Investigations of Nyamuragira and Nyiragongo volcanoes (Democratic Republic of the Congo) using InSAR Sarah Colclough (1) (1) Cambridge University, Earth Sciences, Cambridge, CB2 3EQ, United Kingdom ABSTRACT Nyamuragira and Nyiragongo are neighbouring volcanoes in the DRC (Democratic Republic of the Congo), and are amongst the most active and dangerous in the world. Eleven ERS SAR scenes were obtained, spanning the period between Jun 1997 - Jan 2003, and were combined to form interferograms and associated coherence images. During this period, Nyamuragira erupted four times (in Oct 1998, Jan 2000, Feb 2001 and Jul 2002), and Nyiragongo erupted in Jan 2002. The coherence images enabled mapping of lava flows, and based on assumed flow thicknesses, minimum erupted volumes were estimated to be 22 × 106 m3, 71 × 106 m3 and 133 × 106 m3, for the 2002 Nyiragongo, and 1998 and 2001 Nyamuragira eruptions, respectively. SRTM data were used to remove the topographic signal from the interferograms, revealing previously undetected volcanic deformation signals over both long (years) and short (weeks) time-scales. Persistent subsidence in Nyamuragira’s NE flow field was attributed to flow cooling and substrate relaxation, and localised deformation within Nyamuragira’s summit caldera and NE flow field, were interpreted to result from shallow magma transport. Inflation on the NW flanks of Nyamuragira was interpreted as a response to magma accumulation prior to the 2002 eruption. This study shows that InSAR can provide insights into the behaviour of these otherwise little studied volcanoes, and can usefully support hazard assessment. The study also emphasises the value of InSAR, even when applied to densely vegetated volcanoes in the humid tropics. 1. INTRODUCTION 1.1. Geographic setting Nyamuragira and Nyiragongo are situated within the Western branch of the East African Rift, and are the western-most, and only recently active volcanoes of the Virunga Mountains. Nyamuragira is a shield volcano and Nyiragongo is a strato-volcano (Fig. 1). Historical eruptions have mostly originated from the numerous flank fissures and cinder cones, and less frequently from the summit calderas. Activity is characterised by effusion of basic magmas, and lava lake activity has also occurred at both volcanoes. The lavas are of low viscosity, resulting in flows that extend for tens of kms from the summits. The city of Goma lies just 18 km south of Nyiragongo, and other smaller towns, villages and refugee camps also lie within close proximity, meaning that the hazards posed by these volcanoes threaten a large population [1]. Nyamuragira Nyiragongo N Lake Kivu Goma Fig. 1. August 1987 Landsat scene (RGB: bands 5,4,1), draped over SRTM. Vertical exaggeration x4. 1.2. Data selection and interferometric processing ERS track 228, frame 7155, provided the most complete coverage of the study area, and eleven raw SAR scenes were obtained (Table 1). Scenes were acquired during ascending night time passes, minimising atmospheric contamination. Perpendicular and temporal baselines and DCF variability of potential InSAR pairs were considered to minimise decorrelation, and time spans of potential pairs relative to eruptions were taken into account. Initial results revealed excellent coherence over barren lava surfaces, and limited coherence over vegetated areas and steep upper slopes of the volcanoes. For each interferogram, the overall coherence quality was a trade-off between the temporal and perpendicular baselines, and DCF variability (Table 1). Twenty-five InSAR pairs possessed sufficient coherence, and were processed using GAMMA software. The interferograms were multi-looked at a 2:10 ratio, and the GZW branch cut unwrapping algorithm with a coherence threshold of 0.3 was used. Dense vegetation resulted in isolated patches of coherence, which despite modifications to the processing technique [2] could not be bridged together during unwrapping, without generating artificial phase ramps. Hence, the isolated areas were unwrapped individually. The two-pass technique was employed to obtain the differential interferograms, whereby SRTM data were used to remove the topographic signal. Table 1. Interferograms processed and their properties. (Mean coherence values were extracted from the same location in each image). Perpendicular Temporal Absolute DCF Mean Interferogram pair Baseline (m) Baseline (days) Difference (Hz) Coherence 04Jun’97-22Oct’97 23 140 0.0 0.56 04Jun’97-10Nov’98 102 524 153.6 0.55 04Jun’97-11Nov’98 233 525 0.0 0.53 04Jun’97-06Sep’00 98 1190 355.3 0.53 22Oct’97-10Nov’98 79 384 153.6 0.50 22Oct’97-11Nov’98 256 385 0.0 0.45 22Oct’97-29Jan’03 154 1925 1690.4 0.41 10Nov’98-11Nov’98 335 1 153.6 0.68 10Nov’98-06Sep’00 200 666 201.7 0.54 11Nov’98-19Apr’00 347 525 88.3 0.43 11Nov’98-06Sep’00 135 665 355.3 0.53 11Nov’98-20Dec’00 409 770 175.9 0.46 19Apr’00-06Sep’00 482 140 267.0 0.28 19Apr’00-20Dec’00 62 245 87.6 0.54 19Apr’00-29May’02 52 770 902.4 0.29 06Sep’00-20Dec’00 544 105 179.4 0.28 06Sep’00-13Feb’02 108 525 821.7 0.29 06Sep’00-03Jul’02 46 665 298.9 0.44 20Dec’00-13Feb’02 436 420 1001.1 0.32 20Dec’00-29May’02 10 525 814.8 0.30 13Feb’02-03Jul’02 62 140 522.8 0.47 13Feb’02-29Jan’03 75 350 513.3 0.24 29May’02-03Jul’02 488 35 336.5 0.43 29May’02-29Jan’03 501 245 699.7 0.18 03Jul’02-29Jan’03 13 210 1036.1 0.38 2. METHODOLOGY 2.1. Lava flow mapping Lava flows erupted onto previously vegetated land were revealed as newly coherent areas. However, for newly erupted flows emplaced upon older flows, the contrast in coherence was less marked, and it was not always possible to distinguish their extent/edges. Coherence images were resampled to a UTM projection, and flow areas calculated by addition of the pixel areas comprising the new flows. Mean thicknesses of 1.0, 1.5 and 3.0 m were assumed for Nyiragongo pāhoehoe and ‘a‘a flows [3a], and Nyamuragira flows [4 and 5], respectively. 2.2. Deformation mapping It was assumed that the only contributions to the differential phase signals were true ground displacements, so that each displacement fringe represents 28.3 mm of relative range change in the satellite LOS direction. The 23º ERS nominal look angle means that interferograms are more sensitive to vertical than horizontal deformation, and it was assumed that all movement was vertical. Deformation signals were analysed quantitatively. 3. RESULTS 3.1. Lava flow mapping Three cases of new lava flow emplacements were revealed. Flows to the S of Nyiragongo first appear in InSAR pairs with the earliest scene acquired on or after Feb 2002. The only extra-caldera Nyiragongo eruption between 1997-2003 was in Jan 2002, and emplaced flows on the S flanks. Flows NW of Nyamuragira show coherence for pairs with the earliest scene acquired on or after Nov 1998, and flows to the N and SE of Nyamuragira first appear in pairs with the earliest scene acquired on or after Feb 2002. By similar reasoning to the above, these flows were attributed to the Oct 1998 and Feb 2001 Nyamuragira eruptions, respectively. Associated flow areas that were not evident in the earliest InSAR pair following an eruption, but became coherent in later pairs, were mapped as ‘a‘a flows, where the delay in coherence was attributed to postemplacement rotations and compactions [e.g. 6]. Fig. 2 illustrates the mapped areas, and Table 2 lists the areas and volumes. Flows to the SE of Nyamuragira reveal coherence in pairs for which the earliest and latest scenes were acquired after the 2000 and before the 2001 Nyamuragira eruptions, respectively. These flows can be attributed to the 2000 Nyamuragira eruption, but poor coherence quality precluded mapping. 2a 2b 2002 Nyiragongo pāhoehoe flows 2002 Nyiragongo ‘a‘a flows 2001 SE Nyamuragira flow 2001 N Nyamuragira flow 1998 NW Nyamuragira flow 2c Coherence image RGB combinations: Fig. 2a. R:13Feb’02-3Jul’02, G:29May ’02-3Jul’02, B:29May’02-3Jul’02 Fig. 2b. R:19Apr’00-20Dec’00, G:13Feb ’02-3Jul’02, B:13Feb’02-3Jul’02 Fig. 2c. R:4Jun’97-22Oct’97, G:13Feb ’02-3Jul’02, B:13Feb’02-3Jul’02 Fig. 2. Nyiragongo and Nyamuragira mapped lava flow areas superimposed on coherence images. Fig. 2a: 2002 Nyiragongo lava flows. Fig. 2b: 2001 Nyamuragira N and SE flows. Fig. 2c: 1998 Nyamuragira NW flows. Table 2. Mapped lava flow areas and calculated erupted volumes (to 1 and 0 d.p., respectively). Surface area (m2) Erupted volume (m3) 2002 pāhoehoe flows 8.6 × 106 9 × 106 NYIRAGONGO 2002 ‘a‘a flows 9.1 × 106 14 × 106 2002 Total 17.7 × 106 22 × 106 1998 NW flow 23.7 × 106 71 × 106 2001 N flow 15.6 × 106 47 × 106 NYAMURAGIRA 2001 SE flow 28.7 × 106 86 × 106 2001 Total 44.2 × 106 133 × 106 3.2. Deformation mapping Significant areas of deformation included: a large area of persistent subsidence within the NE Nyamuragira flow field, inflation and deflation signals within Nyamuragira’s caldera and inflation to the NW of Nyamuragira’s caldera (Fig. 3). Surface deformation rates were calculated for transects taken along the NE flow field (Fig. 3a) and the summit caldera (Fig. 3b). Two examples of the resultant deformation rate profiles are depicted in Fig. 4. 3a 3b 2.83 cm Fig. 3a. NE Nyamuragira flow field A1 transects, superimposed on the 22Oct’97-10Nov’98 differential inter- A2 ferogram. UTM projection. 0 B2 Range Change B1 Fig.