All of the data sources used to create Project Badweyn are fully cited in the informational pages associated with the map categories. Analysis of and changes to data by Secure Fisheries, besides purely aesthetic choices, are described here. We limited the spatial extent of data analyzed and displayed to the Somali region. Analyses were executed in ESRI ArcGIS Pro 2.9.2 unless otherwise indicated.
The shoreline shapefile (version 2.3.5) was obtained from the Global Self-consistent, Hierarchical, High-resolution Geography (GSHHG) Database (http://www.soest.hawaii.edu/wessel/gshhg/). We used that shoreline as the baseline for creating a 24 NM buffer to display boundaries of the 24 NM Territorial Sea.
The data to create the line for Claimed Territorial Waters (1972-2014) was downloaded from www.marineregions.org (version 11).
The line for the Exclusive Economic Zone (2014) was created using the coordinates published by the United Nations (http://www.un.org/depts/los/LEGISLATIONANDTREATIES/STATEFILES/SOM.htm) without changes. The outline is displayed without changes.
The line of maritime delimitation between Somalia and Kenya was derived from coordinates in the ruling by the International Court of Justice. Maritime Delimitation in the Indian Ocean (Somalia v. Kenya) judgment on October 12, 2021 (https://www.icj-cij.org/en/case/161).
A recent, original survey of Somali coastal habitats does not exist. We relied on publicly available datasets based on satellite imagery and downloaded from the internet. For ease of use of Project Badweyn in low-bandwidth situations, high resolution raster data was converted into simplified polygons. In the case of the seagrass data, a very detailed polygon file was converted to a lower resolution raster data set, then re-converted back to a polygon for display. Thus, these maps should be used for informational purposes only and not as an exact map of habitat distribution nor for any spatial analysis. Rather, the map depicts the areas of these habitats for the general information of users.
Coral reef extent was modified from the Allen Coral Atlas Coral Reef Extent map created for the Global Coral Reef Monitoring Network (GCRMN) and produced by Center for Global Discovery and Conservation Science (GDCS) at Arizona State University.
Mangroves and herbaceous wetlands were derived from data from © ESA WorldCover project 2020 / Contains modified Copernicus Sentinel data (2020) processed by ESA WorldCover consortium.
Areas of seagrass were generalized from the Allen Coral Atlas habitat mapping. Data were rasterized to 0.005 degree cells, then converted back into polygons for ease of viewing online. The polygons show a simplified seagrass area and should not be taken as exact seagrass locations. The Satellite Coral Reef Mosaic is © 2022 Planet Labs and licensed CC BY-SA-NC 4.0 (https://creativecommons.org/licenses/by-nc-sa/4.0/). See the Allen Coral Atlas for the original data and methods.
Key Biodiversity Areas are only those in the Somali region from the Global Dataset containing the current boundaries for Key Biodiversity Areas (KBAs). Project Badweyn contains the September 2020 edition of the The World Database of Key Biodiversity Areas (WDKBA) Spatial Dataset.
Coastal Fishes, Highly Migratory Fishes, Invertebrates
Fisheries independent data about marine species distributions are not available for the waters off the Horn of Africa. As a proxy, we use the amount of catch estimated by the Sea Around Us project using their reconstruction method (http://www.seaaroundus.org/catch-reconstruction-and-allocation-methods/). These catch numbers provide enough information to show the distribution of commercially important species groups.
To create the distribution maps displayed in Project Badweyn, we downloaded the reconstruction data for the most recent five years available (2010-2014) from the Sea Around Us project’s R library (http://www.seaaroundus.org/tools-guide/#rlib). These data are disaggregated into 0.5 x 0.5 degree cells. The species present in the downloaded data for our study area were aggregated into one of 23 commercially important groups. We summed the catch (in metric tons) for all species in an aggregate group in each cell for each year, then found the average catch per cell over the five years. Because the average was often small, the visualization uses a logarithmic scale (<1, 1-10, 10-100, 100-1000, >1000 up to the maximum for each group) and is color coded accordingly.
Maps of Somali landing sites exist in a variety of publications for different time periods and regions, but we created the first comprehensive, modern map of artisanal landing sites along the Somali coast. For this purpose, we defined landings sites as places where small, artisanal Somali boats bring their catch to shore at some point during the year. These are most often on beaches where fishers pull their boats onto the land accompanied by mooring sites in the water just off the beach. To identify landing sites, we visually searched satellite imagery for small boats on the beach or moored in the sea. Because we have not been able to ground truth these locations or the actual numbers of active fishing vessels, the counts associated with each site are approximations.
To create the map of landing sites, we used plugins for Google and Bing satellite imagery in QGIS, a free and open source Geographic Information System software program (www.qgis.org). At a zoom of around 1:2000, we scanned the Somali shoreline for small boats, starting at the northwestern border with Djibouti and covering the entire coastline to the southern border with Kenya. Where boats were found, the coastline was marked with a point to record latitude and longitude, and the name of the site was recorded if it was documented in the Google maps, from consultation with Somalis or the United Nations Food and Agriculture Organization, and by comparison with UNOCHA’s list of Somali settlements. The number of boats on the beach or within approximately 100m of the shoreline was counted in both Bing and Google images. If boats were visible on the water but there was no obvious landing or mooring site nearby, they were considered active, working boats and were not counted because their landing location was unknown.
The date of each Google image was obtained by loading the sites into Google Earth and zooming in to each location until the date of the image was visible. For Bing images, the date of the image was obtained by finding each location on the Bing Aerial Imagery Analyzer for OpenStreetMap (http://mvexel.dev.openstreetmap.org/bing/). When there was a range of dates given by the imagery analyzer, the earliest date was used. To get one number of boats to include in the data visualization, we used the most recent image if the dates were more than 5 years apart. If they were less than 5 years apart, we used the image with the most boats. This decision rule was based on balancing using the most up-to-date image while recognizing that vessels move regularly. Occasionally, this resulted in a site that had zero boats. In that case, the site remained on the map, but further investigation will be necessary to determine if it is an active, defunct, or temporary landing site.
Size of landing sites was classified by the number of boats. Each site was placed in one of four categories: 0-5 boats, 6-30 boats, 31-60 boats, and 61-210 boats. These designations were determined by the natural breaks in the histogram of number of boats and color coded and sized accordingly.
Location names and districts were determined by spatially joining landing sites to the Somalia settlements data available from the United Nations Office for the Coordination of Humanitarian Affairs Somalia (https://data.humdata.org/dataset/somalia-settlements-2012).
Foreign Fishing Areas – Highly Migratory Species
We created a simplistic representation of the intensity of foreign fishing for highly migratory species off the Somali coast by aggregating the purse seine catch data from the Indian Ocean Tuna Commission’s public database (http://iotc.org/data/datasets). These data are reported in 1 degree latitude and longitude cells. We summed the data in each cell over the entire time period available to clearly show the area of greatest catch. For the visualization, each cell was placed in one of five categories determined by the natural breaks in the histogram of catch and color coded accordingly.
Foreign Fishing Areas – Coastal Species
The map of foreign fishing for coastal species shows estimated areas of trawling based on parameters from the analysis of South Korean trawling vessels off the coast of Somalia published in Securing Somali Fisheries. Those vessels were found to be fishing only off the coast of Puntland, and our analysis found that they fish in areas shallower than 100m. Our research and personal communications indicate that other trawling vessels fish in the southern Somali waters near the border with Kenya and along the northern coast in the Gulf of Aden off Somaliland. Assuming that the trawling vessels operating in Somali waters are of similar or smaller size and use similar fishing gear, we created a polygon depicting the area from the shoreline to the 100m depth contour as a visualization of the potential trawling area in those regions. It is important to note that these areas are not actual trawling areas, but are theoretical fishing grounds given the limited information available.
We are not displaying the range of foreign gillnetters because we do not have reliable accounts of where those vessels operate, although they do fish in Somali waters and are responsible for a large proportion of the foreign catch of coastal species. For more information on those vessels, see Securing Somali Fisheries.
Shipping Traffic Density
In order to visualize the density of shipping traffic in the Horn of Africa region, we used automatic identification system (AIS) data, provided by ExactEarth (www.exactearth.com) for January-April 2016, for all non-fishing vessels in our area of interest. These data are satellite detections of vessels that broadcast identifying information and their position at intervals while they are in transit, anchored, or docked. To show the areas used most frequently by moving vessels, we excluded any points that had an associated speed over ground of zero in order to eliminate vessels that were anchored or in port. There were 49,617 total detections over the entire area and time period.
To display the density of shipping traffic, rather than individual ship detections as points, a 100 x 100 cell fishnet (grid of squares) was created for the area of interest. The AIS data were joined to the fishnet by spatial location to calculate the number of ship detections in each square of the grid. For the visualization, each square was placed in one of five categories determined by the natural breaks in the histogram of detection density and color coded accordingly. Squares with 0 or 1 detection are not colored.
The ports displayed are major ports designated as such in the Somalia Joint Needs Assessment by the United Nations (http://www.somali-jna.org/downloads/SJNAICR090906%20-%20Infrust%20Part%20I.pdf).
The ‘fish wars cycle’ (Pomeroy et al., 2016) provides a foundation for understanding key drivers of fisheries conflict. Conflict around fisheries is perpetuated by three top-level components: competition over fisheries, existing levels of conflict, and fisheries scarcity. Each top-level component is defined by a host of quantifiable variables. For example, competition over fisheries is affected by the presence of both commercial and small-scale fishing fleets. Existing conflicts may include user-group violence and crime against fishers. Fisheries scarcity is affected by poor resource governance. Secure Fisheries researchers adapted this cycle by reducing the number of variables and defining them to be suitable for media coding.
Fisheries conflict was evaluated by reported fisheries dispute events (FDEs). An FDE is an incident in which a fisheries resource is contested, disputed, or the source of conflict between a minimum of two actors, at a discrete temporal moment, and in a discrete location. Actors could be individuals or groups (spontaneous or organized). FDEs occur over a defined period of time and in a bounded location. Media reports were coded for relevant variables to collect event-level fisheries conflict data in six countries in the Horn of Africa-East Africa (Tanzania, Kenya, Somalia, Yemen, Djibouti, and Eritrea) during 1990–2017 (Yemen and Somalia data have been updated through 2020). Researchers quantified the frequency and intensity of these conflicts, categorized their causes, and measured their consequences.
For more information, read “The causes and consequences of fisheries conflict around the Horn of Africa” (Devlin et. al., 2021).
Universities Collecting Catch Data
The locations of universities involved in data collection are general points in the cities where the universities are located and do not represent the actual locations of the universities. Catch data reported in each popup when a university point is clicked are aggregated catch compositions over the entire time period a university has been collecting data. Students collect catch data from three to four boats once per month at each location. The displayed pie charts show the percent by species group of the total number of fish counted since data collection began in late 2018 (the exact date varies by university) until October 2021.