PEER.REVIEWED ARTICI.E
SurfaceTemperature Dynamics of LakeBaikal Observedfrom AVHRR lmages
DavidW. Bolgrien,Nick G. Granin,and Leo Levin
Abstract )o Satellite data are important in understanding the relation- Angoro R. ship between hydrodynamics and biological productivity in a Lcke Ecikol large )ake ecosystem. This was demonstrated using N2AA AvHRR,ond in silu temperature and chlorophyll fluorescence data to describe the seasonal temperature cycle and distribu- tion of algal biomass in Lake Baikal (Russia). Features such as ice cover, thermal fronts, and the dispersion of river water were describedin a seriesof imagesfrom 1990 and 1991. The northern basin retained ice cover until the end of Mav. In lune, thermal ftonts extending PE&RS PEER.REVIEWED ARIICIE Tnere 1. MonpuovrrRrc CnnRncreRrsrrcsoF SELEcTEDLARGE LAKES (FRoM ditions of the Limnological Institute (Irkutsk) and the HurcHrrlsou.1957). Institute of Biophysics (Krasnoyarsk). Since 1988, foreign in- Lake Baikal Michigan Superior vestigators have participated in expeditions under the auspi- ces of the Baikal International Center for Ecological Research Location 53'N 108"8 44'N B7'W 47'N 89'W (Maddox, 19BO).Transect data were collected using a flow- Origin Tectonic Glacial Glacial through thermistor and fluorometer interfaced to a micro- graben corrasion corrasion computer. Chlorophyll fluorescence, calibrated with Volume (km',) 23,000. 5,760 12,221 extracted chlorophyll o samples, provided an estimate of Max Depth (m) 1,638 265 307 't (2.5 Mean Depth (m) 73o 99 45 phytoplankton concentration. The position and depth m) Area (km') 31,500 57,850 82,367 of the intake pipe, flow rate, instrument response time, and Shoreline (km) 2,2oo 2,2"to 3,000 ship speed defined a minimum detectable "patch" size of Max length (km) 674 494 )oJ 200 m (Granin et al.,19BB). The temperature recorded by the Max width (km) 74 1so 257 ship was considered to be a "bulk" surface temoerature as Catchment [km,) 540,000 118,000 1,24,8OO opposed to the "skin" surface temperature meaiured by the *Total : volume of Laurentian Great Lakes 24.600 km3 AVHRR.The location of transect points were estimated by dead-reckoning from navigational fixes. Local Area Coverage (r,ac) avHnn images from the NoaR range fFigure 1). The basins are steep-sided, especially along National Environmental Satellite Data Center were selected the western shore. Lake Baikal has three large bays (Barguzin based on available in situ data, characterization of specific Bay, Chivyrkuiski Bay, and Maloe More) and three main trib- thermal features, and minimal cloudiness. Mountainous ter- utaries (Selenga River, Barguzin River, and the Upper An- rain surrounding the lake exacerbated the problem of cloud gara). Littoral zones are restricted to river deltas, shallow contamination of satellite images. Sitnikova et al. (1s84) bays, and "sores," i.e., isolated coastal wetlands. The north- found surface evaporation and foe in fall and winter to ob- ward-flowing Lower Angara River is the sole outlet (Figure scure aerial observations more thin in the spring and sum- 1.1. mer. The seasonal temperature cycle of Lake Baikal was de- Sea surface temperatures (ssr) were derived from avHRn scribed using satellite data from the Advanced Very High band 4 (1030 to 1130 nm) brightness temperatures (Kidwell, Resolution Radiometer (nvnnn) aboard the National Oceanic 1991). Ice cover was identified from AVHRRband Z (zZOto and Atmospheric Administration (Noan) polar orbiting satel- 1100 nm). Data were not corrected for atmospheric effects or Iites. To our knowledge, AVHRR data have not been previ- sun angle differences, nor were the images geo-rectified. The ously used to describe this lake. Lake Baikal is a rift lake spatial resolution of the AVHRR at nadir is approximately 1.1 located between 51'and 56'N latitude and is surrounded bv km. avsRR data (band 2 for Figure 3, and band 4 for Figures high mountains. This combination permits an ice cover of up 3 to 6 and B to 10) were classified using a sequential cluster- to one m thick to form over the entire lake (Kozhov, 1963). ing technique to create a "landmask," and enhance particu- Complete ice cover is atypical of large deep lakes because lar features in each scene. long fetches usually result in continuous wind mixing. Sitni- kova et a1. (1984) used both aircraft and satellite imagery to Resultsand Discussion detail ice break-up in 1981. They found that the majority of Satellite remote sensing research depends on the availability the lake was ice covered in earlv Aoril. Shore-bound ice in of data at spatial and temporal scales relevant to the feature pack-ice the southern and middle basins. ani broken in the investigated. Further, the feature of interest must be spec- northern basin, remained through the middle of May. Ice ac- trally evident in the image. For large lakes, the AVHRRgener- cumulated along the eastern shore because of the prevailing ally satisfies these criteria. The visible bands are useful for westerlv winds. The entire lake was ice-free at the end of detecting ice cover on lakes (Hanison and Lucas, 19Bg). The May. Aithough shore-based ice observations have been made approximate 1.1-km resolution of the thermal bands of the for most of this century, Sitnikova et ol. (1,s84) appear to be AVHRRhas proven sufficient to describe salient phvsical fea- the only reported use of remote sensing techniques on the tures of the Laurentian Great Lakes (Bolgrien and Brooks, lake. 1992). The availability of U.S. satellite imagery for the Lake Vernal thermal fronts are common features in large Baikal region is summarized in Table 2. Review of image north-temperate lakes. Fronts form in coastal regions through quality, including cloud cover, revealed that <50 percent of a combination of solar warming and spring runoff. They sep- the images listed would be useful for limnological research. arate the thermally stratified inshore region where surface The increased availability of avrnn images for Lake Baikal temperatures are )4"C from the isothermal or inversely strat- in recent years makes this sensor a very useful tool for the ified offshore region where surface temperature are <4"C. systematic monitoring of thermal features of this remote lake. Fronts move offshore until the 4'C isotherm disappears across the lake (Mortimer, L971-;Bolgrien and Brooks, 1992). Convection currents and wind-mixing are the primary mech- Tnere2. NUNIBERoF lMacEsAvnrmerE FRoM U.S. Snrellrres FoR THE LAKE anisms responsible for deep-water mixing (Granin et a1., Bntxt REcrolr(51'-56'N x 103'-110'E.DArA ARE FRoM rHE USGS Groanl 1991; Shimaraev ef al., tgg3; Weise ef al., 1s92). Surface LrruoInronvarroru svsrEu ,3ir?l;"#J[1]Dls, ANDEoSAr (ns or 15 temperature is a good method of delineating the temporal and spatial dynamics of thermal fronts (Bolgrien and Brooks, Year <1988 1988 1989 19SO 1991 7952 1993 1992). AVHRR ?03322212951 LandsatTM 020170517411 Methods LandsatMSS r25031L17200 fluorescence have been col- Surface temoerature and data CZCS 40 (no further data collected) Iected from Lake Baikal for more than L0 years by joint expe- 2L2 PEER.REVIEWED ARIICTE I Figure3. Distributionof ice on northernLake Baikal derivedfrom AVHRRband 2 from 75 May 1991. thin/melting Figure2. Photographtaken from the SpaceShuttle of the northernend of LakeBaikal in late April,1991 (photo- graphssrso39-73-88, -89, -9o). lce condi- J53oN tions interpretedby Evanset al. (L992). 100k- I 100E Figure4. ssr ('C) map of Lake Baikalderived from nvHRR band4 from 09 .June1990. The availability of Russiansatellite data for the Lake Baikal regionis not known. A schemeof the seasonaltemperature cvcle of Lake Bai- kal was createdusing imagesfrorn 1990 and 1991. Figure 2 may have originated on the shallow delta and/or been is a photographicmosaic taken from the U.S. SpaceShuttle "leaked" from Maloe More through the straits south of Olk- (srs-as)in late April 1991 (photographsSrSo3s-23-BB, -ss, -go). hon Island. The northern end of Lake Baikal was coveredby a mixture By 29 June 1990, the offshore SSTwas approximately of ice types (Evanset ol., 1,992).The sill isolating the Upper 4'C and the thermal front had progressed10 to 30 km off- Angara sore from the main Iake basin is noted. An AVHRR shorein the middle basin (Figure5). Cloudsobscured both image(band 2) acquiredon 15 May 1991showed a similar ends of the lake on this date. The observedssr was consis- situation for the entire lake (Figure 3). Broken ice filled Bar- tent with transectdata collectedon 3-5 fuly 1990 (Graninef guzin Bay and Maloe More and some shore-iceremained o.1..19911. The offshore surfacetemperatures in the middle near the SelengaRiver delta. The majority of the southern basin were approximately4"C. The warm water inputs (13'C) and middle basinswere completely free of ice. The northern of the SelengaRiver, seenalso in Figure 4, contributed to the basin was still ice covered although the ice appearedthicker warming in the middle basin. along the easternshore. Surfacewater temperatureswere (4o The AVHRRscene from S fuly 1990 (Figure 0) coincided (band 4 data not shown). The distribution of ice shown in with a field expedition to measurethermal stratification,nu- Figures 2 and 3 was consistentwith the description of Sitni- trient utilization, and phytoplankton productivity throughout kova ef al. (1.98a).Variations in the timing of ice-out in dif- the lake (Granin et ol., 1.991;Goldman et al., t993). These ferent parts of Lake Baikal drive large-scalepatterns of studies found surfacenutrient depletion and phytoplankton thermal stratificationand, therefore,the patterns of biological productivity were directly related to the degreeof thermal productivity (Goldmanef 01.,1gg3). stratification.Warm surfacetemDeratures indicative of strati- Ice was absenton I June 1990 (Figure 4) and thermal fication were most oronounced in the southern and middle fronts were observedas <10-km wide bands of 5 to 6oCwa- basins (Figure O).Data collected along a transectrunning ter bordering much of the southern and middle basins.Both normal to shore in the northern basin showed a thermal in situ and RvriRRdata showed offshoretemperatures were front was located about B km from shore where surfacetem- (4'C. Temperatures>6'C indicated the onset of stratification perature rapidly increasedfrom approximately 4'C to )B"C in Chivyrkuiski Bay and near the SelengaRiver. Water 5 to (transecta - a' in Figure 6; Figure 7a). This front was also 6'C between the Olkhon Island and the SeleneaRiver delta apparent in the AVHRRSST data. Chlorophyll fluorescence PE&RS PEER.REVIEWED ARIIGTE 10 120 108 ', 0 ir',,,, o i,i 'i-.. tluorescence 7 84 @ o 0 72 5 60 o ! 48 o 0 o. 3 36 p o .J o H 2 24 lahhar.fr!ra q 1 A T2 0 0 8 1012141618202224 ?OC km fron shore (a) 10 120 108 o 8 96 o u l::""+ 7 84 o o loson N o 72 5 60 o Figure5. ssr ('C) map of central Lake Baikalderived H 48 0 29 June1990. 0 from nvunRband 4 from 3 36 0 d 2 24 o I 12 b t" n"r^d" b' 0 0 10 20 50 60 70 80 90 100 kn 14oc (b) 10 120 9 108 o 8 o o 7 84 6 o o 1t k a 60 d H o 4 48 o 3 p o H 2 o d 1 L2 0 0 0 10 20 30 40 50 60 70 80 90 100 km (c) Figure7. (a) Surfacetemperature and chlorophyll fluorescencedata collectedalong transect line a in Figure6. Chlorophyllfluorescence values are Figure6. ssr ("C)map of LakeBaikal derived from nvHRn in relativeunits. Data are from Graninet a/. band 4 from 08 July 1990. Transectlines for Figure7 (1991).(b) as in Figure7a exceptdata werecol- parallel are labeledby linesrunning on shore. lectedalong transect line b in Figure6. (c) as in Figure7a exceptdata were collectedalong tran- sect linec in Figure6. gradually increasedfrom offshoreto nearshore.Because phy- toplankton growth in unstratified waters is constrainedby light, resulting from deep vertical mixing, rather than exter- calm and sunny period. The warm water near the north end nal nutrient supply (Goldman et o/., 1993),this increasein of the lake, especially near the Upper Angara sore, resulted chlorophyll fluorescencemay reflect decreasingmixing depth from the innut of warm river waters and limited circulation along the transect. in this relativelyshallow region. A transecttoward the north end of the lake (transectb - A transectfrom the SelengaRiver delta to Olkhon Island b' in Figure 6) showed severaltemperature and chlorophyll (transectc - c' in Figure 6; Figure 7c) showed higher temper- fluorescencefronts (Figure 7b). There was a significant de- atures and chlorophyll fluorescence in the middle basin as creasein fluorescenceas temperaturesincreased from 2 to compared to the northern basin. High phytoplankton concen- 4oCat the north end of the transect.While the AVHRRSST trations likely reflectedthe high nutrient supply of the Se- data show similar temperaturepatterns, there is a 2 to 4"C lenga River. There was a 25-km wide region near Olkhon overestimationof bulk temperatureby ssr. This may indi- Island where in situ temperaturesdecreased to about 4'C and cate daytime heating of the surfaceskin during an extended fluorescencedecreased by >50 percent. This cold-water re- 2L4 PE&RS PEER.REVIEWED ARIICTE 150C + N -.r+':!l1. -+ 53oN '* aT lB 150C 11008 -t* \,1 100km Jss"N , :r''\ ::l:' \ | 11oot -: clouds 100km Figure8. ssT ("C)map of northernLake Baikal derivedfrom nvnRR band 4 from 25 Julv1990. Figure9. SsT("C) map of northernLake Baikal de- rivedfrom AVHRR band 4 from 07 August1991. gion was not observedin the RvHRRimage (Figure 6). This differencemay be explained by the surfaceheating phenom- ena or by the five days between in situ measurementsand and 10). Thermal stratificationtypically lasts from early July (Shimaraev, the avHRRimage acquisition. The temperaturedifference be- until early November 1.977).Ina fashionanalo- tween these datesmay representan ephemeralupwelling gous to vernal warming, autumnal cooling appearedto com- (Figure event along the south shore of the island. Upwelling results mence nearshoreand progressoffshore 10). Surface from the upward displacementof the thermocline by internal waters cool until thermal stratificationbreaks down and deeo waves or the horizontal displacementof the epilimnion by vertical mixing occurs. The upper layer of the lake typically wind. Both situations result in the transient lowering of near- becomesisothermal in November and begins to freezein Jan- (Kozhov, shore surfacetemperatures. Upwelling is common in Lake uary 1963;Shimaraev, 1977). Baikal (Graninet al.,19BB)but was not well illustratedin the availableAVHRR imagery. Conclusions Thermal featuresseen on 25 Iulv L990 and 7 Ausust The seasonalthermal cycle of Lake Baikal was constructed 1991 were consistentwith the schemeconstructed frim the using NOAAAVHRR images. Satellite-derived surface tempera- earlier AVHRRscenes (Figures B and 9). The southern basin tures clearly depicted important thermal featuressuch as ice was obscuredby clouds in both scenes.Typical vertical tem- break-up,thermal fronts, and autumnal cooling. Ship perature profiles in late July and August showed thermocline transectdata describedsurface thermal featuressimilar to depths ranged from <1 to 15 m. A chlorophyll fluorescence those describedby satellite data but at a much finer spatial maximum was often located at or slightly abovethe thermo- scale. Surfacechlorophyll fluorescencewas found to be cline (summer vertical profile data not shown). Surfacetem- higher nearshore,presumably becauseof nutrients supplied peratureheterogeneities ("patches") observedin Lake Baikal by river runoff and restricted vertical mixing. by Gitel'zonet d1.(1991) may be discernibleusing the Hydrodynamic conditions (light availability, mixing, AVHRR.Typical patchesin the relatively productive southern temperature)of large lakes, such as Lake Baikal, dominate bi- and middle basinsof Lake Baikal are 0.7 to 1.5 km in size, ologicalproductivity, especially during the spring.These while they are 3 to 7 km in the less productive northern ba- conditions can be deduced through the systematicanalysis of sin. The lifetime of a typical patch i; 2 to 16 days. Signifi- satellite SSTdata. Satellite remote sensing,therefore, plays an cantly, Gitel'zon et o1,(1991) found no systematic important role in the study of these expansiveand remote relationship between surfacetemperature and chlorophyll ecosystems. fluorescencewhen measuredsimultaneously. There is a sev- eral day lag between a cold water (hencein-creased nutrient Acknowledgments concentration)intrusion and the chlorophvll fluorescencein- The authors wish to thank the Baikal Center for Ecological creaseresulting from phytoplankto.t g.o*ih. Factorssuch as Research;the University of Wisconsin-MadisonEnvironmen- adaptationby phytoplankton to low light may also be impor- tal RemoteSensing Center; Arthur Brooks,Richard Back, and tant. Becausethe heterogeneousdistributions of energy and Kenneth Nealson (University of Wisconsin-MilwaukeeCenter material are important to the functioning of the entire eco- for GreatLakes Studies); and Will Gould (NoAANESDIS). The system,the capacity to monitor these processesutilizing sat- SpaceShuttle photography was kindly provided by Cynthia ellite sensorsis advantageous. Evans(Lockheed Engineering & SciencesCo., Houston, The short Siberian summer was evident in the decrease Texas).Satellite imageswere purchasedwith a scholarship of surfacetemperature from August to September(Figures 9 awardedto D.W. Bolgrienby the InternationalAssociation of PE&RS PEER.REVIEWED ARIICTE Levin, L.A. Levin, V.V. Tsekhanovsky, P.P. Sherstyankin, and M.N. Shimaraev, 1991. Distribution of Lake Baikal pelagic eco- system characteristics in spring convection, Preprint # 158b. In- Sciences (Siberian clouds stitute of Biophysics, USSR Academy of Division), Krasnoyarsk, Russia, 54 p. Harrison, A.R., and R.M. Lucas, 198s. Multichannel classification of snow using NOAA AVHRR imagery, International lournal of Re- mote Sensing,10:907-916. Hutchinson, G.E., 1957. A Treatise on Limnology, Vol. r, f. Wiley t and Sons, New York, 1015 p. N Kidwell, K.B. (editor), 1991. NOAA Polar Orbiter Dato Users Guide, National Environmental Satellite Data and Inforrnation Service, rook- National Climatic Data Center, Satellite Data Services Division, ;f":". Washington, D.C., 280 p. Kozhov, M., 1963. Lake Baikal and its Life, Dr. W. funk Publishers, The Hague, 321 p. Figure10. ssr (oC)map of LakeBaikal derived from Kozhova, O.M., V.N. Pautova, L.R. Izmest'yeva, and LK. Davydova, Lake Baikal, AVHRRband 4 from 04 SeotemberL991. 1985. Chlorophyll a of Hydrobiological lournal,2L: 72-1.9. Lathrop, R.C., I.R. Vande Castle, and T.M. Lillesand, tgg0. Monitoring river plume transport and mesoscale circulation in Great LakesResearch. This work was supported by the Green Bay, Lake Michigan through satellite remote sensing, / Great Lakes Research, TGi471-484. Global ChangeFellowship Programof the National Aeronau- tics and SpaceAdministration. This paper is contribution Maddox, I., 1989. Baikal Center takes step forward, Nofure, 341:754. #364 of the Center for Great Lakes Studies,University of Mortimer, C.H., 1971. Large-Scale Oscillatory Motions and Seasonal Wisconsin - Milwaukee. Temperature Changes in Lake Michigan and Lake Ontario, Spe- cial Report #12 of the University of Wisconsin-Milwaukee, Cen- ter for Great Lakes Studies, 106 p. Refelences 1988. Discoveries and testable hypotheses arising from Coastal Zone Scanner imagery of southern Lake Michigan, Bolgrien, D.W., and A.S. Brooks, 1992. Analysis of thermal features Limnol Oceanogr., 33:2O3-226. of Lake Michigan from AVHRR satellite images, I Great Lakes Research, L8t259-266. Sitnikova. G.V.. M.S. Furman. and N.N. Yanter, 1984. Remote and ground-based methods for researching the dynamics of ice Evans, C.A., M.R. Helfert, and D.R. Helms, 1992. Ice patterns and processes on Lake Baikal, Femofe and Ground-Based Studies of hydro{hermal plumes, Lake Baikal, Russia: lnsights ftom Space the Dynamics of Naturul Processes in Siberia,Institute of Geog- Shuttle hand-held photography, Proceedings of International raphy, USSR Academy of Sciences (Siberian Division), Irkutsk, Geosciences and Remote Sensing Symposium, IEEE, Houston, pp. 72-81.. Texas. Gitel'zon, I.I., N.G. Granin, L.A. Levin, and V.V. Zavoruev, 1991. Shimaraev, M.N., 1977. Elements of the Heat Regime of Lake Baikal, Mechanisms underiying the formation and maintenance of the Nauka, Novosibirsk, r+9 p. spatial distribution nonuniformities of the phytoplankton in Shimaraev, M.N., N.G. Granin, and A.A. Zhdanov, 1993. Deep venti- -32o. Lake Baikal, Hydrobiological lournal, 318:317 Iation of Lake Baikal waters due to spring thermal bars, Litnnol Goldman, C.R., I.]. Elser, R.C. Richards, f.E. Reuter, and A.L. Levin, Oceanogr.,38:1068-1071. 1993. Thermal stratification, nutrient utilization, and phyto- Weiss, R.F., E.C. Carmack, and V.M. Koropalov, 1991. Deep-water re- plankton productivity during the onset of spring phytoplankton newal and biological production in Lake Baikal, Nofure, 349: growth in Lake Baikal, Russia, Limnol. Oceanogr, in press. 665-669. Granin, N.G., L.A. Levin, and V.V. Zavoruyev, 1988. Spatial hetero- geneity of temperature and phytoplankton in the surface layer of Lake Baikal, Preprint # 85b, Institute of Biophysics, USSR Acad- emy of Sciences (Siberian Division), Krasnoyarsk, Russia, 36 p. (Received 11 February 1993; accepted 28 April 1993; revised 16 fune Granin. N.G.. A.A. Zhdanov, V.V. Zavoruev, G.G. Zinenko, A.L. 1993) To receiveyour free copy of the ASPRSPublications Catalog, write to: ASPRS Attn: JulieHill 5410Grosvenor Lane, Suite 210 Bethesda.MD 20814-2160 Pleaseindicate your areaof interest:Photogrammetry, Remote Sensing, GIS, Cartography,all of the above. 2L6 PE&RS