1、 CCOP/SOPAC Preliminary Report 16 April 1989 GROUND TRUTH SURVEY TO CALIBRATE SPOT SATELLITE IMAGE OF THE BA RIVER DELTA AREA By Pascal COLLOTTE Techsec Prepared for: Committee for Co-ordination of Joint Prospecting for Mineral Resources in the South Pacific Offshore Areas (CCOP/SOPAC) Fiji Project:
2、 FJ.19. Contributed by: Mineral Resources Department of Fiji TABLE OF CONTENTS SUMMARY. 3 INTRODUCTION .4 OBJECTIVES . . 4 METHODS . 4 - Land, forest . 4 - Wetland vegetation and sedimentation . 4 - Deltaic tidal flats and lagoon reef patches 4 - The Ba River mouth . 4 RESULTS . .5 WETLAND . .6 LAGO
3、ON . . .7 RIVER MOUTH . .7 CONCLUSIONS . .8 REFERENCES 9 ANNEXES .10 ILLUSTRATION FIGURE 1: Bare soils and sugar cane fields FIGURE 2: Pine tree forest FIGURE 3: Rhizofora mucronata and mangle FIGURE 4: Rhizofora stylosa FI GURE 5: Carbonate sand beach under the fossil reef formation FIGURE 6: Rhizo
4、fora stylosa growing on the carbonate sand of Nukulivono Island FIGURE 7: Relic fringing reef edge inside the present lagoon FIGURE 8: Map showing current and flood plain patterns at the mouth of Ba River. 3 SUMMARY A ground truth survey has been carried out to calibrate the satellite thematic map r
5、esulting from the processing undertaken on a selected area of the SPOT image of NW Viti Levu. This area covers the Ba River Delta and its adjacent lagoon and land areas. The survey has been carried out under the Fiji SPOT Pilot Project, as a new task of the project FJ 19: “Baseline studies of inshor
6、e and nearshore areas in Fiji for coastal development programmes“. As a result, it was found that a very good correlation exists between the image interpretation and the field reality. Some additional sub-facies were differentiated during the survey. Although some of them did not appear from process
7、ing used for the map, they mostly existing image data. More processing has to be made to obtain this information from the satellite data. 4INTRODUCTION The ground truth survey in the Ba Area was conducted from 10 to 15 April 1989 as a request of the Mineral Resources Department of Fiji (MRD). The gr
8、ound truth survey was carried out using the SPOT thematic map, as the image has been already pre-processed at Papeete, using IFREMER processing facilities (Collotte 1989). Therefore we had the opportunity to check the accuracy of the interpretation made as a thematic map from the image processing. M
9、y participation in the survey was requested by Mineral Resources Department as the SPOT image processor and interpreter. Four persons made up the survey team: Rajendra Singh, Tevita Vuibau, Eroni Tupua, all of Mineral Resources Department and Pascal Collotte, (CCOP/SOPAC). OBJECTIVES This survey has
10、 been carried out under the Fiji SPOT Pilot Project. as a new task of the project FJ 19 : “Baseline studies of inshore and nearshore areas in Fiji for coastal development programmes“. The objective was to calibrate the SPOT satellite information with field observations, with regard to the analysis m
11、ade from image processing. METHODS During the survey, we calibrated the thematic map (see annex 1), by comparing map facies with their reality on the field. We took photographs of each facies and sub-facies. Panoramic photos were also taken from high viewpoints to cover the area of interest. The are
12、a of survey was divided in four sub-areas: - land and forest, - wetland vegetation and sedimentation, - deltaic tidal flats and lagoon reef patches, - the Ba River mouth. 5RESULTS LAND The eastern-most area appearing as LAND and FOREST as a bushy vegetation on the SPOT map is actually covered by sma
13、ll-size sugar cane fields at different stages of growth. Meeting with farmers in this area, it was found that at the time the image was acquired (April 1988), the local area was facing a severe drought. As a result, many fields that are presently planted with cane crops were barren. Also, the cane c
14、rops in April 1988, were under stress and therefore growth was slow and retarded. The crops under stress generally appeared yellowish compared to the dark green crops not under stress (see Figures 1 and 2). This explains the differentiation made by the satellite. The cane fields on either side of th
15、e Ba River are well delimited on the processed image, because of the abundance of moisture due to the proximity of the river. After checking with farmers living close to the river, it was found that their cane production was reduced by only 10% last year despite the drought. In this area the blue pa
16、tches mapped as bare soils from SPOT were effectively either hare land or uncultivated fields (see Fig. 1). The red facies defined as forest from image processing are effectively forest, hut can be sub-divided in three forest subfacies. The first was differentiated on the thematic map, but not disti
17、nguished in the legend. 1. The very dark red covering much of the hills represents pine tree forest (coniferous, see Fig. 2). The two other types of red sub-facies, not differentiated on the map, are related to deciduous tree forest. 2. The bright red, found in the river meander, represents a tall d
18、eciduous type of tree versatile to any soil moisture (Albizia procera and lebbeck). 3. The same color facies also represents three other types of deciduous trees: - Serianthes melanesica, - Samanea saman, - Casuarina equisetifolia. 6 WETLAND The land/mangrove contact, which appears sharp on the SPOT
19、 processed image, is sharp and shows the same sequence at the five spots checked. following this order: mangrove trees, bare mudyflat and sugar cane fields. This sequence extends over 10 m only, so that the mud-flat does not appear on SPOT imagery, because the SPOT multispectral resolution is 20 x 2
20、0 m. The two types of mangrove defined by processing as shrubby and bushy mangrove are respectively (see Fig. 3 and 4): - Rhizofora mucronata (tall trees with dark small green leaves), - Rhizofora mangle (smaller trees with light green leaves). The R. Mangle tree presents networked branching roots w
21、hich plunge into the water. The R. Mucronata is tall; its roots run underground with aerial tips to absorb oxygen. Rhizofora mangle grows in areas less swampy than the Rhizofora mucronata. In that respect, these two types of vegetation are directly related to the type of soil they grow on. The Rhizo
22、fora mangle grows on sandy soil, and require less moisture than the Rhizofora mucronata. A narrow ring of Rhizofora mucronata surround the 100 m and 40 m high hills, bordering the eastern creek of the delta (see annex). This is shown on the SPOT map as well. Panoramic photos of Vatutavui area were t
23、aken when crossing the lagoon. The sandy beach of Vatutavui village looks like an exception in the muddy deltaic environment. Evidently, it is a fossil formation buried under a reef formation, which may be connected with the dead reef edge appearing in the lagoon. Evidence of this fringing reef are
24、found in the area (see Fig. 5). Wave action has eroded the reef formation, and the relic carbonate sand formation is now reworked as a beach deposit. This is of much interest with regard to the tectonic uplifting theory of Viti Levu. The village of Vatutavui has been settled since 600 BC and the old
25、 cemetery emplacement had since long been eroded, although cemeteries are generally built up on high areas safe from coastal erosion. In addition, as noticed from the Latui/83 cruise seismic records, it appears that at least 2 river channel networks are superimposed in the eastern side of the delta,
26、 up to a sedimentary thickness of 45 to 60 m (Collotte, 1988). Those two events are evidence of subsidence of this area, or fast sea level rise. The sea level curve made from samples from the south coast of Viti Levu shows a sea level rise of 12 m over the last 8,000 years (Maeda et al., 1986). 7 Th
27、e lagoon is now reworking relic sand which used to be buried under the reef moat. The closes t hill from the Raviravi prawn farm gives us a good viewpoint of the western area. We checked the trueness of small islands right inside the mangrove area. The access is quite difficult, but the islands are
28、real and their differentiation, from SPOT, from the mangrove is explained by the difference of soil and vegetation. LAGOON While surveying Nukulivovo Island, we noticed a good correlation of mangrove vegetation and coral sand cay with the thematic mapping information. Even the small tongues of sand
29、are mapped. The mangrove type is of Rhizofora stylosa, well developed on carbonate sand soil (see Fig. 6). At low tide, photos were taken of the dead fringing reef edge (see Fig. 7) which appears on the SPOT map only sporadically, and of the Tavuta reef contour which appears very clearly on the map.
30、 West to Vatutavui village (see annex) is the oldest river channel system. The tidal flats are sandy-mud flats lying at more than 1 m deep at high tide. From the hill on the east of the prawn farm (viewpoint), the reef patch mapped at the immediate west of the farm appears at low tide. It is brownis
31、h and evidently dead because covered by mud. By talking with fishermen, it appears that the main flood current comes from the west passage through the discontinuous barrier reef. It could be responsible for the east curving motion of the submerged river channels (see Fig. 7). A weaker flow comes in
32、the lagoon through the Tavuta reef passage (north). The ebb currents flow goes west alongside the delta front, responsible for the loop motion of the turbid plume when meeting the flood current. It is also responsible for the plume coming from the river mouth and moving west along the shore. However
33、, the main river flow direction is toward the Tavuta reef passage (see Fig 7). RIVER MOUTH The eastern round island close to the river mouth appears, at low tide, largely surrounded by mud-flats. The western island and its adjacent tidal flats are smaller. On the SPOT thematic map these islands and
34、the submerged river channels appear in complete correlation with the field observation (see Annex). 8 Along the seaward part of the river, the mangrove is of Rhizofora mucronata and stylosa types, which correspond exactly to the SPOT information. CONCLUSIONS Three sub-facies have been differentiated
35、 in the forest facies of which two could have been defined from SPOT. This differentiation is not clear on the SPOT thematic map, more processing has to be made to obtain this information from the satellite data. The stages of growth and the healthiness of plants can be recognised from satellite ima
36、gery. All types of facies differentiated from image processing are real. The mangrove facies can be used to identify difference in the types of 80il they are growing on. However, two sub-facies do not appear on the thematic map. Small islands (500 m diameter) are perfectly differentiated by their so
37、il and vegetation difference. Long sand tongues 2 m wide, are mapped by SPOT multispectral image (20 m resolution). The facies defined, by processing the image, on the thematic map show very good correlation with the field reality. RECOMMENDATIONS 1. More processing should be made to improve the dif
38、ferentiation of vegetal sub-facies. 2. Aerial reconnaissance survey at low tide would be useful to refine further details in lagoon and mangrove areas. 9 REFERENCES COLLOTTE Pascal, “Summary report on the Ba River Delta data review“ . CCOP/SOPAC Preliminary Report No.9, 1988, 16 pages, 7 figures. CO
39、LLOTTE Pascal, “Trip report on remote sensing activities at the Station Polynesienne de Teledection (SPT), Papeete, Tahiti - 10 January / 15 February 1989“. CCOP/SOPAC report, 11 pages, 1 Table, 1989. MAEDA Y., MIYATA T., RODDA P., SUGIMURA A., MATSUMOTO E. and MATSUSHIMA Y., “Holocene Sea-level Changes in Viti Levu and Vanua Levu, Fiji“. In Sea-level Changes and Tectonics in the middle Pacific, Report of the HIPAC Project - 1984 and 1985 - 126 pages, 1986. 10 ANNEXE
