Presentation on theme: "The Evolution of Aerial Strip Sampling into the Digital Era Adaptations over 40 Years and Contemporary Relevance Somalia and the Horn of Africa R.M.Watson."— Presentation transcript:
The Evolution of Aerial Strip Sampling into the Digital Era Adaptations over 40 Years and Contemporary Relevance Somalia and the Horn of Africa R.M.Watson J.M.Nimmo P.D.Crees email firstname.lastname@example.org, email@example.com@firstname.lastname@example.org www.rmruk.com;www.balloonaerialphoto.com RESOURCE MANAGEMENT AND RESEARCH Social & Environmental Information Specialists Littlebourne Forest Green Dorking Surrey RH5 5SQ UK
About 60 years ago scientists started to use aircraft to count wild animals. Counts of entire populations were usually too difficult, so types of sampling were developed. Strip samples used the aircraft’s inherent movement most efficiently. i.e. maximum area observed per hour of flight Sampling design, bias and precision estimations, operational aspects (speed, height, location, use of photography etc.) were studied and the method has become one of the standard tools of large animal research
Figure 1-Simplified Diagram Aerial Strip Sampling To Count Animals
The semi-arid deserts, savannahs and woodlands of the Horn of Africa were ideal for expanding the method to cover a wide number of resource features. KREMU (now DRSRS) from about 1977 used teams of technicians in 6-seat aircraft to manage the increased demands of multi-feature spotting and enumeration/ attribute definition/ quantification. RMR used a single pilot-observer using cameras, tape-recorders, stop-watches, recording radar altimeters in an aircraft modified for very low speeds and long endurance between 1969 and 1985, to make 2.5% to 5% sampling surveys over a large part of the Horn of Africa, about 4.5 million km 2. (About 1 million km of flying over about 10,000 hours).
Figure 2 – RMR strip sampling resource surveys carried out in the Horn of Africa 1968 - 1989 1974-76 1971-73 1971-74 1968-73 1978-89 Shirre Lowlands Central Highlands N.E.Rangelands Giggiga Region
A typical output table from the 1977-85 SOMRAAS is shown below. The full set of reports can be seen at this workshop by anyone interested at the end of the session.
The 1979-85 survey raw data provided estimates, for each of 496 land system units of the country, of 136 features, some of which were assigned up to 4 attributes.
GIS technology and methods were in their infancy at the start of these surveys. If the data tables of these reports were to be digitised now using the LSU boundaries as a geo reference framework they could provide an invaluable record of the status of Somalia’s rural population and production systems before the current upheavals. The 1991 D.Phil Thesis of the second author shows how a geo-referenced database of the survey results could become an instrument for planning and development. Contemporary digital methods now make this goal much easier to accomplish.
These methods were developed and the surveys were made when national long term planning still took place, or was, at least, an aspiration. They were paid for by the World Bank, the Governments, USAID and FAO. They were not used in the Horn of Africa for forming policies and plans to the extent possible (and expected) because: 1.Financial resources aimed at longer term, high risk and low yielding opportunities, typical of “development” investment, have become progressively more difficult to find. 2.National planning, and the application of a national resource “information infrastructure” to decision making does not attract political support under current governance conditions. 3.Multi-feature aerial strip sampling by analogue methods was too difficult and became too dangerous to be a commercial method. It remained a research tool.
Some idea of the technical difficulties can be assessed from the next slides which show: 1. The passage of a typical aircraft sampling strip over a model landscape, with an explanation on the following slides of the processes and analytical procedures needed in the pre-digital era, both in the aircraft and in the survey office. 2. Photographs, tape recordings, annotated maps and radar altimetry records in analogue forms were the products of an aerial strip sampling. 3. Each product had to further analysed before it reached a stage it could become digitisable. 4. When PC’s became available the final stages of analysis were fully digitised.
How aerial strip sampling for resource surveys was carried out in the 1980’s A methodology at its analogue limits.
An animated model of an aerial strip sampling from 1983
Example ‘Photograph’ frame with animals marked and counted.
ObservationSound TrackTranscriptionPost-Transcription Aircraft positioned at the start of the sampling transect. Start Transect 1 in Stratum 5 on 6 February 1983. Time 08. 09.53 1/5, 6-02-83, 08.09.53 Mark on 1:100,000 map. Enter time. (T1) Sheep and goats photographed with 2 frames : herd is fully in the strip. Frame 3 and 4 of film 36 sheep and goats in one in thorn enclosure F36/3,4, S+G (1) (E) Photographs (marked for herd overlap) counted under microscope. Number entered as n* vegetation/species bias factor based on photograph and computed as density (no./km 2 ). Attributes computed as % for sheep and goats. Inner transect marker cuts across a grazing reserve and is timed 3 seconds of fenced grazing reserve with 5% unfenced garden area 3sGR(5%Gdn)(F) Enter and compute as % of total time flying transect. Attributes computed as % for GR Balli (natural pond) with water in strip One unenclosed balli with water in one 1 Bl (1) (W) (NE) Enter and compute as density (no./km 2 ). Attributes computed as % for Bl Sheep and goats photographed with 1 frame: herd is partly in strip when cut by decision line. Frame 2 of film 36 sheep and goats in two and riverine pool in two. The riverine pool and the goats-sheep are cut by the outer demarcator F36/2, S+G (2) (RP) (UE) 1RP (2) (UE) Photograph counted under microscope (see Slide 9). Number entered as n* vegetation/species bias factor based on photograph* 0.5 and computed as density (no./km 2 ). Attributes computed as % for sheep and goats. Riverine pool entered as 1*0.5 and computed as density (no./km 2 ). Attributes computed as % for RP. 12 camels counted by eye in strip 12 camels in one in riverine palm grove 12Cm (1)\(RvPG) Enter and compute as density (no * palm grove vegetation/species bias factor/km 2 ). Attributes computed as % for camels Riverine pool in stripOne riverine pool in one1 RP (1)(W) Enter and compute as density (no./km 2 ). Attribute computed as % for RP Inner transect marker cuts across a dry river bed One dry river bed1 RvB (D) Enter and compute as no. of intersects/km of transect
ObservationSound TrackTranscriptionPost-Transcription Inner transect marker cuts across a balli One enclosed balli with water in two 1 Bl (2) (W) (E) Enter and compute as density (no.* 0.5/km 2 ). Attributes computed as % for Bl Sheep and goats photographed with 3 frames : herd was in two strips when cut by decision line Frame 6 to 8 goats in two browsing in Salvadora thickets F36/6-8(2) S+G (Br Slv.p thkt) Photographs (marked for herd overlap) counted under microscope. Number entered as n* vegetation/species bias factor based on photograph*0.5 and computed as density (no./km 2 ). Attributes computed as % for sheep and goats. Inner transect marker cuts across a fenced field and is timed (Outer transect becomes the demarcation line on next transect) 8 seconds of recent fallow in fenced 8s RF (F) Enter and compute as % of total time flying transect. Attribute computed as % for RF. 6 seconds of growing beans in fenced 6s GB (F) Enter and compute as % of total time flying transect. Attribute computed as % for GB. 10 seconds of harvested maize in fenced 10s HMz (F) Enter and compute as % of total time flying transect. Attribute computed as % for HMz. Other fields outside stripNone House in the strip 1 Ridge roofed house with grass in one, not enclosed. 1RRG (1)(UE) Enter and compute as density (no /km 2 ). Attribute computed as % for RRG 1 house cut by strip demarcator 1 Ridge roofed house with tin in two, not enclosed. 1RRT(2)(UE) Enter and compute as density (no *0.5/km 2 ). Attribute computed as % for RRT One 4WD vehicle in stripOne 4wd vehicle in one1 V(1) (4WD) Enter and compute as density (no./km 2 ). Attribute computed as % for V Inner transect marker cuts across a vehicle track One vehicle track recently used VTr (R) Enter and compute as no. of intersects/km of transect. Attribute computed as % for tracks. Well in the strip 1 hand-dug well in use for watering 1HDW (1) (W) Enter and compute as density (no /km 2 ). Attributes computed as % for HDW. Cattle counted by eye in the strip 9 cattle in 1 being watered at well in open 9 C (1) (HDW) Enter and compute as density (no * open vegetation/species bias factor/km 2 ). Attributes computed as % for cattle. Donkeys counted in strip 8 donkeys in one close to hand-dug well, open vegetation 8 D (1) (HDW)(OV) Enter and compute as density (no * open vegetation/species bias factor/km 2 ). Attributes computed as % for donkeys. 1 abandoned livestock enclosure in strip 1 long abandoned livestock enclosure in one 1B(1)(AbAb) Enter and compute as density (no /km 2 ). Attribute computed as % for B
ObservationSound TrackTranscriptionPost-Transcription 2 berkado in strip and one outside with no water 2 dry berkado unfenced with sediment removed in one 2Bk(1)(2D,1SdExc )(UE) Enter and compute as density (no /km 2 ). Attributes computed as % for Bk 1 berkad in strip with water in it 1 wet berkad unfenced in one1Bk(1)(W) Enter and compute as density (no /km 2 ). Attribute computed as % for Bk Cattle counted by eye at the berkad 3 cattle in one at wet berkad being watered in the open 3C(1)(Bk)(W) Enter and compute as density (no * open vegetation/species bias factor/km 2 ). Attributes computed as % for cattle. 2 aqallo in strip no longer occupied 2 abandoned aqallo in one, no enclosure 2Aq(1)(Ab)(UE) Enter and compute as density (no /km 2 ). Attributes computed as % for Aq. Cluster of 6 aqallo in an enclosure cut by outer demarcator 6 occupied aqallo in two in an enclosure 6Aq(2)(E) Enter and compute as density (no * 0.5/km 2 ). Attribute computed as % for Aq. 1 livestock enclosure currently in use but no livestock present in 2 strips 1 livestock enclosure in use in two no livestock present 1B(2)(Oc) Enter and compute as density (no * 0.5/km 2 ). Attribute computed as % for Aq. Inner transect marker cuts across an A. bussei woodland and is timed 6 seconds of A. bussei woodland6s A.bssEnter and compute as % of total time flying transect Herd of topi in the strip photographed Frame 5 topi in one grazing in A.bussei woodland F36/5 (1) Tp (Gr) (A.bss) Photograph counted under microscope. Number entered as n* vegetation/species bias factor based on photograph and computed as density (no./km 2 ). Attributes computed as % for topi. Charcoal kilns counted in stip & those outside strip ignored 3 Charcoal kilns in one in A.bussei woodland 3 ChK (1) (A.bss) Enter and compute as density (no.* /km 2 ). Attributes computed as % for ChK Aircraft reaches end of transect by map navigation End of Transect 1 in Stratum 5 on 6 February 1983. Time 08 09 1/5 5-02-83 08.28.40 Enter time (T2) and compute total duration time for use in land use estimations. Mark on map and measure transect length to be used with mean altitude based of radar altimetry to compute area of transect to be used for density estimations.
The conditions in the Horn of Africa in 2007 have changed, particularly in Somalia 1.New levels of urgency to promote social changes in the direction of more representative and accountable governance. 2. Poor levels of credible resource information which is growing increasingly out of date. 3. Information about resources is less subject to sovereign rights. 4. It is now safe to fly in the airspace at over 500m. 5. GIS and other software now allows previous and new aerial strip sampling resource surveys to be quickly turned into products useful for planning sustainable and equitably distributed developments. 6. New digital equipment and techniques, including airborne and satellite remote sensing and analytical software, can take remote strip-sampling surveys of resources past its analogue barriers.
How multi-feature aerial strip sampling for resource surveys would be conducted today. The aerial strip shown comes from a survey carried out in Laos in 2002 using a hydrogen balloon and a 4MP camera. ArcMap is used to digitize the resources on the photos A demonstration of an aerial strip sampling analysis work station has been set up and can be seen by interested parties at the end of the formal sessions.
Tin Roof Barracks Flowing River Fallow Crop Field Tamarind Fruit Tree Cattle Crush Wood Tiled House Un-rehabilitated Borrow Area Livestock Drinking Pond 4WD Vehicle Goat Compound & House Grass Roofed House Toilet Logs Irrigated Fields Fish Pond Irrigation Canal Weir Irrigated Fields Tin Roofed House Livestock Watering Ponds Steel Bridge Water Buffalo Tin Roofed House Irrigated Gardens
LogsFenceBanana Grove Grass Roofed House 14 Cattle 12 Pine Logs 3 Cattle Footpath Irrigated Fields Tin Roofed House Charcoal Pit Toilet Fish Pond Track 1 Elephant Tin Roofed House Rice Barn Toilet Fence Goat Pen + 6 Goats 14 LogsToilet Buffalo Banana Grove Fence Tin Roofed House FootpathBuffalo Pen + 1 Buffalo 6 Goats
Abandoned Truck Most Houses Looted No Roofs MosqueTruck Hotel PeopleLooted Roofless Houses 20’ and 40’ Containers used as Shops People Looted Roofless Houses Track WallFootpath
There are further possibilities for using digital – GIS methodologies to enhance a resource survey. These enable the survey to become an acceptable census. Acceptability is today a vital feature of information. It is particularly problematic in areas where governance does not carry well-tested and stable consent. Acceptability is achieved by post-survey visits to cognate sub-samples from the sampled strips. These visits can be made to house clusters or herd or water points or any other of the features already captured in the survey, so that the essential information needed to turn a survey into a census can be added to the attributes table. So ownership and other “tenure-status” can be attributed to the feature, recent history of use and movements, demographics, heath status, demographics etc. can be added to the feature data. This information can simultaneously establish the survey’s acceptability and turn its feature data into the tool necessary to make equitable and sustainable interventions.
It is now possible to carry out a Somali Resources Assessment by Aerial Sampling (SOMRAAS). Using a King Air 200 with a cruising speed of 520 km/h it would take 500 to 540 hours of flying, and could take about 100 days. The aircraft would carry a 22MP camera taking a continuous strip of images 200m to 400m wide with resolution of about 1 to 2 cm pixel size. A 39MP camera in the same aircraft could simultaneously make a Somali-wide orthophoto survey with 50cm resolution. This could be a basis for a sub metric DTM if 50 to 100 post –photography survey points were to be established. The SOMRAAS survey would generate 54 terrabytes of data, and could be stored on 55 hard drives of 500 gigabytes.
Full interpretation and quantification of the 230,000 kms of transects can be made at between 7 to 24 kms/hour, using a work station of the type shown in a demonstration which interested parties can see at the end of the session. It is estimated that a SOMRAAS would take 37,500 hours of interpreter time. This could be managed by 20 interpreters in a year. The total cost of a SOMRAAS would be about 1.1million Euro. The cost of turning the SOMRAAS Survey into National Census would be about 5 to 10 million Euro.