Can citizen scientists provide a reliable geo-hydrological hazard inventory? An analysis of biases, sensitivity and precision for the Rwenzori Mountains, Uganda

Spatio-temporal inventory of natural hazards is a challenging task especially in rural or remote areas in the Global South where data collection at regional scale is difficult. Citizen science, i.e. involvement of no-experts in collecting information and co-creation of knowledge with experts to solve societal and environmental problems, has been suggested as a viable approach to tackle this bottleneck, although the reliability of the resulting data is often questioned. Here we analyse an inventory of geo-hydrological hazards (landslides and floods) reported by a network of citizen scientists in the Rwenzori Mountains, Uganda, established since 2017. We assess the precision, sensitivity and potential biases affecting this citizen science-based hazard inventory. We compare the citizen science-based records with two independent inventories, one collected through systematic fieldwork and another by PlanetScope satellite imagery mapping for the period between May 2019 and May 2020. The precision of the geo-observer data is higher (99% and 100% for landslides and floods, respectively) than that of satellite-based data (44% and 84%, respectively) indicative of fewer false positives in the former inventory. Also, citizen scientists have a higher sensitivity in reporting landslides (51%) compared to satellite imagery (39%) in addition to being able to report the events a few days after the occurrence. In contrast, the sensitivity of satellite-based flood detection is higher than that of citizen scientists. The probability of landslide events being reported by citizen scientists depends both on citizen scientists and hazard specific features (impact, landslide-citizen scientist home distance, landslide-road access distance and altitude). Although satellite imagery mapping could result in a spatially less biased inventory, small landslides are often missed while shallow ones can easily be confused with freshly cleared vegetation. Also, in a dominantly cloudy environment, it can take several days to weeks before a cloud-free satellite image can be obtained. In summary, the typically rapid response time of citizen scientists can result in faster information with high reliability at the risk of missing out almost half of the occurrences. Citizen scientists also provide more data on impact and type of land use, something difficult to achieve using satellite imagery. Working with farmers at village level as citizen scientists can facilitate covering a wider geographical area while reducing the area monitored by each citizen scientist at the same time.


Introduction
Climate-related disasters are on the rise (IFRC 2016, UNISDR 2015a, Papathoma-Köhle et al 2016. They continue to claim lives of many, destroy infrastructure, damage property, and disrupt livelihood activities that eventually undermine the welfare of affected societies. According to the Sendai framework for disaster risk reduction (2015)(2016)(2017)(2018)(2019)(2020)(2021)(2022)(2023)(2024)(2025)(2026)(2027)(2028)(2029)(2030), increasing exposure and vulnerabilities have steadily increased the risk to natural hazards and the associated disasters (UNISDR 2015b). To be able to foster national disaster risk reduction strategies, it is necessary to promote local risk assessment studies based on continuous data collection on disaster occurrences and the impacts they cause. Disaster risk is categorized into intensive and extensive risk. Intensive risk is associated with large impact events of low frequency (Hamdan 2015). Extensive risk on the other hand involves small, frequent and widely spread hazard events such as landslides and floods (UNDRR 2015). Such geo-hydrological hazard events rarely make it to national media, making their assessment challenging despite their cumulative impact being large (Hamdan 2015). This is particularly the case in developing nations especially the tropical mountainous regions of Africa where spatially dispersed small hazards are common but their impact remain under-documented (UNISDR 2015a, UNDRR 2019a. In these mountainous regions, high population densities are often found, frequently on the rise and combined with high societal vulnerabilities (UNISDR 2015a, CRED 2017, Zhou et al 2019. Bottlenecks in regional and subnational Disaster Risk Reduction (DRR) include insufficient institutional capacities, lack of data and inadequate budget allocations (UNISDR 2015b, AUC 2018, UNDRR 2019a. Comprehensive multi-hazard analysis can contribute to better risk management through: (a) a clear understanding of geographical concentrations of the disaster risk, (b) quantification of potential impacts, and (c) identification of key risk drivers (Pondard and Daly 2011). This requires spatially and temporally explicit data records including impact caused. The increasing availability of very high spatial resolution satellite imagery (such as IKONOS, QuickBird, PlanetScope, and Pléiades) presents an opportunity for rapid and accurate visual geo-hydrological hazard mapping at different spatial and temporal scales (Rabonza et al 2016, Monsieurs et al 2018, Talisay et al 2019, Thirumurugan and Krishnaveni 2019, Jain et al 2021. However, in tropical environments, persistent cloud cover, often limits use of optical satellite remote sensing for hazard detection (Robinson et al 2019). Visual satellite image interpretation can also be affected by mis-classifications as human-induced seasonal land use changes can be perceived as potential geohydrological hazards (Dewitte et al 2022, Jain et al 2021. In the recent past, citizen science (CS) has gained tremendous recognition (Danielsen et al 2014, Aceves-Bueno et al 2017, McKinley et al 2017, Hicks et al 2019 although it has existed for long time (Silvertown 2009, Eitzel et al 2017. This is partly due to the increasing realization among scientists of the benefits of this practice wherein non-experts and experts engage and collaborate to solve real-world problems (Cohn 2008). Citizen scientists (CSs) can engage in various stages of the research cycle ranging from collecting data to disseminating study findings (Dickinson et al 2012, Aceves-Bueno et al 2017, Cunha et al 2017, Phillips et al 2018. CS therefore has been interpreted and defined differently depending on context and field in which it is being applied. According to Kullenberg and Kasperowski (2016), CS is understood to involve participation of citizens in the collection and analysis of data for research in different fields including ecology, biology and conservation (Kullenberg and Kasperowski 2016). It can also involve environmental monitoring and inventory projects (Haklay et al 2021). In the context of DRR, the involvement of citizens in recording or mapping hazards can be an example of CS projects (Hicks et al 2019). In this research, the project involves the participation of non-expert citizens in mapping geohydrological hazards and their impacts with goal of scientifically understanding the spatio-temporal distributions, vulnerabilities and risks imposed by the disasters. Participatory approaches enable data collection on a larger geographic scale and over a longer time period than is possible in more conventional scientific research (Cohn 2008). Furthermore, collaborating with community members in disaster assessment facilitates complete understanding of disaster risk in a local context (Bwambale et al 2022).
In 2017, a CS network, coined the geo-observer network, was established in the Rwenzori Mountains, Uganda (Jacobs et al 2019) mainly to (a) record and report environmental hazards (including dates of occurrence and impact caused), (b) disseminate scientific research findings and participate in developing and implementing DRR strategies. This was based on an understanding that people in the affected communities are better placed to monitor their own places and participate in developing DRR strategies. The network currently consists of 30 CSs, trained to detect, and report information on eight different hazards (landslides, floods (here referring to overflow of large volumes of water beyond the usual normal levels mainly during or after rainfall), droughts, crop and livestock pests and diseases, earthquakes, lightnings, windstorms, and hailstorms) by filling a simple structured questionnaire programmed in a KoboCollect application using smartphone technology . With a mobile internet connection, the information recorded in a standard form (including location, type of hazard, size, picture of the affected elements and impact on housing, population, crops and physical infrastructures) is uploaded to a central KoboCollect account managed by a research assistant based at Mountains of the Moon University. This online database contains all information recorded for each event, including GPS coordinate, pictures, and the ID of the reporting citizen scientist. Each geo-observer of the network is responsible to report all events for a specific territory, typically one or two parish (es). A parish is a second smallest administrative unit in Uganda next to a village. The parish sizes are not homogeneous as the boundaries are mostly defined by physical features such as rivers, streams, wetlands, roads and gorges. Although the network has demonstrated potential for CS-based hazard mapping Jacobs et al (2019), the growing inventory has to date not been used for hazard or risk assessment. Prior to valorisation of the data, i.e. using the CS-based hazard location and impact on society to constrain risk assessment and inform risk reduction strategies, in hazard and risk assessments, the reliability of the citizen-based dataset needs to be assessed  as citizen science based data might be affected by biases or issues of reliability or completeness (Bird et al 2014, Lewandowski and Specht 2015, Haworth 2016, Aceves-Bueno et al 2017, Haworth et al 2018. This study therefore is aimed at evaluating the data quality of the citizen-based inventory in the Rwenzori Mountains. Focusing on landslides and floods, we validate the CSs reports through systematic fieldwork, analyse the precision and sensitivity of the inventory, and compare these to an inventory constructed using visual interpretation of very high-resolution satellite imagery. Finally, we identify potential underlying factors determining the detectability of landslides by CSs.

Study area
Rwenzori Mountains are located in tropical Africa (Peel et al 2007) across the southwestern border of Uganda and eastern D.R Congo. This mountainous region suffers from frequent landslide (Jacobs et al 2016a) and floods (Jacobs et al 2016b). The Ugandan side of the mountains is characterized mainly by protected zones (national parks, forest, and wildlife reserves) and inhabited zones (figure 1). This study focuses on landslide and flood events on the inhabited Ugandan side of the Rwenzori Mountains, specifically those occurring in the two districts of Kasese and Bundibugyo.

Landslide and flood inventory 2.2.1. Citizen science-based inventory
The CS based inventory was established from a selection of 18 CSs active in 30 selected parishes out of the 41 covered by CSs in the study districts (figure 1). In selecting the parishes, in addition to ensuring a good spatial spread over the two study districts, considerations were made to ensure representation of both parishes from which many reports had been received and those with little or no reports at all. The methodology for data-collection on landslides and floods is described in . This dataset contains among others, details on the location (GPS coordinates), timing and damage inflicted by a hazard event (individual landslide or flood occurrence) and is typically collected within a timespan of a few days after occurrence. Although the data collection started in February 2017, we here narrow the dataset to landslides and floods reported between 1 May 2019 and 30 May 2020 because landslide scars, particularly for shallow and smaller landslides, are not always preserved in the landscape due to quick vegetation regrowth or reclamation by agriculture. As soon as the hazard reports are submitted by the CSs, they go through a manual validation process by a research assistant to remove reports with obvious errors like presence of incorrect hazard pictures (pictures not related to the hazard being reported) and suspicious impact . This was followed with automated screening using a Python script to identify potential inconsistences (including GPS position precision, date of recording compared with date of occurrence). In the 1 year period considered in this study, 292 and 75 landslide and flood reports had been submitted by the CSs. After the manual and automated screening, the consolidated dataset consisted of 255 landslide and 59 flood reports.

Satellite imagery-based inventory
The remote sensing (RS)-based inventory was built for the same period (1 May 2019 to 30 May 2020) as the retained CS-based inventory from orthorectified 3 m resolution visual optical images of Plan-etScope (Planet 2016). PlanetScope offers among other products visual geometrically and atmospherically corrected 3-band; Red Green Blue (RGB) and analytic 4-band (RGB and Near Infra-Red) images for different places across the globe with a one-day revisit time (Planet 2021). The imagery is collected as a series of framed overlapping scenes from a single satellite in a single pass (Planet 2019). To ensure the images acquired were suitable for visualization application, a 15% cloud threshold was used before download. However, due to persistent cloud cover, it was not possible to get cloud free images covering the two study districts at the same time at all times. For each clear image of one district, the nearest in time image of the other district was used. From each image (appendix A), potential landslides and flood events were digitized in a GIS environment. By overlaying the already constructed shapefile of digitized events on a successive satellite image, double digitizing was avoided. A total of 458 landslide and 59 flood potential events respectively were digitized from the satellite images.

Expert-based field assessment and validation
From mid-September to mid-December 2020, a field assessment and validation was conducted by the first author to systematically map landslides and flood occurrences (figure 2) in the 30 selected parishes and verify the events reported in the CS database and those events identified through visual interpretation of satellite imagery. For each parish, a local guide was hired to ensure all the possible areas of the administrative unit were reached. This facilitated mapping as many landslide and flood events that occurred in the period of May 2019 and May 2020 as possible. The CSs reports, and the RS-based inventory were also verified by systematically visiting these locations.

Data validation
Each event that would be found to exist in the field and correctly identified by the CSs and/or mapped from the satellite imagery, was labelled True Positive (TP). Those erroneously mapped as landslide or flood events or found not to exist in the field, were labelled False Positives (FP). Events that were found to exist in the field but had been missed by either the CSs or the RS inventory process were labelled False Negatives (FN). The process is illustrated using a hypothetical field space in figure 3. These classifications were then used in subsequent analyses outlined below and summarized in the schematic figure 4.

Sensitivity, precision, F1
Sensitivity (also called true positive rate or recall rate) is used to refer to the proportion of existing positives that are correctly identified or detected as positives (Powers 2008, Tharwat 2018, Chicco and Jurman 2020. In previous studies, sensitivity has been used to measure the performance of diagnostic tests (Genders et al 2012) and classification performance in computational linguistics (Powers 2008). In this study, sensitivity is used to refer to the proportion of landslides/floods that occurred and were correctly reported by CSs and/or mapped from satellite imagery. It is defined as given in equation (1).
Precision (confidence) has been applied in information retrieval and data mining studies to measure the predictive performance of models. It refers to the proportion of predicted positives that are truly positives (Powers 2008, Boughorbel et al 2017. In this study it denotes the proportion of detected landslides or floods that were real landslides or floods in the field. It is calculated as given in equation (2). .
(2) Figure 3. Illustration of a hypothetical field space where brown dots represent events (landslide/floods) that occurred, green dots represent CSs mapped hazards while blue dots represent RS-based inventory. TP represent events found to exist in the field and correctly identified by the CSs or mapped from the satellite imagery. FP represent features erroneously mapped as landslide or flood events or found not to exist in the field while FN represent events that were detected in the field but had been missed by either the CSs or the RS inventory process. The F 1 score (F measure) is an integration of both sensitivity and precision. It gives the harmonic mean of sensitivity and precision and ranges between 0 and 1 (Tharwat 2018). Although the metric has received some criticism (Flach and Kull 2015), it is still a widely used measure of classification or predictive performance studies (Genders et al 2012, Chicco and Jurman 2020, Chicco et al 2021. Here it was applied to measure the overall performance of CSs and satellite in detecting landslide and flood occurrences. The higher the F 1 score, the better the detective performance of the hazard inventory approach. It is calculated as in equation (3).
The comparison of the CS-based-and satellite-based inventories with the field data and subsequent calculation of sensitivity, precision and F 1 statistics allowed a direct contrasting of the data quality of the two former inventories (figure 4).

Spatial bias analysis
CS-based inventories can be affected by spatial biases (Zhang and Zhu 2018). In our study, the landslide inventory in particular could be affected by biases: landslide events are spatially more dispersed than floods and are more frequent but smaller in size, which could render them less noticeable. CSs might have more knowledge or be more likely to visit areas closer to their homestead (Pernat et al 2021). In addition, landslide inventories by CSs but also by experts (Dickinson et al 2010) can be affected by a bias towards those areas closer to the road network (Wang et al 2020). This creates incompleteness issues in such citizen-based databases. Here we examine the potential sources of spatial biases by analysing factors that determine whether a citizen scientist detects and records a hazard event. To do this, a binary logistic regression model was applied. Logistic regression is a widely used modelling approach in determining relationships between a binary dummy dependent variable and explanatory variables (Coughlin et al 1992, Genders et al 2012, Nishadi 2019). In our case, the binary dependent variable is whether or not a landslide that was confirmed in the field was reported by the citizen scientist (1) or not (0). If landslide reporting ability is represented as Ls report and X 1 , X 2 , X 3 , …, X n represent a set of potential explanatory variables, the probability of a landslide being reported by CSs is given as shown in equation (4). α+β1X1+β2X2+β3X3+.............βnXn) . (4) where P ( Ls report ) is the probability of a landslide being detected and β 1 to β n are regression coefficients. The first set of explanatory variables used in equation (4) model were landslide related factors. These included landslide size, slope, altitude, landslidecitizen scientist home distance, landslide-road access distance and impact (weighted sum of effects on humans, infrastructures and crops, appendix B). Although even a low impact landslide event can be detected and reported by a citizen scientist, the more disastrous an event is, the higher is the chance that such an event will be discussed on different platforms and the easier it might be for a citizen scientist to learn about it and detect it. The regression analysis was repeated while accounting for a fixed effect of CSs (citizen scientist specific unique attributes). To achieve this, unique identity (citizen scientist ID) numbers were introduced in the model. To gain an insight on how much variance is explained by each of the factors included in the regression analysis, a univariate logistic regression analysis was performed for each of the variables. The Tjur's coefficient of discrimination, D, was then used as an indicator of the individual contribution for a specific variable in explaining variance in landslide detectability by the CSs (Tjur 2009, Allison 2014). We then investigated which factors could explain variations in the observed individual odds ratios of CSs. This explorative analysis was done through a Spearman rank correlation matrix involving individual odds ratios of CSs, age,

Landslide and flood detection
Out of the 255 and 59 landslides and floods retained in the CS-based inventory after manual review and automated screening of incoming reports, 221 landslides and 46 floods were verified in the field. Of the 459 landslides and 59 floods identified on satellite images, 373 and 57 respectively could be verified in the field (table 1). The remaining instances could not be verified due to accessibility constraints in the field and were not considered in further analysis. Results show that CSs identify and report landslide and flood hazards in their localities with 99% and 100% precision respectively (table 1). Table 1 however indicates also a relatively low sensitivity for CS-based landslide and flood detection (51% and 70% respectively). RS-based landslide and flood detection is less precise particularly for landslides (39%) in addition to having even lower sensitivity. RS-based flood detection has a higher sensitivity compared with CS. Overall, floods are more easily detected by both CSs and RS as implied by F 1 scores of 0.8 and 0.8 respectively. Table 2 shows a direct comparison between CS and RS inventories in the form of a confusion matrix. Only 16% of the 425 landslides mapped in the field were detected by both satellite and CS and 26% were detected by neither inventories. Remarkably, 35% of the landslides was detected by the CS but not by the satellite interpretation and 23% of satellite detected landslides were not identified by CSs. The high sum of this diagonal demonstrates that both inventories are to a large extent complementary and typically do not detect the same events. While the average size of the landslides detected and not detected by the CSs is not statistically significant (Prob > z = 0.088), satellite imagery detects larger events and tends to miss out on the small ones (Prob > z = 0.000). Through logistic regression (appendix C; table 8), it was found out that landslide detection by RS is significantly influenced by  landslide size, slope and altitude (Tjur's D = 0.226). A marginal effect post estimation showed that the chances of missing out some landslides on a satellite image increase as the landslide area reduces.

Factors that influence citizen science-based landslide detectability
Results of the logistic regression show that the variation in landslide detectability by CSs is significantly influenced by the landslide's impact, the distance between citizen scientist's residence and the landslide, the distance from the access roads to landslide and altitude (table 3). The odds ratios confirm the hypothesis that larger impact and shorter distance increase the likelihood of landslide events being detected. However, based on the Tjur's D, these factors only explain a small proportion (Tjur's D = 0.138) of the variations in landslide detectability. Through univariate analysis, a Tjur's D for each factor was generated to gain an understanding of the maximum contribution of each individual factor to the variance in landslide detectability by CSs (table 3 column 3). Based on the univariate analysis results, the landslide distance from the citizen scientist's home plays the most important role, followed by impact and slope in influencing landslide detectability.
To check potentially controlling factors related to the individual CSs included a fixed effect for each individual in the logistic regression. This increased the explanatory power of the model (Tjur's D = 0.435 from 0.138). In a final step, we explored which personal characteristics could influence the resulting odds ratios by simple correlation with the variables as presented in table 4.

Sensitivity and precision in CS-and RS-based hazard detection
CSs can correctly identify and report georeferenced hazard information comparable to research experts. Although no control trial was done to quantify the added value of the training workshops, this high precision rate after training workshops and in-field practice concurs with what Lewandowsli and Spencht (2015) report on the importance of adjusting protocols and volunteers training in improving data quality (Lewandowski and Specht 2015). According to Paul et al (2014), CS-based data collection programs result in datasets with quality attributes similar to those established systematically by experts (Paul et al 2014). However, their inventories (CSs) suffer incompleteness challenges especially those that involve mapping small, frequent but spatially spread events like landslides (Rohan et al 2021).
Despite the relatively low sensitivity, the CS database outperformed the RS-based inventory in terms of sensitivity, precision and F 1 scores when considering landslides. The low sensitivity and precision of RS-based inventory compared to CSs and other studies are due to several factors. We used satellite images at a 3 m resolution while other studies used Google Earth images that offer a 0.3-0.6 m resolution (Hao et al 2020, Ubaidulloev et al 2021. The images available in Google Earth are usually acquired with a lower temporal resolution compared with the PlanetScope images, and therefore may capture landslide scars that are already (partially) concealed by the regeneration of the vegetation (Dewitte et al 2022). This reduction of the spectral signature of the landslides identified via Google Earth is however counterbalanced by the higher spatial resolution of the images. In addition, CSs visit the field after a few days, allowing them to identify very small landslides that cannot be seen on a satellite image. Furthermore, the inventory was limited to areas monitored by CSs. These areas are characterized by cultivated landscapes made of parcels that present shape that could be mistaken for landslides on the satellite image. The confusion matrix (table 2) demonstrates that the CS inventory and the RS inventory are largely complementary. However, the CS inventory in turn is affected by biases predominantly relating to a fixed individual citizen scientist effect and to lesser extent relating to the landslide characteristics, impact, and location. The further the event from the residence of the citizen scientist or the community access roads, the higher are the chances of such an event being missed. The magnitude of impact and nature of the element impacted are also important. This supports our hypothesis that more destructive events are more readily communicated or discussed on different platforms and so easier to detect by CSs. People are more motivated to report events that affect their houses or cause injuries/death than those that affect crops and pastures. This could partially be due to the perception that CSs reports reach humanitarian agencies who eventually distribute relief items mainly to those whose houses get damaged. Most landslide events that affects houses in the study area are usually small resulting from (a) steep slope cuts while establishing platforms for house construction and (b) small scale cropping farming on the steep slopes. This partially explains why CSs report more of small events. The observed improvement in the explanatory power of the model after introducing the citizen scientist ID implies that CSs specific characteristics (here bundled into one fixed effect) play an important role in influencing the likelihood of landslides being detected. Thus, while CSbased inventories might depend on socio-cultural, infrastructural, motivational factors of the CSs and nature of the hazard, satellite imagery largely depends on nature of the target object in the feature space (size of the object) and the prevailing atmospheric conditions (Agapiou et al 2011, Selva 2021. In general, it should be noted that the ability of CSs to detect landslides that have occurred largely depend on whether they get informed about their occurrence. The positive correlation between number of reports and age of CSs (table 4) can be explained by the fact that the older the CSs are, the higher the chance of them having a stable residence in their village and the wider the social network built from whom news about landslide occurrences can be received. In addition, older CSs tend to stay longer in the reporting network as indicated by a strong positive correlation (r = 0.55) between years of operation and CSs age (table 4).
CS-based inventories and RS-based inventories not only differ in which landslides are detected, but both inventory methods also fundamentally differ in the type of information that can be collected: CSs can collect data on the impact associated with the hazards that is not visible through remotely sensed information. RS-based mapping on the other hand provides an opportunity to record geo-hydrological hazards even in inaccessible areas (Ji et al 2020). On the other hand, while revisit periods of very high-resolution satellites are shortening, image interpretation is oftentimes still hampered by cloud coverage, while CSs tend to visit, and map affected sites within few days after occurrence . The implication here is that CS-based landslide detection and very high-resolution (⩽3 m) satellite imagery are complementary in spatial extents, types of landslides and types of information collected, but not substitutes. Because of this complementarity, CS data represent an added value to hazard assessment and risk analysis for example in determining thresholds for landslide triggering factors (Monsieurs et al 2019) and supplementing incomplete landslide inventories (Samodra et al 2018). CSs generate reliable georeferenced hazard information including information on timing and damage inflicted, particularly in data-poor countries like Uganda where extensive disaster risk is high. Such events do not easily make it to social or formal media platforms in addition to not meeting the impact thresholds of the existing national (DesInventar) data repositories (UNDRR 2019b) and global (EM-DAT) databases (IFRC). To reduce incompleteness challenges, it is relevant for projects and programs aimed at spatiotemporal explicit hazard assessment to complement high spatial resolution satellite image-based hazard inventories with CS hazard sensing. By visiting the affected scenes within a few days and documenting the impact caused, CS-based inventory can be useful in temporal landslide assessment and vulnerability analysis which are prerequisites for risk analysis.

Limitations of the data used in the analysis
Although we find the quality of the CS-based inventory of good quality, maintaining a big network of CSs to cover a large geographical area over time can be challenging and require investment to equip, train and continuously monitor the activities of CSs. Although CSs report geo-hydrological hazards with high precision, the inventory should be carefully used for regional susceptibility and risk assessments as it is not free from spatial biases. In addition, the satellite imagery digitization was entirely made by the main author of this manuscript. Although efforts were made to do it systematically, it does not completely erase possibility of omitting or including certain features that would respectively be or not be included by other users.

Conclusion
This study demonstrated the reliability of CS-based hazard detection and its complementarity with RSbased inventory. It also showed that by integrating CS-based hazard inventories with existing high resolution satellite imagery in spatio-temporal geohydrological hazard studies can help to overcome the incompleteness challenges. The Sendai framework for DRR advocates for inclusive national and local hazard assessment and risk management (UNISDR 2015b) and so CS provides a viable approach not only for rural and remote mountainous areas such as the Rwenzori ranges, but for all at-risk regions where traditional observation networks and infrastructures are weak or absent and local communities directly impacted. The network presents a desired multidimensional approach in cogenerating knowledge and disseminating research findings and could enhance participatory disaster risk management at local level.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors. Appendix A. List of dates and months (table 5) in which images were acquired to build the RS-based inventory Landslide impact is an important variable to consider while analysing factors that influence whether an event can be detected and reported or not in citizen science (CS)-based hazard assessment. Although event a low impact landslide event can be detected and reported by a citizen scientist, we hypothesize that the more disastrous an event is, the higher is the chance that such an event will be discussed on different platforms and the easier it will be for a citizen scientist to detect it. The nature of impact and type of element affected also matters. For instance, an event that involves a fatality attracts more public attention than the one involving only crop damage. Similarly, an event that blocks an access road can be easier to detect than the one that only cause it to crack. Therefore, using type of element affected, nature of impact and magnitude of impact caused criteria, different weights were assigned based on experience and expert judgement as shown in tables 6 and 7. For each type of element affected, the weights assigned were converted to relative weights by dividing the biggest wight assigned to the smallest so that all values range from 0 to 1. Since one landslide event can impact on multiple elements, the relative weights of different affected elements for each of the respective event were added together to come up with the total impact value for that particular landslide event.
Appendix C. Logistic regression results (table 8) and marginal effect post estimation (figure 5) to determine threshold for landslide area effect on landslide detectability using satellite imagery A logistic regression was performed for landslide detectability through satellite imagery as a dependent variable and slope, landslide area, altitude and NDVI as independent variables. Slope, landslide area and altitude influence positively on the detectability. The marginal effect post estimation results (figure 5) show that as the landslide area reduces, the chances of missing out some landslides on a satellite image increase.