**5. Discussion**

This paper was designed to assess user needs (in terms of land administration functions), how the three remote sensing methodologies under development meet these needs, and finally what governance aspects would be critical in widespread update. As a result of the workshops in Kenya, a SWOT analysis was created for each developed application. The results of those SWOT analyses as well as from the fieldwork are summarized and visualized in Table 8: the adherence to 10 aspects of the assessment criteria, derived from the user needs assessment, land management paradigm [5], and FFP requirements [4] is shown.

**Table 8.** Assessment of remote sensing methodologies with regard to fit-for-purpose land rights mapping in Kenya. Green indicates compliance with an aspect, yellow indicates that the application partially complies with an aspect.


With regards to SmartSkeMa, is seems clear that this is not a methodology aiming to replace data collection via aerial images or other surveying techniques, but sketch maps can be used to complement and support collecting data about the relationship of people with respect to land. When SmartSkeMa is considered as methodology for documenting community land tenure in Kenya, its ease of use makes it a cheap option as, once set up, it allows communities to document their land with little additional cost. Its level of accuracy can also be tailored to the task at hand since communities can sketch on top of an aerial image allowing higher precision than is obtained using a plain sketch map. Finally, because a community can use SmartSkeMa with relative independence it may produce data faster than would be possible using traditional land survey methods where the skilled personnel in Kenya are scarce. From the results obtained, SmartSkeMa's functionalities contribute to meeting most of the 10 aspects. The wide range of spatial precision covered by SmartSkeMa presents a great opportunity for incremental and progressive land data acquisition. However, data produced by SmartSkeMa is not very well suited for land valuation in the sense of calculating objective monetary equivalents. The data however may include information about relative values as perceived by land users within a cultural context. More work is needed to determine the extent to which these land values can be captured in the data and how they can be interpreted.

For UAVs, during the workshop most interest was conveyed in the provision of an up-to-date map. Various local government entities such as the department of urban planning and spatial development identified the potential of UAV data to derive information on the current land use and for monitoring urban developments. Furthermore, the immediateness of the data provision was seen to be very beneficial to investigate and solve land disputes within group ranches. However, since the registry index maps are paper-based, the entry barrier to adopt UAV technology is very high. Good visibility of rooftops and information on the height of buildings was found to support land valuation processes. The exploratory case study in Kenya showed that most of the 10 aspects can be met. As an indirect surveying technique, the concept of using UAV technology in cadastral mapping is based on the existence of visible boundaries which can either be extracted by automated image analysis or manual delineation. However, it was also found that a precise and accurate generated orthophoto allows

extraction of boundaries that are not necessarily visible, such as combining features that demarcate the corner point of the parcel even though the line in between is not visible. The ease of use and the flexible setup in terms of the technical standards of the sensor and platform allows covering a large range of different purposes. In terms of scalability, UAV technology only serves a limited range of different scales as costly and lengthy flight authorization procedures hinder an efficient application. Furthermore, in many countries current regulations require to fly missions which are in visual line of sight, allowing only some hundred meters of a possible flight trajectory. According to Kenyan stakeholders, limited battery capacity was found to be the second bottleneck currently impeding large scale implementation.

For ABE, the results from the workshop proved that our proposed automatic boundary extraction approach facilitating the delineation of visible objects and cadastral boundaries can be used to collect information on land tenure, land value, and land use. It further aligns well with the FFP spatial and scalability requirements: it allows a cheap, fast, and accurate delineation of visible boundaries from aerial imagery. However, costs, speed, and accuracy can vary depending on the capture and processing of the aerial imagery and the implementation of the automated boundary extraction: the approach is currently open source, which seems low-cost, but might require more time in acceptance as the SWOT analysis revealed. Given the complexity of cadastral boundaries, automating their delineation remains challenging: the variability of objects and extraction methods reflect the problem's complexity, consisting of extracting different objects with varying characteristics. These circumstances impede the compilation of a generic model for a cadastral boundary and thus the development of a generic method. These remarks come back to the limitations of general boundaries: no standardized specifications exist for boundary features, boundaries are often not marked continuously and maintained poorly [59]. To further develop automated boundary extraction in indirect surveying, we suggest considering the extractable boundary rather than the visible boundary alone (Figure 14): instead of focusing on the visible boundary comprising of outlines of physical objects, automated boundary extraction should focus on the extractable boundary that incorporates local knowledge and context. This information is not inherent in the concept of the visible boundary, but it is extractable from remote sensing imagery.

**Figure 14.** From physical object to cadastral boundary: reformulated boundary concepts for indirect surveying.

Overall, our approach that couples a machine-based automatic feature extraction with a delineator-based interactive delineation can be used to map extractable boundaries. The delineation cannot be fully automated at the current state since the extracted outlines require (legal) adjudication and incorporation of local knowledge from human operators to create final cadastral boundaries. Image-based approaches bear potential to automatically extract use rights, which do not necessarily represent legal rights. These circumstances limit the scope of automated approaches. We observed that automating boundary extraction dealing with sensitive land rights can only be successful, when the interactive part that bridges the gap between automatically generated results and the final cadastral boundary is designed and implemented in correspondence to user needs. Our work revealed limitations of the current approach and ideas for improvements to be addressed in future work, in order to advance the current approach regarding efficiency and acceptance. This would promote the paradigm shift towards cadastral intelligence that integrates human-based expert knowledge with automatically

generated machine-based knowledge. Additionally, future studies should provide approaches to capture requirements from existing technical, legal, financial, and institutional frameworks to be considered when aiming to implement innovative cadastral mapping procedures successfully.

Finally, for governance aspects, further notes on legislative and financial aspects are worth expanding upon. Implementing any of the remote sensing methodologies at any scale, without an appropriate legislative framework, appears fraught. This partly relates to modernizing existing laws and regulations to open up to innovative approaches, and partly to new rules for new challenges. Especially when this new legislation would give clarity on the responsibilities of the different actors, prioritize cheap and open source technologies and stimulate and facilitate partnerships between the governmental and non-governmental actors that would make the uptake and upscaling of the remote sensing methodologies much more likely. Without this an occasional "pilot" might continue and show what can and cannot be achieved within a certain setting, but for true upscaling, a supportive environment will be needed; appropriate laws and regulations and a collaborative attitude among national and local government, as well as with non-government actors. As Kenya has a long land administration history, there is the human capacity in the field, however it seems this country is lacking the political will to introduce that supportive environment to a large extent. Focusing specifically on UAV legislation, getting to balanced legislation that allows a responsible use of UAVs without truly compromising the other issues is not easy. This can be seen worldwide, but even more in countries like Kenya, which struggle to get political will to make clear instructions for UAVs. The implementation of the UAVs could be improved by increasing collaboration between the national and local governments with the non-governmental actors. This collaboration could help to solve the lack of important financial resources. Resources are needed to hire new staff, training, certification, among others. In this sense, the national level can also play an important role as a facilitator to allow private companies to participate. Some non-governmental actors such as private companies could have the resources to use the UAVs; however, they require certain governmental support such as the issuing of the permits or incentives to invest. Meanwhile, on the financial aspect, both proprietary and open source options present challenges: actors payments for software, licenses, and the required updates prohibitive; however, even with open source software, the lack of IT infrastructure and internet access still impacts negatively on scaled uptake. In Kenya, the current resources are not enough to establish a sustainable implementation.
