EFFICIENT AND FAST PRODUCTION OF CADASTRAL
Transcription
EFFICIENT AND FAST PRODUCTION OF CADASTRAL
EFFICIENT AND FAST PRODUCTION OF CADASTRAL MAPS IN ETHIOPIA Thomas Dubois International Cadastral Mapping Adviser, REILA project NIRAS Finland Oy [email protected] Paper prepared for presentation at the “2016 WORLD BANK CONFERENCE ON LAND AND POVERTY” The World Bank - Washington DC, March 14-18, 2016 Copyright 2016 by author(s). All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies. Abstract Ethiopia is estimated to have 50 million parcels. Of these, only a fraction are registered in any kind of cadastral register. The rest of the parcels can only be distinguished by informal agreements between the land holders in the country. The effects of this are numerous land conflicts and a high tenure insecurity. It has been shown by many international studies that the introduction of a cadastral system generally gives a greatly improved tenure security. In real terms, a cadastral register can therefore mean less conflicts, increased food production and an improved environmental protection. It also empowers women and increases gender equality. The Responsible and Innovative Land Administration (REILA) project is a cooperation between the Finnish and Ethiopian governments. It is implemented by the consultant company NIRAS and has been active since 2011. Its main focus has been to support the development and performance of rural cadastral mapping in Ethiopia. This paper will explain how the REILA project recently has effectivized the mapping process to greatly minimize the time and effort needed to produce the cadastral maps. This is done with a method that is easy to teach and that can be performed by using only open source softwares. Key Words: Land administration, Cadastre, Ethiopia, Parcel maps, Efficient production 2 Table of Contents 1 Introduction ........................................................................................................................................... 4 1.1 The General Cadastral Situation in Ethiopia ................................................................................. 4 1.2 The REILA Project ....................................................................................................................... 5 1.3 Cooperations ................................................................................................................................. 6 1.4 Field Surveying Methods .............................................................................................................. 6 1.5 Imagery-based Field Surveying of Parcels ................................................................................... 7 1.6 Editing and Printing of Imagery-based Field Data ....................................................................... 7 1.7 Information About the Author ...................................................................................................... 8 1.8 Acknowledgements ....................................................................................................................... 8 2 General Description of the Workflow ................................................................................................... 9 3 Production of Field Maps.................................................................................................................... 11 4 Field Work .......................................................................................................................................... 14 5 Office Editing...................................................................................................................................... 16 5.1 Scanning and Georeferencing ..................................................................................................... 16 5.2 Quality Control ........................................................................................................................... 18 6 Public Display and Additional Editing ............................................................................................... 20 7 Printing of Parcel Maps....................................................................................................................... 21 8 Handover and Maintenance ................................................................................................................ 22 9 Possible Future Improvements ............................................................................................................ 23 10 Conclusions ..................................................................................................................................... 26 11 Table of Figures .............................................................................................................................. 27 3 1 Introduction 1.1 The General Cadastral Situation in Ethiopia In Ethiopia, it is estimated that there are approximately 50 million parcels. Of these, only a fraction are registered in any kind of cadastral register. The rest of the parcels can only be distinguished by informal agreements between the land holders in the country. The effects of this are numerous land conflicts and a high tenure insecurity, which have the result that the land holders use the land in a very short term perspective. It has been shown by many international studies that the introduction of a cadastral system gives a greatly improved tenure security. In real terms, a cadastral register can therefore mean less conflicts, increased food production and an improved environmental protection. Since it is made mandatory by law to register also the spouse in joint (family) holderships, it also empowers women and increases gender equality. To complete a modern digital cadastre that covers the whole country has therefore been high on the agenda for the Ethiopian government on both national and regional level. The scale of this task means that any improvement in the production speed is of utmost importance. It should be mentioned that parcels are held (not owned) from the Government of Ethiopia by the land holders through a lease system that normally spans between 25 and 99 years, depending on the type of holdership. In addition, holding of land without time limit exists in rural areas for peasants and pastoralists. In some areas, registration of the parcel attributes (land holder names, land use, parcel ID etc) have been performed. This is titled First level registration, and does not include the measurement of the parcel boundaries. However, it has been found that there is an urgent need of a Second level registration, which includes the registration of the parcel boundaries. The main use of the geographical features of the parcels are general planning, valuation, land disputes and reallocation. Figure 1. The regions of Ethiopia. 4 The country is divided into administrational units at four levels: Region, Zone, Woreda (“district”) and Kebele (“sub-district”). Parcel IDs are created by using codes (numbers or letters separated by slashes) for the four administrational units, followed by either a Holding and Parcel ID, or only a Parcel ID (this differs between the regions). E.g. in the Benishangul-Gumuz region, a full parcel ID could be 06/02/02/14/1502. 1.2 The REILA Project The Responsible and Innovative Land Administration (REILA) project is a cooperation between the Finnish and Ethiopian governments, and has been active since August 2011. Its main focus has been to support the development and performance of rural cadastral mapping in Ethiopia. The project will finish in June 2016, but different extension possibilities are now evaluated. REILA involves the following Ethiopian institutions: The Ministry of Agriculture and Natural Resource, Land Administration and Use Directorate (MoARD-LAUD). The Ethiopian Mapping Agency (EMA). The Amhara Bureau of Rural Land Administration and Use (BoRLAU). The Benishangul-Gumuz Bureau of Environment, Forestry and Land Administration (BoEFLA). In addition, the Information Network Security Agency (INSA) is an indirect beneficiary of the project. One of the main tasks of the REILA project is to establish a cadastral system in its focus areas. Initially this included the Amhara and Benishangul-Gumuz regions. However, REILA was given the responsibility by the Ethiopian Ministry of Agriculture and Natural Resource to evaluate imagery based cadastral mapping in different geographical and cultural environments using different methods countrywide. This expanded the activities to the Oromia, SNNP and Tigray regions. This task resulted in reports, seminars and an Operations Manual, as well as valuable experience for the project staff and its counterparts. Among other project activities, the following can be mentioned. Supporting the acquisition of aerial and satellite imagery that covers parts of Amhara and Benishangul-Gumuz, e.g. by assisting the planning and performance of ground control point surveying. Establishing of a TVET Land Administration education in Assosa, Benishangul-Gumuz. 5 Establishment of a new geodetic national zero-order reference network in cooperation with EMA. Establishment of a new National Rural Land Administration Information System (NRLAIS) in cooperation with mainly EMA, MoARD-LAUD and INSA. Capacity building through cooperation with universities inside (e.g. Bahir Dar University) and outside (e.g. ITC Holland) of Ethiopia. Arrangement of seminars and study visits (e.g. to Holland, Germany, Rwanda and Finland) for persons with key positions in the mentioned Ethiopian counterpart organizations. Presentations and publications of documents at e.g. World Bank- and FIG events. 1.3 Cooperations Except cooperation with the mentioned counterpart institutions, the following programmes/projects can be mentioned: Land Investment for Transformation (LIFT). This six-year DFID-funded project started 2014 and focuses on the issuance of second level certification in Ethiopia. The cooperation between REILA and LIFT is mainly through sharing of experience and planning of related cadastral activities. Sustainable Land Management Programme (SLMP). This programme is mainly carried out by MoARD-LAUD. The cooperation with REILA has included sharing of staff and experiences. 1.4 Field Surveying Methods Since around 10 years, different mapping methods have been tested in Ethiopia to evaluate which method that is most suitable for rural cadastral mapping. The traditional ground surveying techniques (using total station or precision GPS) are in many cases expensive and difficult to teach in a short time. They are very accurate though, so they are preferred in infrastructure and (peri)-urban mapping. The handheld GPS is comparatively inexpensive and easy to use, but has a problem with random inaccuracy when used outside North America, Europe and Eastern Asia. However, for navigation purposes in the field work, it is a very useful tool. Satellite and aerial image techniques are comparatively inexpensive and accurate, easy to teach and have seen a rapid development the recent years. The accuracy and resolution of the imagery have been drastically improved the recent years, and are suitable for rural cadastral mapping nowadays. For these reasons, imagery based rural second level registration was selected as the main cadastral mapping method of the REILA project. 6 1.5 Imagery-based Field Surveying of Parcels The images are printed out and brought to the field for identification of the parcel boundaries. The parcel borders are confirmed at the location of the parcel by the land holder, neighbors and a Land Administration Committee member (the legal representative). The borders are then simply drawn with a pen directly on the image printout. This method is proved to be very successful, since it is fast, easy to teach and has an accuracy (normally 1-2 m) that fits the purpose for rural cadastral mapping. It also has a high acceptance from the land holders, since they can easily verify that the parcel borders are correctly demarcated on the image printout by comparing the drawn border lines with features (e.g. roads, trees) in the image printout. 1.6 Editing and Printing of Imagery-based Field Data The office part of the process involves scanning and referencing of the printed image which includes the drawn border lines, followed by digitization of the borders in a GIS software. Until recently, this process had several bottlenecks. The production of the image printouts took a long time, involving the creation of an atlas of images covering the selected sub-district. The recent development of the methodology now means that the process has an improved quality control and greatly reduces the time needed to produce the map printouts needed for the work. This is all done based on open source software, mainly using the latest version of the Quantum GIS (QGIS) software. By using a combination of built-in functions in the mentioned software, an atlas consisting of hundreds of field maps can now automatically be produced in a few hours, using digital orthophotos as the main resource. The parcel map production is even more improved, with functions that automatically creates a parcel map atlas in a pdf file containing all parcel maps. For each individual parcel map it automatically links in the attribute information about the parcel and the land holder, finds an appropriate zoom level for the parcel size and adds the map to the pdf file. The user can then simply print the whole file to produce the maps in one step. Both the map generation and the printing can be performed in one day for a subdistrict, which typically consists of 5.000-6.000 parcels. With the previous method, this typically took 4-5 weeks of full time work for one person. Other improvements include an effective quality control which can be fully performed within the QGIS software. The quality control includes topology checking, detection of outlying attribute values as well as detection of duplicate parcel IDs. 7 1.7 Information About the Author The author of this document, Thomas Dubois holds an M. Sc. in Surveying (specializing on Geodesy and GIS) from the Royal Inst. of Technology (KTH) in Stockholm, Sweden, and is a permanent employee of the consultant company NIRAS. He is the International Cadastral Mapping Adviser in REILA since the beginning of the project, and is stationed as a component team leader in Bahir Dar, Amhara region. 1.8 Acknowledgements The author wishes to thank all staff working within REILA and NIRAS as well as all staff and managements within the Ethiopian and Finnish counterpart organizations for their invaluable contribution to the development of the methods described in this document. 8 2 General Description of the Workflow This document will focus on the technical parts of the project methodology. However, to give the full picture of the implementation of the second level registration in the project areas, the major steps are listed here. In the preparatory phase activities that form the basis of the certification should be conducted. These include: Selection of a kebele (sub-district) where the second level registration will be performed. Training of trainers and operation managers. Recruitment and training of contractual staff. Strengthen/establish and train Kebele Land Administration Committee members. Update the registry from the 1st level registration (parcel attributes, e.g. land holder names, land use), if it exists. Verify the current extents of the kebele boundary. Settle as many conflicting interests/disputes as possible. Perform procurement and setting up of all the necessary equipment and materials. Public Information and Awareness (PIA) creation for different target groups (e.g. decision makers, land holders). Arrangement of transportation facilities. Preparing and printing of field maps and different field registration forms. The field work can be divided into two separate parts: Surveying and mapping of the land parcel boundaries. Adjudication - identification, assessment, and verification of the legal holders, the legal rights and encumbrances for each parcel. The following are the major activities performed at the office level: Scan and geo-reference the printed out field maps including parcel borders demarcated with a pen. Digitize parcel boundaries. Construct and fill in the parcel attribute table. Scan all existing records and field registration forms. 9 Perform quality control of all the information. Print out the information for Public Display. During this, the land holders verify their registered boundaries and parcel attributes. Make corrections of the information following the Public Display. Produce a map for each parcel. Prepare for the transition to the maintenance phase, so that the cadastre can be continuously updated easily. Figure 2. A farmer is participating in the measurements of his parcel. 10 3 Production of Field Maps The first step is the acquisition of the imagery. One source is Satellite images that are normally ordered from a digital image provider (e.g. Digital Globe). Another option is to use Aerial images. These require thorough preparations (e.g. planning and acquiring airborne photogrammetry equipment), but will give a higher resolution and avoids large displacements in the image due to incorrect compensation of displacements caused by the terrain undulation. The development of enhanced and accurate global terrain models has limited the effects of this error also for the satellite images. When the images are received, an area to be mapped has to be selected. In the REILA case, this is normally a kebele (sub-district). The approximate boundaries of the selected kebele then have to be obtained. These are used to select the parts of the images that will be printed out. By using the approximate boundary of the selected kebele, imagery is made available for the field map production. The selection is normally done by opening the image database and the kebele boundary in the QGIS software. Figure 3. Kebele boundaries are used to select images. The image contrast, brightness and saturation are adjusted individually to enhance the image details. The kebele boundary is then adjusted after discussions with local Land Administration experts. If a detail in the image (e.g. a river) is known to be the exact boundary, the borders are adjusted to fit that feature. 11 By using the QGIS Vector tools, a buffer zone with a distance of e.g. 1500 m is then created around the given boundary. This zone will be used later to create additional field maps on demand if the approximate boundary is not covering the actual kebele borders. In QGIS, a rectangular vector grid is created with a suitable dimension to cover the whole kebele including the buffer zone. Each rectangle gets a dimension to fit the selected map scale (normally 1:1000 or 1:2000) and paper size (normally A3 or A2). The initial point in the grid (upper left corner) gets coordinates rounded off to even 100-values to enable an identical position of the coordinate grid lines on each field map (the grid coordinates will of course be different for different maps). By using Spatial query functions, two layers of rectangles are created. The first contains all rectangles that are on or inside the kebele boundary. The second layer contains all remaining rectangles that are on or inside the buffer zone. Figure 4. Grid rectangles for field map creation in the kebele and the buffer zone. The layer with the grid inside the kebele boundary is then used in the QGIS Print composer as the Atlas Generation Coverage layer. By using a predefined Composer Template, the field maps are harmonized, and it is simple to fill in the needed information (e.g. the kebele name, image acquisition date etc) by using existing and formatted labels in the template. The Atlas Generation function will generate one pdf file that contains all field maps inside the kebele. The maps are automatically numbered in a sequence from 1 and up. The field maps can then conveniently be printed out from e.g. Adobe Reader with the selected paper format. 12 Figure 5. An example of an A3 sized field map. After this, the maps in the buffer zone layer are generated into a pdf file (but not printed). These maps are also numbered from 1 and up but beginning with the letter ‘A’ (for Additional) to distinguish them from the maps inside the kebele. An Overview map is also produced in the QGIS Print Composer (using a separate template), displaying the numbers of each field map from both the kebele- and buffer series. In case the real kebele boundary is found to exceed the area covered by the printed out maps, additional maps can conveniently be printed out simply by finding the numbers of the needed additional maps in the overview. Since they are all prepared in a pdf file, individual maps can then quickly be printed out. This also assures that a minimum of paper and toner is used, since additional maps are only printed on demand. Figure 6. Example of an overview map (left) and an enlargement (right) to display the sheet numbers. 13 4 Field Work Since the technical aspects are the focus of this document, the field work is only briefly described. The prerequisites are the following: A Public Information and Awareness campaign has to be performed in the location of interest. If a first level cadastral register (parcel attributes) exists, it should be updated with current information from recent land administration activities (e.g. from inheritance, subdivision etc). A field team has to be in place, consisting of a surveyor (to draw the boundaries on the field map), a field registrar (to register all parcel attributes (land holder name, parcel ID etc), a team leader (to control the acquired field data and to plan the team activities and write reports), and finally a Land Administration Committee member, who is the legal representative from the kebele. Equipment has to be available, consisting of a selection of field maps (covering the area to be visited the coming days), field registration forms, clipboards, bags, stationery, 50 m measurement tape, ruler and a handheld GPS. A means of transportation (normally a car) also has to be available. The team leader and the committee member plan the daily activities and make sure that the land holders are present when the team moves systematically from parcel to parcel. The land holder points out boundary features (e.g. roads) and the indicated borders are drawn in the image. The parcel ID is then written in the middle of the parcel to enable identification during the digitizing process. After the parcel is demarcated and all data is collected, the field form is signed by the land holder, team leader and committee member. Figure 7. A field team during the demarcation of a parcel (left), and an example of a parcel demarcated in the field map. 14 The measurement tape is used if a border corner cannot be seen in the field map image. In these cases, the distance is measured to an object that can be identified both in the image and on the ground. The distance is converted to the map scale, and by using a ruler, the correct place can be located in the image. In case the direction is uncertain, two or more additional measurements are performed in the same way to obtain the location through intersection. The handheld GPS is only used for navigation purposes and for tasks with a lower accuracy demand (like the demarcation of the kebele boundary). Figure 8. Example of the use of a 50 m tape and a ruler to demarcate hidden parcel corners. The land holder is asked to present any documentation that shows that he/she is the legal holder. If a first level registration exists in the area, it is normally a holding- or parcel registry book that has been issued to the farmer previously by the Woreda (district) Land Administration office. Regarding field forms, two main possibilities exist. If there is an existing database with updated parcel attributes, the data can be printed out and brought to the field. The field registrar will then correct any erroneous information and simply confirm correct information. This will normally increase the speed since less writing is needed, but can be limited by uncertainties and errors in the printed out database information. The second option is to use blank field registration forms and create a new database. This is preferable if any exiting cadastral information is limited or untrustable. The advantage with the second option is also that it is possible to create a ‘tailor-made’ structure with desired contents. This is of special importance when the new national rural cadastral methodology (NRLAIS) is developed. When a field map is completed, it should be delivered to the office without delay. The field crew then continues until the kebele is completed. They can then move to the next kebele and start there if all preparations are done. 15 5 Office Editing 5.1 Scanning and Georeferencing When the field maps are delivered to the office, they are first scanned with an appropriate scanner. For A3 size maps, a simple A3 scanner will perform the task. To reference the scanned image, the Georeferencing function in QGIS is used. To perform this, nine evenly distributed grid crosses from the printed coordinate grid are selected. The comparatively high number of points is chosen to ensure that any grosserror is easily detected, and to be able to find local deformations in the map (e.g. due to folding damages). Each grid cross is zoomed in and digitized, followed by the entering of the grid cross coordinates, which are printed in the margins of the field maps. If any large residuals appear (larger than 3 pixels), all grid gross demarcations and entered coordinates are checked. If physical deformations of the map are found to be the reason, the map is re-scanned with additional pressure put on it to flatten it out. Minor errors have been found to be the consequence of folding the map, so this should be avoided. Figure 9. Selected (circled) grid crosses for georeferencing (left) and a digitized grid cross center (right). When the georeferencing is satisfactory, the new referenced image is added to the main view in QGIS. A simple way to check the quality of the georeferenced image is to compare it to a reference grid (with the same grid interval as the field maps) that can be generated in QGIS. Any deviations are then clearly visible if the scanned map- and reference grid are deviating from each other. 16 Figure 10. Acceptable accuracy (left) and not acceptable (right), when the reference grid line is completely outside the image grid. If the referencing quality is acceptable, the digitizing of the parcels can start. A polygon layer with a predefined attribute table is created for the purpose of gathering the parcel information. The digitizing is done by choosing the polygon tool and then carefully clicking on each border corner that is demarcated with a pen in the field. For curved borders a sufficient number of points are digitized to represent the boundary well. A new function that speeds up the parcel digitizing is the Autotrace function. When demarcating the borders of the neighbor to an existing parcel, it is enough to click at both common corners of the existing parcel. The function will then trace and snap on all points between the selected corners of the existing parcel. In the figure below, a new parcel (shown in red) is attached to point A and B. The existing points in between are snapped to automatically. A B Figure 11. The Autotrace function. Automatic point snapping is made from A to B for the new parcel. 17 A very convenient feature in QGIS is the style files. They can be created as templates, and by activating a style file for a certain layer, all colors, line widths etc are set to the predefined style. The advantage of this is that the graphic display of different layer categories (grids, parcels, boundaries etc) can be harmonized and is also very quick to implement. 5.2 Quality Control To avoid gaps, overlaps and invalid parcel geometries, the QGIS Topology checker is used. This function should be regularly used during the parcel digitizing to avoid topology errors. A B C Figure 12. An example of overlaps (A), gaps (B) and invalid geometries (C, the parcel border is crossing itself). The attributes are first checked by sorting each column in the table. For values, errors are then found by looking at the top and bottom of each column if the erroneous input has an extreme value (e.g. a team number that is higher than the number of teams). Also text columns can be checked by sorting them and then scrolling through them. Since they are sorted, misspellings are easily found. By using the function Select features by using an expression, it is also possible to detect duplicate parcel IDs, since they should have unique values. Figure 13. Detection of duplicate parcel IDs. 18 If several data encoders are digitizing the same kebele, it is important to coordinate and supervise the digitizing, since topology errors will occur when files from different encoders are joined together. One person will then be responsible to create on é final file containing the whole kebele (sub-district). The quality control is also performed by re-checking the information in selected parcels. When the digitizing is complete and the quality control is finished, the Final kebele boundary is demarcated. It is of utmost importance to take regular backups throughout the process. 19 6 Public Display and Additional Editing Even if the quality control in the field and in the office is rigorous, misunderstandings in the field and undetected errors will remain. To detect these errors and to enable transparency in the registration, a Public Display map is printed out and displayed in the kebele for a certain amount of time (e.g. one month). It is then possible for each land holder to verify that the boundaries of his/her own parcel(s), as well as others in the kebele (sub-district), are correct. In addition, the collected attributes of each parcel are printed out and displayed, and a field team is stationed at the selected site in the kebele throughout the display period. Each land holder must now verify that the border- and attribute information are correct. The task of the field team is to collect the signatures and note any incorrect data that is found. They also guide the land holders so that they find their parcels in the kebele map and the attribute tables. Figure 14. A land holder is shown the location of her parcel on the public display map. When the display period is over, the errors that were found are corrected in the office. If necessary, field visits have to be performed to sort out any problems that cannot be solved in the office. It should be noted however, that it is not possible to achieve a successful and correct registration of 100% of the parcels in a kebele. Land disputes and complicated editing errors has to be put in the maintenance list so that the second level certification can take place for the completed majority of the parcels in the kebele. 20 7 Printing of Parcel Maps One major bottleneck in the second level certification process was the creation of the individual parcel maps. The procedure was to manually panorate the Print Composer view in QGIS and adjust the zoom level to find a suitable scale for the parcel. The attribute data (e.g. parcel ID, land holder name) then had to be manually copied from the attribute table into the parcel map, after which it could be printed. This was a tedious process that took 5-10 minutes per map. With a normal volume of 5.000 parcels for a kebele (sub-district), this took several weeks to perform. The use of map templates and the atlas function dramatically reduced the time and effort to generate the parcel maps, by using the following steps: The shapefile containing the parcels is selected for the atlas grid generation. The map is set to be Controlled by atlas, and to use a range of Predefined scales. This will enable the software to automatically find an appropriate scale and position for each map depending on the parcel size and location. The attribute information for the current parcel is obtained by adding a table with a link to the desired column in the attribute table, and selecting Current atlas feature as the source. After entering the common information for all parcels (e.g. kebele name, date of printing) in the predefined labels in the template, a pdf file can be generated. If Single file export is selected, the result will be one pdf file containing all parcels in ID or land holder name order. The parcel maps in the pdf file can then be printed out in sequence or by selecting individual maps in e.g. Acrobat Reader. The main remaining bottleneck is the needed signatures and stamps on each map. It is under investigation if this could be simplified (e.g. by including scanned signatures) in a way that is supported by the legislation and trusted by the land holders. Figure 15. An example of a parcel map. 21 8 Handover and Maintenance When all parcel maps are finished, the whole work including the digital cadastre files, maps, field forms and backups should be taken care of by the implementing organization. In the REILA case, this is normally the Woreda, which is an administrational unit on district level, normally containing 10-50 kebeles (sub-districts). Among the most important parts and the greatest risks of the whole project is the maintenance. If proper routines are not followed and the cadastre is not properly and regularly updated, the whole work may have to be repeated after some years. Many changes have then occurred due to e.g. inheritance and infrastructural development. Ethiopia has started an enormous task which is of utmost importance to finish as soon as possible. Therefore, huge resources in manpower and equipment are needed for the continuous mapping of new areas. It is then a risk that the maintenance is partly neglected due to lack of resources. To distribute them so that the maintenance can still be performed is therefore of highest importance for the future use and reliability of the cadastre. To support the maintenance, it is important to ensure that the updating routines are initiated immediately after delivery and included as a part of the daily work for the land administration staff in the woredas and kebeles. Capacity building is also very important, since staff turnover might mean that a knowledge gap suddenly appears when no one at the land administration office can handle a certain part of the production chain. 22 9 Possible Future Improvements When the production chain is created, it is important not to be locked to a certain software or data format, since this might mean that new methods or softwares cannot be implemented. It is also important to be attentive to any technical developments within land administration, since technical innovations can further effectivize the process. In the methodology presented in this document, some possible future improvements can be mentioned: The georeferencing module that is used now in QGIS is generalized to work for all normal types of georeferencing, with different number of points and point distributions. The module is therefore not ideal for the specific use of referencing the field maps with nine grid crosses. In the present module, it is necessary to input the coordinates of each of the nine grid crosses, and also to zoom in and out to find them. Since the printed coordinate grid on the scanned field maps have a regular grid cross distribution and scale, it would be enough to indicate the position in the image and enter the coordinates for the upper left and lower right grid crosses and finally give the grid interval. The software could calculate the location of the remaining seven points in the ground coordinate system. It could then make an approximate zoomed in display on each point (since their approximate locations are known) to simplify the digitizing of the exact position of each grid cross. This would make the georeferencing remarkably quicker and more fail-safe (since fewer coordinates are entered) for the specific purpose of referencing field maps. Figure 16. An illustration of zoomed in grid crosses in a possible future georeferencing module. 23 One of the most time consuming parts of the office process is the digitizing of parcels. It is somewhat simplified by the previously mentioned Autotrace function, but in most cases it still needs the manual digitizing of each parcel corner. The regular checking of topology errors is also time consuming. To use some kind of automatic line extraction in images have been investigated on several occasions, but there are a limited number of free softwares that can perform the process (Incscape and WinTopo are among the few). If the landscape is free of trees, flat and dominated by a checkerboard of parcels used for agriculture, most borders that are visible in the image are also the real boundaries, and the extracted lines cold then simplify the work. However, in many areas trees are used for border demarcation, and together with other natural features like rivers and gullies, they will generate a large number of erroneous borders if an automatic line tracing function is used. This demands extensive and time consuming editing, and also demands experienced users. An approach that might be promising in the future is to use a type of semi-automatic line tracing. This method begins with the assumption that the field maps are scanned, and that all parcel borders are demarcated with a pen. Every parcel corner that is shared with another parcel is called a node. In the example in the following figure, they are labeled A to E for the parcel 1101/04. Between the nodes, there is normally a need to digitize a number of points to follow the border shape. With semi-automatic tracing, the function would start with digitizing the nodes A and B only. The function first retrieves information about the pen color (with a certain tolerance) where the nodes are digitized. The function can then try to create a straight line from A in the attempt to connect to B, following the color of the pen line. When the line turns (no more pen color in that direction), a border point is automatically created, and a new straight line is started from there. Finally, after creating more border points, the function will in most cases find B and the border section between A and B is automatically completed. 24 A E D A C B B Figure 17. Nodes labeled A-E in a parcel (left), semi-automatic line tracing between nodes (right). After the A-B section is completed, the node C is digitized and automatically connected to B through additional border points. The process is repeated until the whole parcel is demarcated. In case the line passes through e.g. a tree shadow with the same color as the pen, some manual digitizing is needed. However, for most parcels only the border nodes need to be digitized, which significantly speeds up the process. If a neighboring parcel is digitized using the same nodes as a finished parcel, the function can reuse the calculated border points between the nodes so that gaps and overlaps are avoided. Handheld GPS equipment’s suffer at present from random inaccuracy that can reach up to 10 meters in some cases when used in Ethiopia. The main reason for this is that it is not possible to receive any correction signals in this country that are otherwise found in Europe (EGNOS) or North America (WAAS). If a similar signal world start to be transmitted with a range including Ethiopia, the errors would be greatly reduced and the handheld GPS could then be much more suitable for some tasks within cadastral mapping. Satellite images might have a problem with displacements of details in areas with steep terrain with many natural breaklines. The main reasons for this is that a terrain model is used to compensate for these displacements caused by the perspective from the satellite. If the terrain model is not accurate enough, displacement errors in the final picture will occur. However, the international terrain models are continuously improved, and if future terrain models go down to around meter accuracy, satellite images will compete on almost equal terms with aerial images, since the image resolution is already at around 30 cm for e.g. the new satellite WorldView 3. 25 10 Conclusions This document describes how the REILA project has developed simple, fast and efficient methods to perform cadastral mapping in Ethiopia. In general, the recent developments of the cadastral mapping process have greatly reduced time and effort needed to produce the cadastral register in an Ethiopian kebele (sub-district). Previously, it was a tedious operation to print out field- and parcel maps, but due to developments of the methodology it is now a very fast and automated process, which is performed with open-source software (Quantum GIS). The quality control of geographical and attribute data for the parcels has been simplified by the use of topology checking functions for the parcel boundaries and sorting of individual columns in the attribute table. One of the major threats to a developed cadastral database is when the maintenance of the register is not handled properly. It is therefore of great importance to support that resources are set aside for this. Capacity building to avoid problems during staff turnover is also a major issue. Future technical developments will certainly allow an even faster, flexible and reliable methodology. It is therefore important to monitor the development of software and equipment and update the procedures regularly. 26 11 Table of Figures Figure 1. The regions of Ethiopia. ................................................................................................................ 4 Figure 2. A farmer is participating in the measurements of his parcel. ...................................................... 10 Figure 3. Kebele boundaries are used to select images. .............................................................................. 11 Figure 4. Grid rectangles for field map creation in the kebele and the buffer zone. ................................... 12 Figure 5. An example of an A3 sized field map. ........................................................................................ 13 Figure 6. Example of an overview map (left) and an enlargement (right) to display the sheet numbers. .. 13 Figure 7. A field team during the demarcation of a parcel (left), and an example of a parcel demarcated in the field map. .............................................................................................................................................. 14 Figure 8. Example of the use of a 50 m tape and a ruler to demarcate hidden parcel corners. ................... 15 Figure 9. Selected (circled) grid crosses for georeferencing (left) and a digitized grid cross center (right). .................................................................................................................................................................... 16 Figure 10. Acceptable accuracy (left) and not acceptable (right), when the reference grid line is completely outside the image grid. ............................................................................................................. 17 Figure 11. The Autotrace function. Automatic point snapping is made from A to B for the new parcel. .. 17 Figure 12. An example of overlaps (A), gaps (B) and invalid geometries (C, the parcel border is crossing itself). .......................................................................................................................................................... 18 Figure 13. Detection of duplicate parcel IDs. ............................................................................................. 18 Figure 14. A land holder is shown the location of her parcel on the public display map. .......................... 20 Figure 15. An example of a parcel map. ..................................................................................................... 21 Figure 16. An illustration of zoomed in grid crosses in a possible future georeferencing module. ............ 23 Figure 17. Nodes labeled A-E in a parcel (left), semi-automatic line tracing between nodes (right). ........ 25 27