above ground living organic material contained in trees which is expressed as dry weight per unit area. Forest biomass acts as substantial terrestrial carbon sinks, they are estimated to absorb 2.7 Petagrams of carbon per year, as such accurate estimation of forest carbon stock is very important. The estimation of biomass is also important because of its application in commercial exploitation as well as in global carbon cycle. Particularly in the latter context, the estimation of the total above-ground biomass (TAGB) with sufficient accuracy is vital in reporting the spatial and temporal state of forest under the United Nations Framework Convention on Climate Change (UNFCCC), Reducing Emissions from Deforestation in Developing Countries (REDD). In this research, tree height, DBH and crown cover were measured using field instruments. Individual ultra-high-resolution UAV images acquired using customized Visible-NIR, were georeferenced and tree crown were extracted using multi-resolution segmentation. A regression equation between field measured biomass and Crown Projection Area (CPA) was developed. The paper presents results from Barandabhar Forest of Chitwan District, Nepal. RMSE of ortho-mosaic was found to be 18 cm. While R² value of 89% was obtained for relationship between DBH and biomass, that of 61% was attained for relationship between CPA and biomass.