We're sorry but this page doesn't work properly without JavaScript enabled. Please enable it to continue.
Feedback

Analysis of Free and open Land Cover maps for agricultural land use planning at the local level

00:00

Formal Metadata

Title
Analysis of Free and open Land Cover maps for agricultural land use planning at the local level
Title of Series
Number of Parts
351
Author
License
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Identifiers
Publisher
Release Date
Language
Production Year2022

Content Metadata

Subject Area
Genre
Abstract
Analysis of Free and open Land Cover maps for agricultural land use planning at the local level The area of agricultural land for food production is limited and is constantly decreasing both in the world and in Bosnia and Herzegovina. According to the National Action Plan in Bosnia and Herzegovina (B&H), up to 1,600 ha of land are lost annually (NEAP BiH 2003). The prevention of degradation and sustainably controlled land use should be the most important parts of the land protection policy of every country and the local community. In order for this policy to be implemented properly, relevant indicators of the state of land resources are necessary (Predic et al. 2021). According to the Law on Agricultural Land of the Republic of Srpska, municipalities and cities are obliged to prepare a planning document “Groundwork for Agricultural Land Protection, Use and Restructuring (The groundwork)”. The Groundwork is made according to the FAO (Food and Agriculture Organization) model which consists of an inventory of land and climate resources, agro-ecological zoning, and economic-ecological zoning. With GIS modeling of existing data (pedology, digital elevation model, climate data,...) new relevant data were created (bonity, agro-ecological zoning, suitability of cultivation…). It is intended for municipal authorities in decisions making in the process of land use and protection. GIS layer of the current condition of land cover and land use (hereinafter LC/LU) is one of the most important GIS layers for creating Groundwork. It is necessary to make a precise GIS layer on a large scale in order to obtain relevant data on agricultural land and land use. The most precise method of making LC/LU is manual mapping of LC/LU classes with orthophotos and high-resolution satellite images combined with field verification. The critical point of this method is that it is time consuming. On the other hand, "free" land cover data is available, such as Corine Land Cover (hereinafter CLC), OpenStreetMap,... In this paper, using free open source programs, a comparison of two sets of data representing land cover was performed: manually vectorized data with an orthophoto image of LC/LU and CLC. The aim of this paper is to determine the relevance of CLC data for the needs of land use planning at the level of administrative units in B&H. The study area is the municipality of Laktaši with an area of 38807 ha for which the LC/LU was created in 2018 at the same time as the CLC for B&H. The first phase of the comparison is the synchronization of LC/LU-CLC classifications. LC/LU classification is The Land Cover Classification System, (FAO LCCS, 2000) which is modifiable for the conditions of B&H. Both the LC/LU and the CLC classifications consist of classes divided into three levels. The main difference between LC/LU and CLC is that the LC/LU classification is primarily intended for the detailed identification of agricultural land. The LC/LU nomenclature is dominated by classes that represent agricultural land both in terms of land cover and in terms of use (18 out of a total of 36 classes). The smallest mapped area in LC/LU depends on the significance of a LC/LU class. For example, for the arable land class, it is 0.5 ha, and for the permanent crops class, it is 0.1 ha. The main reason is the fragmentation of properties in B&H (85% is dominated by less than 0.5 ha plots). Unlike the CLC classification, which discusses artificial surfaces in great detail and has 11 classes in the third level (111 Continuous urban fabric,.., 121 Industrial or commercial units,…,142 Sport and leisure facilities), LC/LU classification has only 2 classes for artificial surfaces: Built up and Built up dominates. In this class, the minimum mapped area is 0.025 ha because it is necessary to accurately separate land areas that are temporarily or permanently lost to agriculture. Regardless of the above differences, it is possible to synchronize LC/LU and CLC classifications through third level classes. In the study area (Laktaši municipality) LC/LU GIS layer contains 23 out of 36 LC/LU classes (10707 polygons), and CLC layer 16 out of 44 classes (177 polygons). In the study area, the CLC classification did not recognize 11 classes of LC/LU, of which 8 classes are precisely characterized by agricultural areas (greenhouses, vineyards, nurseries, meadows…). The entire process of comparing and analyzing data was performed using QGIS with the support of the Python programming language. Using QGIS, the union of LC/LU and CLC polygons (14044 polygons) was performed. Using the Python programming language, an error matrix was created and the parameters of the quality of land cover maps were recalculated (Bratic et al., 2020). The obtained results show the accuracy of CLC with respect to LC/LU reference. Although the overall accuracy is 70%, the class-level results are showing that during the creation of CLC layers, a significant part of non-agricultural areas was marked as agricultural classes. For example, 19.4% LC/LU forest class and 42.4% LC/LU class built up, in CLC were mapped as arable dominated class. From the above example, in the studied area, a significalntly larger area of agricultural land was present in relataion to the actual state. After analyzing the results, it was concluded that the CLC in the studied area is not a sufficiently precise GIS basis for agricultural land use planning at the local level. However, it can be a good starting point for making of LC/LU, which would significantly shorten the time of its creating.
Keywords
PlanningLocal ringMathematical analysisOpen setFreewareCivil engineeringFederation of Bosnia and HerzegovinaProcess modelingEuclidean vectorPhysical systemExecution unitAbelian categoryPermanentPolygonAreaObservational studyCategory of beingSocial classTexture mappingClique-widthSatelliteForestGraph (mathematics)Element (mathematics)Projective planeCovering spaceLevel (video gaming)Process (computing)Social classError messageResultantMatrix (mathematics)NeuroinformatikVector spaceSurfaceMereologySynchronizationForestOpen setExecution unitPoint (geometry)Product (business)Connected spaceAreaObservational studyFigurate numberSubject indexingFrequency2 (number)InformationStudent's t-testPolygonSoftwareFocus (optics)Procedural programmingSign (mathematics)MappingComputer animation
ForestOpen setEmail
Transcript: English(auto-generated)
Thank you, Marco, for the introduction. Welcome everyone. My name is Stefan Jovanovic and I will present the work done with the chief of Agroecology department within the Agricultural Institute of Republic of Srpska and with Gorica Bratish PhD student from Politecnico di Milano. This one doesn't work. Okay, just few information about Agricultural Institute. It is founded in
1947 and within 9th department there is a unit that deals with GIS. The story of GIS started 10 years ago with one FAO project supported by Italy. Thank you, Italy, one second. Practically after post-war period
people needed to create land cover and land use map in order to protect agricultural land. So starting with this project we developed projects and documentation for the country level and afterwards for the local level. The goal of these projects and this documentation, as I said before, is
to create policies that will protect agricultural land and one of the crucial elements was land cover, land use map layer. So in this research that I'm presenting today
we tested three and open layers with respect to the one that are proprietary, the one that we created as the Institute and the study area is Lakhdashi municipality which is placed in the northern part of Bosnia. As you can see here some details about the surface and
population in this region. So the land cover map on the left side is the one that we have created according to FAO classification that is modified for Bosnia as a country. Within this classification we have two main categories, non-agriculture and agricultural, which is our focus,
then nine main classes and then individual classes. Also on the other side, you can see corina land cover, which is free and open data that we examined during this research. Practically in this methodology we used free and open JS software, QGIS, where we unified
vectors represented with these two land cover maps. We did synchronization with respect to land cover, land use, the one that we created and in Python we did the reclassification in order to find easier way to create error matrix and after creation of error matrix
we did the computation of land cover indexes. So on the figure you can see the proportions of the classes generated with these two maps. As you can see corina land cover
has mapped significantly bigger area of agricultural land, which is first issue. Regarding procedure, sorry, producer accuracy for certain classes, cultivated dominates show the best results. The issue that we found that some of the classes, agricultural classes, were not recognized within corina, which for us is quite the big deal and
the class with biggest portion of omission error was cultivated dominates. Regarding user accuracy, the best one was forest built up and orchard and
regarding omission error for classes, pastures and built-up converted in the surface is around 350 hectares. In conclusion I left this at the end. The overall accuracy was 71%. The issue as I mentioned are the missing classes within corina land cover and
what we found out for certain classes we can use for this kind of documentation corina land cover as a good starting point. Also in our country a lot of private companies are trying to deal with jazz and somehow with their connections to the politics they are making this corina a proprietary data and they are trying to sell product like this,
which is not correct and I want to stress that so we have to rely on data that are based on science not just to sell something that is for our process fine or not. Thank you for the attention.