The most important part in forest inventory based on remote sensing data is individual tree identification, because only when the tree is identified, we can try to determine its characteristic features. The objective of research was to explore remote sensing methods to determine individual tree position using LiDAR and digital aerial photography in Latvian forest conditions. The study site was a forest in the middle of Latvia – in Jelgava district (56º39’ N, 23º47’ E). Aerial photography camera (ADS 40) and laser scanner (ALS 50 II) were used to capture the data. LiDAR resolution was 9p m2 (500 m altitude). The image data is RGB, NIR and PAN spectrum with 20 cm pixel resolution. Image processing was made using Fourier transform, frequency filtering, and reverse Fourier transform. LiDAR data processing methods was based on canopy height model, Gaussian mask, and local maxima. Field measurements were tree coordinates, species, height, diameter at breast height, crown width and length. Using combined LiDAR and optical imagery data allows detecting at least 63% of all trees and about 85% of the dominant trees.

The study represents results of theoretical evaluation of forest biomass available for solid biofuel production in pre-commercial thinning in Latvia. The study is based on the National forest inventory (NFI) data; calculations are done for each NFI plot separately. The calculation is done in three steps – selection of the NFI sample plots, which fulfils criteria for the pre-commercial thinning, development of the diameter distribution table, setting the criteria of the thinning intensity, calculation of extractable biomass. Thinning from below (removal of the smallest trees) is considered in calculation. Two types of biomass are accounted – full tree (aboveground biomass) and stem-wood (stem biomass). The study demonstrates that pre-commercial thinning could become an important source of forest biomass in Latvia (15400 GWh of primary energy according to current situation in forests); however, dimensions of trees and harvesting conditions might be challenging for production. The most of the potential biofuel resources are located in stands with average tree higher than 8 m; therefore, it is reasonably to develop and introduce technologies applicable for production of partially delimbed trees.

The aim of paper is a comparative analysis of on-farm greenhouse gas emissions across family farm types and farm size classes using FADN data in Lithuania. To achieve this, Lithuanian FADN data of 2014 were used for the analysis. The research draws on a sample of 1304 family farms. The methodology is based on an adaptation of the IPCC guidelines using Lithuanian emission factors from Lithuania’s National Inventory Report and the activity data of family farms derived from Lithuanian FADN. The GHG emissions were analysed per farm (t CO2eq farmE-1) and per hectare (CO2eq haE-1 of UAA). The research found out that the major sources of GHG emissions are related to the use of chemical fertilizers on farms comprising 52.6% of the total emissions from family farms. The performed analysis shows that GHG emissions per farm depended on the farm size and ranged from 63.3 t CO2eq farmE-1 to 479.6 t CO2eq farmE-1, on farm size class less than 30 ha UAA and from 500 ha UAA or over, respectively. The GHG emissions on family farms totalled 184.2 t CO2eq farmE-1 and ranged from 5.8 t CO2eq farm E-1 to 234.6 t CO2eq farmE-1, in the permanent crops farms and in the specialist dairying farms, respectively.

Every day millions of new data records with spatial component are produced in the world, which provide valuable information to make decisions and solve business-related issues. However, a large part of this data is hardly analysed because of their different structures and schemas. The aim of the paper is to improve the integration, processing and analysis of unstructured spatial data. During the research, the author analysed geospatial data types and sources, explored NoSQL solutions for geospatial data processing and chose the open-source tools which are the most appropriate for the stated goals, as well as analysed the coverage of forest areas with protected zones using MongoDB database capabilities and visualized results in a map, using QGIS software. MongoDB is a useful tool for geospatial data analysis and has a large number of embedded topology analysis functions and has drivers for widespread programming languages like JavaScript, Python, PHP, Java, Scala, CNo., C, C + +, etc. QGIS has extensions that allow to make connections to databases, including a connection with MongoDB. Using these features, the developers can develop geographic information systems to analyse geospatial data – structured, semi-structured and unstructured. Generally MongoDB is used for real-time data analysis; however, complicated analysis of large data sets can take up to hours and even days, so it is still necessary to find the best solution to get results in an acceptable time for users. Using MongoDB together with Apache Hadoop – the framework to support big data applications – could be a possible solution for this problem.

The objective of precision beekeeping is to minimize resource consumption and maximize productivity of bees. This is achieved by detecting and predicting beehive states by monitoring apiary and beehive related parameters like temperature, weight, humidity, noise, vibrations, air pollution, wind, precipitation, etc. These parameters are collected as a raw input data by use of multiple different sensory devices, and are often imperfect and require creation of correlation between time data series. Currently, most researches focus on monitoring and processing each parameter separately, whereas combination of multiple parameters produces information that is more sophisticated. Raw input data sets that complement one another could be pre-processed by applying data fusion methods to achieve understanding about global research subject. There are multiple data fusion methods and classification models, distinguished by raw input data type or device usage, whereas sensor related data fusion is called sensor fusion. This paper analyses existing data fusion methods and process in order to identify data fusion challenges and correlate them with precision beekeeping objectives. The research was conducted over a period of 5 months, starting from October, 2019 and was based on analysis and synthesis of scientific literature. The conclusion was made that requirement of data fusion appliance in precision beekeeping is determined by a global research objective, whereas input data introduces main challenges of data and sensor fusion, as its attributes correlate with potential result.

Increase in use of biomass as renewable source of energy in Europe is tightly linked to the policies aimed at mitigation of climate changes i.e. reductions of greenhouse gas emissions. Both for assessment of the carbon sequestration and emissions as well as for assessment of potential amounts of biomass for renewable energy, information of land cover dynamics are essential. Therefore, the aim of our study was to improve accuracy of estimates of the land use changes in the time period between 1990 and 2014. Land use categories were determined in accordance to UNFCCC: wetland, cropland (arable land, bare field), forest, grassland, settlements (urban/suburban area), and other land. Combination of data from National forest inventory (NFI) sample plots and analysis of Landsat images were used. For the classification based on Landsat images vegetation index (NDVI) was estimated and linked to known information on the land use type from NFI sample plot data. In the analysed period, the most significant changes were found for forest lands – the total area of forest land during the last two decades had increased by 1% (64.5 thousand ha). Similar increase (1.2%) was observed also in the area of cropland. Both of these tendencies were primarily the result of marginal field area reduction (by 2.6%). Increase in forest area and thus annual increment has led to an increase in above-ground biomass by 10.2 m**3 haE-1.

The distribution and abundance of Barbarea arcuata (Opiz ex J. et C. Presl) Rchb. were investigated throughout the territory of Latvia. The field survey was carried out to estimate the abundance patterns, and the herbarium materials were used to compile a distribution map. In total 411 localities were recorded in the period from 2015 to 2017. The species has been commonly found on roadsides, which accounts for 66% of the localities. Seventeen percent of the localities occurred in grasslands, 10% – in croplands, 4% – in fallows, 2% – on road embankment slopes, and 1% – on railway embankments. The highest density of B. arcuata were found in new fallows where it forms large populations. Whole field localities account for 5% of the total localities. Medium-sized stands are found in about 20% of localities and are mostly found in grasslands, roadsides, as well as croplands which include cereal fields and oilseed rape fields. Individual specimens are mostly found on roadside habitats and grasslands and account for 75% of the total number of localities. As dominant weed species it is found on fields of oilseed rape, cereal fields and fallows. Herbarium data and the Institute of Biology, University of Latvia lists of species show that B. arcuata distribution was frequent during the period from 1970 to 2014.

The development of rural territories in the European Union (EU) plays an essential role, and agricultural development can largely contribute to this process. To project agricultural trends, a number of models have been developed in the EU, while in Latvia the LASAM model was developed in 2016 to generate projections for agricultural sector development in Latvia until 2050. In 2017, LASAM was extended by a module for socio-economic assessment that allows projecting productivity for various types of farming. The research aim is to develop a model for productivity simulation for various specialisation types of farms in order to project their development in Latvia. To achieve the aim, two specific research tasks were set: 1) to develop a model for productivity simulation for various specialisation types of farms in Latvia; 2) to identify the key results of the simulation of productivity for various specialisation types of farms in Latvia. The research found that in the period 2005 – 2016 the value added of agriculture tended to slightly increase in Latvia, whereas an opposite trend was observed for the number of persons employed in agriculture, which tended to decrease in the period of analysis. Both trends determine the agricultural productivity trend as well. A projection of productivity measured as value added per AWU for various farming types in Latvia by means of the LASAM model has revealed that it is different, and the highest level of productivity in 2030 and 2050 is projected for granivores as well as field crop farms.

Forestry in Latvia in the 20th century was strongly focused on the establishment and management of pure Scots pine and Norway spruce stands trying to avoid any admixture of other tree species. Knowledge on the economic feasibility of the mixed stands’ management is still rather poor in Latvia, while at the same time the establishment of mixed stands of Norway spruce and birch species has become an attractive management objective in Finland and Sweden. This paper used the data from the Latvian National Forest inventory to quantify the amount of birch stands with the second layer of spruce, as the first step to justify the development of recommendations for alternative management options in this type of stands. According to the results, there are 121 752 ha of birch stands with the second layer of Norway spruce, and most of those are located in Hylocomiosa, Oxalidosa, Myrtillosa mel. and Myrtillosa turf.mel. site types. The mean standing volume of birch stands with Norway spruce understorey was higher than in birch stands with no spruce understorey, and Hylocomiosa, Oxalidosa, Myrtillosa mel. were the most productive site types both in terms of total standing volume and that of the Norway spruce growing in the second layer. Analysed data also revealed that the management of birch stands already now differs strongly in state and private forests, in the latter being more focused on selective fellings. It is possible to develop and test alternative management methods of birch stands with the second layer of Norway spruce to maximise yield and reduce expenses of forest regeneration.

In this study, forest type mapping data set taken from UCI (University of California, Irvine) machine learning repository database has been classified using different machine learning algorithms including Multilayer Perceptron, k-NN, J48, Naïve Bayes, Bayes Net and KStar. In this dataset, there are 27 spectral values showing the type of three different forests (Sugi, Hinoki, mixed broadleaf). As the performance measure criteria, the classification accuracy has been used to evaluate the classifier algorithms and then to select the best method. The best classification rates have been obtained 90.43% with MLP, and 89.1013% with k-NN classifier (for k=5). As can be seen from the obtained results, the machine learning algorithms including MLP and k-NN classifier have obtained very promising results in the classification of forest type with 27 spectral features.