FluxLetter

The Newsletter of FLUXNET Vol. 2 No. 3, October, 2009

Highlight FLUXNET sites Highlighting Sites in tropical SE Asian Tropical Network by Takashi Hirano and Ryuichi Hirata

In Southeast (SE) Asia, tropical strong dry season exists causing In This Issue: occupy a terrestrial area defoliation in the dry season. of about 2 million km2, which The tropical rain forest is mainly accounts for 10% of the total distributed in Malaysia and Indo- FLUXNET site: SE Asian Tropical Forest Net- tropical forest area in the . nesia, where seasonal variations work The monsoon governs the cli- in precipitation are relatively Hirano T and Hirata R……...Pages 1-4 mate of SE Asia and makes large small. In lowlands of Sumatra, Opinion: geographical differences in pre- Borneo and the Malay Peninsula, Flux towers in the Amazon for- swamp forests extensively coex- cipitation, such as seasonal varia- ests tion and the annual sum. Ac- ist with tropical peat. However, Kruijt B………………………Page 5-6 cording to precipitation regimes, the tropical forests have been Data Resources: rapidly deforested at a rate of two types of forests are distrib- MODIS Data Complements Flux -1 uted in SE Asia: tropical seasonal more than 1% yr in SE Asia Tower Information forest and tropical rainforest. (FAO FRA 2005-global tables, Cook RB, Santhana Vannan SK, and http://fao.org/forestry/32033/en/) T Beaty...... Page 7-8 The former dominates in the Indochina Peninsula, including for logging and land-use change FLUXNET Young Scientists Thailand and Cambodia, where a into farmland. In addition, El Network Vargas R, Wolf S, and Zeeman M……………………………...Page 9

FLUXNET young scientist: Natsuko Yoshifuji ……..….Pages 10-11 Rodrigo da Silva ...... Pages 12-13

Research:

CO2 fluxes of tropical eco- Figure 2: The tower in KWM. systems with different land- use in Panama Niño events change precipitation Wolf S, Eugster W, and Buchmann N regimes and bring drought espe- ……………….……....….Pages 14-16

cially to the southern part of SE Between deforestation and Asia at intervals of several years, climate impact: the Bariri and consequently enlarge field Flux tower site in the pri- mary montane rainforest of fires, which are usually ignited at Central Sulawesi, Indonesia first for farmland management. Panferov O, Ibrom I, Kreilein H, Oltchev As a result, a huge amount of A., Rauf A., June, T, Gravenhorst G and A. Knohl...... Pages 17-19 smoke is emitted from large- scale biomass burning and shades Looking back and looking the ground as haze. Sumatra and forward at flux studies in forests Borneo become hot islands in El David R. Fitzjarrald...... Pages 20-22 Niño years, because flammable peat is widely distributed there. Figure 1: Map of sites (1. KWM, 2. SKR, 3. MKL, 4. MMP, 5. Kampong Chhnang, 6. Kampong Thom, 7. PSO, 8. Lambir, 9. Undrained swamp forest in Palangkaraya, 10. PDF, 11. drained cutover in Palangkaraya) Page 2

SE Asian Tropical Forest Network

Figure 3: The tower in SKR. Figure 4: The tower in MKL.

Haze diffuses leeward to the to advance such studies, we Moh (MMP, 18º25´N, 99º43´E, Palangkaraya, Central Kaliman- Malay Peninsula with the trade established a tower network of Figure 5), which is deciduous. tan (southern Borneo) within and can cover a wide area flux measurement in SE Asia Two towers were recently built 15 km, whereas their distur- of SE Asia. under the support of Grant-in- in tropical seasonal forests in bance levels are different. They Canopy-scale gas exchanges of Aid for Scientific Research (No. Cambodia, which are a decidu- are an undrained swamp forest SE Asian tropical forests have 21255001) from JSPS. The net- ous forest (Kampong Chhnang) (02º19´S, 113º54´E, Figure 8), a been studied since International work consists of 11 flux towers (11º59´N, 104º44´E) and an drained swamp forest (PDF, Biological Program (IBP) in the of JapanFlux, which were built in evergreen forest (Kampong 02º21´S, 114º02´E) and a 1970s mainly by Japanese re- tropical forests by Japanese re- Thom) (12º44´N, 105º28´E). drained cutover (02º20´S, searchers. However, the geo- searchers (Figure 1). Four tower There are two towers in Malay- 114º02´E). An abbreviation with graphical distribution of the gas sites are located in tropical sea- sian rainforests; one is Pasoh three letters in each parenthe- exchanges and their controls are sonal forests in Thailand, includ- (PSO, 2º58´N, 102º18´E, Figure sis is a site code registered into not yet well described, because ing two evergreen forests: Kog- 6) in the Malay Peninsula and AsiaFlux. More information is the studies were so far site- Ma (KMW, 18º48´N, 98º54´E, another is Lambir (4º12´N, available for such sites through specific. Also, the effects of dis- Figure 2) and Sakaerat (SKR, 114º02´E, Figure 7) in northern the AsiaFlux web site (http:// turbances, such as El Niño 14º30´N, 101º55´E, Figure 3), a Borneo. Three sites join the asiaflux.yonsei.kr/

drought and haze, on CO2 bal- deciduous forest: Mae Klong network from Indonesian sf_southeast.html). Annual ances and evapotranspiration are (MKL, 14º35´N, 98º51´E, Figure swamp forests. All the sites are mean air temperature and an- still under investigation. In order 4), and a teak plantation: Mae located in tropical peatlands in nual precipitation range from 20 Page 3

SE Asian Tropical Forest Network

Figure 7: The tower in Lambir.

to 27ºC and from 1400 to 2800 mm yr-1, respectively, within the sites, thus the sites can cover typical climate divisions in SE Asia.

Research team: Takashi Hirano and Ryuichi Hirata (Hokkaido University) Hiroaki Kondo and Takahisa Figure 5: The tower in MMP. Maeda (National Institute of Advanced Industrial Science and Technology, AIST) Yoshiko Kosugi (Kyoto Univer- sity) Koji Tamai and Satoru Takana- shi (Forestry and Forest Prod- ucts Research Institute, FFPRI) Nobuaki Tanaka (The Univer- sity of Tokyo) Tomo’omi Kumagai and Na- tsuko Yoshifuji (Kyushu Univer- sity) Nobuko Saigusa (National Insti- tute for , NIES)

Figure 6: The tower in PSO. contact: Takashi Hirano [email protected]

Page 4

SE Asian Tropical Forest Network

Figure 8: The tower in an undrained swamp forest in Palangkaraya.

Further Reading Kumagai, T., Ichie, T., Yoshimura, M., CO2 flux in boreal, temperate, and Hirano, T., Segah, H., Harada, T., Yamashita, M., Kenzo, T., Saitoh, T. M., tropical forests in East Asia. Agric. Limin, S., June, T., Hirata, R., Osaki, M., Ohashi, M., Suzuki, M., Koike, T., Koma- Forest Meteorol. 148, 700-713. 2007. balance of a tsu, H., 2006. Modeling CO2 exchange Takanashi, S., Kosugi, Y., Tani, M., tropical peat swamp forest in Kaliman- over a Bornean tropical rain forest Matsuo, N., Mitani, T., 2005. Abdul tan, Indonesia. Global Change Biol. 13, using measured vertical and horizontal Rahim Nik, Characteristics of the gas 412-425. variations in leaf-level physiological exchange of a tropical rain forest in Hirata, R., Saigusa, N., Yamamoto, S., parameters and leaf area densities. J. Peninsular Malaysia. Phyton. 45, 61-66. Ohtani, Y., Idea, R., Asanuma, J., Geophys. Res. 111(D10107), Yoshifuji, N., Kumagai, T., Tanaka, K., Gamo, M., Hirano, T., Kondo, H., doi:10.1029/2005JD006676. Tanaka, N., Komatsu, H., Suzuki, M., Kosugi, Y., Nakai, Y., Takagi, K., Tani, Kume, T., Takizawa, H., Yoshifuji, N., Tantasirin, C., 2006. Inter-annual varia- M., Wang, H., 2008. Spatial distribution Tanaka, K., Tantasirin, C., Tanaka, N., tion in growing season length of a of carbon balance in forest Suzuki, M., 2007. Impact of drought tropical seasonal forest in northern across East Asia. Agric. Forest Meteo- on sap flow and water status of ever- Thailand. Forest Ecol. Manage. 229, 333 rol. 148, 761-775. green trees in a tropical monsoon -359. Kosugi, Y., Takanashi, S., Ohkubo, S., forest in northern Thailand. Forest Matsuo, N., Tani, M., Mitani, T., Ecol. Manage. 238: 220-230. Tsutsumi, D, Abdul Rahim, N., 2008. Saigusa, N., Yamamoto, S., Hirata, R., Ohtani, Y., Ide, R., Asanuma, J., Gamo, CO2 exchange of a tropical rainforest at Pasoh in Peninsular Malaysia. Agric. M., Hirano, T., Kondo, H., Kosugi, Y., Forest Meteorol. 148, 439-452. Li, S., Nakai, Y., Takagi, K., Tani, M., Wang, H., 2008. Temporal and spatial variations in the seasonal patterns of Page 5

Opinion Flux towers in the Amazon forests Bart Kruijt

‘Tropical rain forests are climax and slippery unpaved road jour- ests were a net carbon sink was, porting projects, most of these ecosystems, in equilibrium with neys made maintenance quite an though new and exciting, pretty towers have survived until now their environment and thus they adventure. We needed to keep premature. and data sets are now around cannot be a source or a sink of laptops and pumps going in the After these first results it lasted ten years long! Meanwhile, middle of wet and hot forests, a few years before finally long more towers have joined the CO2’. This was the dominant view on the role of rain forests and carry in batteries for power term measurements were Amazon network. There are in the global carbon balance and all the time. To the surprise of started. Within the scope of the towers in transitional forest its ‘missing sink’, until the mid- many, these measurements ap- Large-Scale - (Mato Grosso state), in season- nineties. Then John Grace, Anto- peared to show substantial car- Experiment in Ama- ally inundated savanna (Ilha do nio Miranda, Carlos Nobre, bon uptake, questioning the zonia (LBA), a large cooperation Bananal, Tocantins), and even in Antonio Nobre and John Gash, climax hypothesis! Looking back between Brazilian, South - the more remote and extremely and their teams from the UK at work by early pioneers, Steve American, North-American and wet north-western Amazon (Sao and Brazil started to measure Wofsy and David Fitzjarrald (the European scientists, six tower Gabriel da Cachoeira). ABLE project, but here we have sites were planned and installed So, what are they telling us? All CO2 fluxes with tall flux towers in the Amazon, within the AB- only a few hours to days of between 1999 and 2001, in sea- in all, the final word on Amazon RACOS project (abraços is Bra- data!), this uptake even seemed sonal, non-seasonal, central and forest carbon uptake has not zilian for ‘hugs’). First in the to be rising over time. Because eastern forest as well as in a fallen yet. First of all, most flux Southwestern Amazon state of the maintenance of these remote savanna (‘Cerrrado’) region. towers have shown a sustained Rondonia, later in central Ama- towers was very demanding only Apart from that measurements carbon uptake. This uptake is zonia north of Manaus. Getting a few months of data could be were also started in two still controversial, however, to these towers was not a sim- collected at these sites, so the (deforested) pastures. Over because calm nights are domi- ple trip: whole-day canoe trips conclusion that Amazonian for- years, and across various sup- nant and traditional u*-based corrections lead to large correc- tions. Depending on where the threshold for acceptable u* val- ues lies, most apparent sinks turn into a zero balance or even source of carbon. This realiza- tion triggered several research projects that looked into the night-time carbon budget in more detail. For example, Jon Lloyd, during the CLAIRE-II atmospheric sampling project, made vertical airplane CBL pro- files above the forests north of Manaus, and concluded that the boundary-layer budgets imply that there is no net sink or source of carbon. Studies with Radon profiles in the area south of Santarem (Tapajos national forest) showed that at least in

Figure 1: Amazon plateau-valley landscape shrouded in mist. Page 6

Flux towers in the Amazon forests

that area, there are indeed large zon forests around 1 (+/- 1) that some of the forests appear previously assumed. Most likely night-time drainage losses of tonne per hectare per year. more productive (in NEE) in the this is because rooting is deep.

CO2, confirming the u* correc- Apart from the ‘grand’ question dry seasons, even in the very dry But that is not enough to explain tion. In the area north of about the Amazon carbon year 2005, than in the wet sea- the seasonal dynamics, as pro- Manaus, however the tower is budget, the accumulating eddy sons, peaking towards the end of ductivity does not simply peak situated in moderately complex correlation data are of course the dry season. This especially with peaking radiation instead. terrain, on top of a plateau in- providing a wealth of other in- applied to the Central-Eastern Currently the various research- cised by valleys of around 40 m formation. One important as- forests of Tapajos and to a lesser ers are looking into the predic- deep. The tower fluxes are sug- pect, on which we have focused extent also to the forests north tion of leaf area phenology as a gesting that strong underestima- for the past few years, is of Manaus, as was shown by clue. tions of fluxes during most nights whether we can use these data Scott Saleska and colleagues. Where is the Amazon tower are partly compensated by to assess the effects of climatic Many vegetation models have network going in the fu- flushes of CO2 during the morn- ture? The existing towers are ing hours. By measuring the CO2 still kept going actively by the concentrations along the slope Brazilian research institutes from plateaus to valley bottoms, INPA (Antonio Manzi), INPE Alessandro Araújo indeed (Celso von Randow), MPEG

showed that that CO2 drains (Leonardo Sá), USP (Humberto downhill and pools in valleys Rocha) and others. At the same during nights, but we also show time, the network is being ex-

that these stores of CO2 con- panded, mainly with Brazilian tribute significantly to the fluxes national and private funding, to measured on the tower during better cover the enormous the morning. Julio Tota per- variability in forest types and eco formed a series of detailed in- -climatic zones. Priorities for this canopy advection measurements. are southern Amazonas state, These suggest that what we with distinct clay , the west- consider drainage flow or uphill ern Amazon (Peru, UK- flow, in fact is part of a complex AMAZONICA project) and along-valley flow pattern trans- secondary forests. With recently

porting the CO2 across the land- even the inclusion of scape, sometimes uphill, some- flux measurements, and possibly times downhill. This does not in the near future a really tall Figure 2: Tower photo of K34 with Antonio Nobre on top make things easier: we need to tower (300 m) we are now look- find a way to correct for only drought. There are a few active difficulty in representing this. ing at a mature, still very vibrant part of the drainage! To assess experiments in the Amazon, The more strongly seasonal network of greenhouse-gas flux the carbon budgets, we cannot employing rainfall exclusion (run forests in SW Amazonia are, on research sites, worth to work do without also considering by Daniel Nepstad and col- the contrary, clearly more pro- with and to invest in! results of biometric plot studies, leagues and by The University of ductive in the wet season, and as for example are being assem- Para and Patrick Meir of Edin- the RAINFOR plot data did contact: bled in the RAINFOR network, burgh University) but the multi- show net carbon losses during Bart Kruijt initiated by , year flux data could also give us the dry year 2005. The hunt is [email protected] Oliver Phillips and colleagues. All some clues, through the analysis now on to elucidate the reasons Wageningen University and Research information taken together, we of dry seasons and dry years. for the apparent ‘reverse sea- Centre, Netherlands might reach a consensus esti- Now the somewhat surprising sonality’. Cleary, soil water avail- mate for average NEE of Ama- conclusion of this analysis was ability is less important than Page 7

Data Resources MODIS Data Complements Flux Tower Information by Robert Cook, Suresh K. Santhana Vannan, and Tammy Beaty

The MODIS (Moderate Resolu- oped a tool that prepares and fered in both tabular ASCII for- also provided. An example of tion Imaging Spectroradiometer) distributes subsets of selected mat and in GIS compatible Geo- this is shown in Figures 1 and 2 aboard NASA’s Terra MODIS Land Products in a scale TIFF formats. Time series plots for the Collelongo-Selva Piana, and Aqua satellites provides and format useful for field re- and grid visualizations, along with Itlay flux tower site. These information about Earth’s eco- searchers. The subsets are of- quality assurance indicators are MODIS subsets are produced system characteristics and dy- namics on a near-daily basis for a resolution of 250 to 1000 m. The MODIS sensor acquires data in 36 spectral bands which are used to produce standardized products including phenology, land cover, primary production, vegetation characteristics (leaf area index and vegetation indi- ces), reflectance, and albedo. The MODIS products provide information about dynamics at a space and time scale appropriate for the flux tower community. MODIS data can be used to scale point meas- urements to areas comparable to a tower’s footprint and they can be used to drive models to understand dynamics of carbon, water, and energy. However, the MODIS products require specialized software and considerable time to process. These products are distributed as data files covering a 1200 km square area of the earth in a data format intended for regional analysis and modeling. These large data files are not optimal for use in field investigations at individual sites (100- x 100-km or smaller). In order to provide MODIS data for field investigations, the Oak Ridge National Laboratory (ORNL) Distributed Active Ar- Figure 1: Location information for Collelongo-Selva Piana, Italy flux tower available from the MODIS Subsetting Tool. Top left-hand panel is a chive Center (DAAC) has devel- Google Map showing the areal extent of the MODIS subset. Top right-hand panel is a LandSat image served from NASA’s Jet Propulsion Lab to the ORNL DAAC’s Web Site. Bottom panel is a Google Map of the flux tower location. Page 8

MODIS Data Complements Flux Tower Information

The tool enables students to use the data including quality flags, without having to know image processing. Kirsten de Beurs’ Remote Sensing and Phenology class at Virginia Tech compared MODIS vegetation data to ob- servations of phenological events on campus, such as the bloom of dogwood trees, to obtain a broader interpretation of these start-of-season signals (Beitler, in press). Tristan Quaife used the MODIS Web Service as part of the Summer Course in Flux Measurements and Advanced Modeling, (University of Colo- rado) to examine the relation- ship between remote sensing and flux tower observations. More information about the MODIS subsets and tools to access them can be found at the Figure 2: An example MODIS data subset visualization from 2000 to 2006 of Gross Primary Productivity (cumulative 8-day values in kg C m-2) for the Collelongo- Selva Piana, Italy flux tower site. link for “MODIS Subsets” on the NASA ORNL DAAC site at automatically for more than compatible data files for the Kepler or Matlab; to integrate http://daac.ornl.gov/MODIS/ 1,200 sites around the world, customized subsets are also the data into workflows on their including the FLUXNET tower distributed through this tool. desktop computer; and to write sites. The locations of FLUX- Users place an order for a custom code to use the subsets contact: NET research sites were regis- MODIS data subset with the for visualization or data refor- Robert Cook tered with the ORNL DAAC so subsetting tool online and the matting. For example, research- Division, that these data are now available subset is typically produced ers are using the MODIS Web Oak Ridge National Laboratory to researchers from the begin- within 60 minutes. An email is Service as part of their data [email protected] ning of MODIS data acquisition sent to the user with the neces- assimilation / data fusion meth- in the year 2000 to the present. sary information to access the ods to improve predictions from References In addition to offering subsets custom data product. ecosystem dynamics models as Beitler, J. 2009. Notebook and Satel- for fixed locations, the ORNL Researchers can also use Web well as to improve process rep- lite. Sensing Our Planet: NASA Research Features 2009.On-line DAAC also offers the capability Services to access subsets of resentation within the models. article: to create user-defined subsets these MODIS data products. A number of professors are http://nasadaacs.eos.nasa.gov/articl for any location worldwide. The Web Services facilitate data using the MODIS Subsetting es/2009/2009_plants.html DAAC’s MODIS Global subset- processing directly on the user’s tools in graduate and under- ting tool provides data subsets local computer by providing graduate courses, to demon- from a single pixel up to a 201- x access functions: to retrieve strate use of remote sensing 201-km area for any user- subsets through command line data to understand ecosystem defined time range. Data statis- operations; to import subsets dynamics, phenology, and carbon tics, time series plots, and GIS directly into software such as dioxide fluxes (Beitler, 2009).. Page 9

Editorial FLUXNET Young Scientists Network by Rodrigo Vargas, Sebastian Wolf, Matthias Zeeman

Everyone has been a young The YSN provides an informal page. [email protected] . scientist at one point in their exchange platform for questions To participate in the YSN , a We invite all young scientists to registration for the mailing list academic carrier. The FLUXNET regarding research, career and join or continue with this net- Young Scientists Network (YSN) funding opportunities for young Is needed at https://acs.lbl.gov/ working opportunity. The suc- was established in 2004 with the scientists. Young scientist can mailman/listinfo/young-scientist. cess of the YSN will depend on objective to connect young sci- present themselves and meet Afterwards, the subscriber will the feedback, interaction, partici- entist within the FLUXNET others in the YSN with interac- receive details to register for pation, and energy of young community and the regional flux tive discussions and personal access to the interactive website scientists. The interactive web- networks. The forum was initi- websites. Furthermore, estab- (password protected). This new site contains multiple tools for ated by Alexander Knohl and a lished researchers [without website consists of forums for networking, but only with active group of ex-”young” scientists. being “young” in age or spirit] questions and ideas, blogs, wiki, participation it can be a useful Thanks to their commitment can search for potential candi- calendar of events, and a file networking and scientific re- many of the former FLUXNET dates to join their research repository maintained by an source. young scientists are now well groups. active community of young sci- The YSN is open for new ideas integrated in the scientific com- The YSN has grown and evolved entists. and feedback and we hope they munity. along with the FLUXNET com- With the establishment of a new can be expressed in the mailing The YSN serves as a platform munity as new young scientists interactive and password pro- list and the interactive website. for young scientists within the incorporate to the network. tected website it seems appro- We look forward to see the FLUXNET community. This After the first 5 years of the priate to start a new mailing list YSN grow and evolve with the network goes beyond “biological YSN the accomplishments were to keep track of members and active participation of young and age” and is focused rather on assessed and ideas were dis- updates. Thus, the old FLUX- not so young scientists. “academic age”, Thus, FLUXNET cussed on how to continue and NET young scientist mailing list Further information on the YSN young scientists are students improve these networking activi- [[email protected]] can be found at: (from undergraduate to PhD) ties. The YSN is now comprised will be closed in January 2010 and post docs who do not lead of two main tools: a new mailing and fully replaced by the new FLUXNET Website – Program their own research groups. list and a new interactive web- mailing list: Information – Young Scientists Forum (http://www.fluxnet.ornl.gov/flux net/youngscientists.cfm)

Acknowledgements We thank Deb Agarwal, Robert Cook, Marty Humphrey and Susan Holladay for invaluable support to the FLUXNET Young Scientists Network. Dennis Bal- docchi and Alexander Knohl provided feedback and experi- ences for updating the YSN.

Figure 1: Screen shot of home page of the interactive Young Scientist Forum website Page 10

Highlight Young Scientist Natsuko Yoshifuji

I belong to Kyushu University, Nature and the Earth. After that drological study at a tropical and carbon exchange, but in the Japan, and work on ecohydrol- experience I became interest in forest in Southeast Asia. My first beginning there was not a tower ogy of tropical forests in South- geography, , and geo- work in the project was to for measure- east Asia with Dr. Tomo’omi . Environmental problems measure sap flow at a hill ever- ments. Therefore, I began with Kumagai and many other re- such as deforestation and global green forest in northern Thai- preliminary measurement of sap searchers. Looking back over the warming were among the hot- land to understand the seasonal flow and monitoring leaf phenol- past, the first chance that lead test topics on mass media in variation in transpiration. Hill ogy to understand the seasonal me to become an environmental those days, and that also influ- evergreen forest is one of the pattern of transpiration and its scientist goes back to my first enced my interest. Thus, I de- typical types of forests in Thai- relationship to leaf phenology trip abroad when I was 12 years cided to study environmental land distributed on the moun- and rainfall. After three years, I old. I flew over Siberia at that science and enrolled at the Uni- tainous area over 1000 m above found that year-to-year variation time, and I was surprised to see versity of Tokyo. I joined the sea level (Tanaka et al., 2008). in the length of the canopy dura- boundless forest, which I had laboratory of forest at The climate of this area is influ- tion and transpiration period never seen in Japan as a small the University because I thought enced by Asian monsoons, spanned about 40 and 60 days at island nation. I had already that I could conduct a study where there is a clear dry season most, respectively, as a result of learned at school that there was about interaction of biotic and lasting as long as 6 months when the changes in the timings of leaf a big forest called Tiga in Siberia abiotic elements at ecosystem the mean monthly rainfall is -out, leaf-fall, and the beginning below 100 mm. At first, we and the end of transpiration expected that tree transpiration corresponding to the year-to- of a hill evergreen forest de- year changes in the timings of creases in the dry season be- rainfall occurrence (Yoshifuji et cause of the soil drought based al., 2006). Before the start of the on a few previous reports at measurement, I had thought that different types of forest in Thai- it was possible that the length of land. Contrary to our expecta- transpiration period and canopy tion, I found that transpiration duration would vary with climate peak appeared in the late dry condition, but I was surprised season when atmosphere and because the extent of the change surface soil were the driest was large considering that the (Tanaka et al., 2003). Through mean length of canopy duration this experience, I learned that it and transpiration period was is a fun to find an unexpected about 310 and 260 days, respec- result from my own data. tively. This surprise gave me new I also started ecohydrological questions. How much is the research at teak plantation in impact of such a large inter- northern Thailand with many annual variation in phenology on other researchers at the end of annual gas exchange? Is the year- my Master studies. Teak (Tectona to-year variation in phenology Figure 1: Natsuko Yoshifuji grandis Linn. f.) is a deciduous large in other deciduous forests By looking at the Tiga I under- scale. The style of research ac- species occurring naturally in in Southeast Asia? It is exciting stood that empirical experience tivities based on the field obser- Thailand, and these plantations that one result produces new is quite different from book vations was also attractive for are now prevalent in northern questions, thus I still continue learning. I was shocked and ex- me. I studied in this lab until I Thailand. The main goal of the my research because my ques- cited to realize that there was a was awarded my Ph.D. research there was also to un- tions are not ending. vast land filled with the un- While I was pursuing my Master derstand the impact of seasonal Research at the hill evergreen knowns for me in the world, and studies, I was given an opportu- and inter-annual variation in forest and the teak plantation is I wanted to know more about nity to take part in an ecohy- precipitation on energy, water, being continued with many re- Page 11

Highlight Young Scientist

Figure 2: Natsuko Yoshifuji in Thailand. searchers and students from response of transpiration to Literature Japan and Thailand, and continu- environmental variation at tropi- Tanaka K, Takizawa H, Tanaka N. ous measurement of water and cal rain forests in Borneo Island Kosaka I, Yoshifuji N, Tantasirin C, Pitman S, Suzuki M, Tangtham N, 2003. carbon fluxes by eddy covariance based on sap flow measurement. Transpiration peak over a hill evergreen method have already been I hope that I can continue my forest in northern Thailand in the late started. Both sites are included ecohydrological studies mainly dry season: Assessing the seasonal in the network of flux measure- based on the field measurement change in evapotranspiration using a ment in Southeast Asia, which by myself and make novel analy- multilayer model. Journal of Geophysical Research - , 108(D17): was established to encourage sis using remote sensing data and 4533. DOI: 10.1029/2002JD003028. comprehensive understanding of modeling. I hope new data and Tanaka N, Kume T, Yoshifuji N, Tanaka gas exchange in Southeast Asia expected or unexpected results K, Takizawa H, Shiraki K, Tantasirin C, (see previous article in this is- in the future will keep giving me Tangtham N, Suzuki M, 2008. A review of evapotranspiration estimates from sue). I am very lucky that I am further questions and passion for tropical forests in Thailand and adjacent given a chance to continue my science. regions. Agricultural and Forest Mete- studies in Thailand and other orology 148: 807-819. forest sites in this network. Flux contact: Yoshifuji N, Kumagai T, Tanaka K, data will enables me to solve the Natsuko Yoshifuji Tanaka N, Komatsu H, Suzuki M, Tanta- sirin C, 2006. Inter-annual variation in first question above. I am analyz- [email protected] growing season length of a tropical ing satellite remote sensing data seasonal forest in northern Thailand. to answer the second question. I Forest Ecology and Management, 229: also started a study about the 333-359 Page 12

Highlight Young Scientist Rodrigo da Silva

My name is Rodrigo da Silva, and at Albany, to participate in the and other in a gap. The tower in were very important to my live. I I will tell you how I made the Amazon region of the research the pasture site has 20 meters actually couldn’t believe that I Amazon Region my place. I was carried out within the LBA pro- tall. One day at my return from was getting paid for having so born in Brazil (07-Oct-1971) in a ject (Large Scale Biosphere At- working at the sites, Dr. Sakai wonderful experiences. Late in place called Rio Grande do Sul, mospheric experiment in Ama- took me to a paradisiacal place 2003 after participated to so I am Gaucho (people from zon). I fell extremely privileged called Alter do Chão. I still re- HVANS Experiment (Hudson the south). The gauchos are land to have the opportunity to study member my thoughts when I Valley Experiment) I come back man, farmers, and cowboys. that incredible region and work saw that place… “That’s the to Brazil to work in Santarém to Thus I grew up living many ad- in that large project. I went to place to live and see the kids finish my thesis. In my studies, ventures riding horses, playing Santarem in January 2001 where grow up”. The forest is powerful we showed the results from soccer, swimming in rivers and I met Dr. Ricardo Sakai who and the Tapajos River is unfor- three campaigns on which the lakes, but I always find time to showed me all sites into the gettable. During this year I went nocturnal vertical profiles of stop and listen to the . Eve- jungle and the network of four times to Santarém to work carbon dioxide, temperature, ryone thought I will be a biolo- weather stations installed in the in different projects with the flux , and winds were gist or a veterinary or some like region. Near Santarém, the LBA towers. However, in this first measured at two sites at the that, even a soccer player. But Project have four flux towers, year of my doctoral studies I Amazonian region, one above not!... I graduated in Physics, so I three in the National Forest of wasn’t sure what question I will the forest, the other at a defor- am a physicist. I made my studies Tapajos (FLONA Tapajos) and try to answer, because I had so ested region. The purpose of the in the Federal University of Santa one in the open farming field. many options and all were chal- observations was twofold: identi- Maria (UFSM) where I made my The forest towers have 68 me- lenging. Osvaldo Moraes and fying the surface accumulation Master degree studying turbu- ters tall and they are measuring David Fitzjarrald, my advisors, layer and comparing the fluxes lence and the boundary layer. In and monitoring the exchange decided that I had to go to US obtained from the boundary that time (1998) I started to processes of mass and energy for one or two years. That was layer budget to those deter- develop my skills with flux tower between surface and atmos- scary but decisive to define what mined from the eddy covariance instrumentations working in the phere, one in an old growth I would like to do. Both years at the nearby micrometeorologi- Micrometeorology Laboratory spot, one in a selective logging 2002 and 2003 in Albany, NY cal towers. More than merely coordinate by Dr. Osvaldo Moraes. Together we installed a flux tower in the middle of a valley near a hydroelectric to study turbulence over complex terrain. This first experience was amazing because I discovered that my researcher was going to be outdoors to seek information about the climate and the envi- ronment. After this first year of learning, we made several inten- sive campaigns of measurements of nocturnal boundary layer using tethered balloon in differ- ent regions of the Rio Grande do Sul, always associated with measurements of flux towers. In January 2001, at the beginning of my doctorate I was invited by David R. Fitzjarrald from S.U.N.Y Figure 1: Rodrigo da Silva Page 13

Highlight Young Scientist providing a means of comparing I and other partner (Dra. Rosa Further Reading nal surface fluxes from vertical profiles the fluxes, this study clarifies the Mourão) made a project to cre- Acevedo, O. C., R. da Silva, D. R. of scalars in an Amazon pasture, Global Change , 10(5), 886-894. physical processes driving the ate the first Masters Course in Fitzjarrald, O. L. L. Moraes, R. K. Sakai, and M. J. Czikowsky (2008), Nocturnal Morales, O. L. L.; Acevedo, O. C.; Silva, exchange at such environments. Natural Resource of Amazon in vertical CO2 accumulation in two R.; Magnago, R.; Siqueira, A. C. Notur- This is because vertical gradients Santarém. This project was ap- Amazonian ecosystems, Journal of Geo- nal Surface-Layer Characteristics at the are available at a much larger proved by the Ministry of Educa- physical Research-Biogeosciences, 113 Bottom of a Valley. Boundary-Layer , v. 112, p. 159-177, 2004. scale than provided by the tion and the first class started in (null). tower. Furthermore, we used 2009. In March I was convened Acevedo, O. C., O. Moraes, D. Fitzjar- rald, R. Sakai, and L. Mahrt (2007), the same technique used by by Dr. Antonio Manzi to take Turbulent carbon exchange in very Mahrt and Vickers (2006) and by care of all LBA Project in San- stable conditions, Boundary-Layer Meteor- Acevedo et al. (2007) to decom- tarém. Today I have so many ology, 125(1), 49-61. pose the exchange in its tempo- responsibilities teaching, training, Moraes, O. L. L.; Acevedo, O. C.; Degrazia, G. ; Anfossi, D. ; Silva, R. ; ral scales, allowing the proper coordinating and make re- Anabor, V. Surface layer turbulence identification of the turbulent searches in Santarém. I’m very parameters over a complex terrain. exchange and how it is interacts excited about the future here Atmospheric Environment, v. 39, p. with the entire accumulation and would like to encourage 3103-3112, 2005. layer. In July 2005, Dr. Antonio future collaborations as well as SAKAI, Ricardo K ; Fitzajarrald, D. R ; MORAES, Osvaldo Luis Leal de ; STAE- Manzi invited me to coordinate keep strong the established. BLER, Ralf ; AcevedoO, Otavio C ; the flux towers in Santarém, Czikowsky, Matt ; Silva, R. . Land-use working in the LBA Office at contact: change effects on local energy, water, Santarém. In August 2006, I got a Rodrigo da Silva and carbon balances in an Amazoian position in the Federal University [email protected] agricultural field. Global Change Biol- ogy, v. 10, p. 1-10, 2004. of Pará at Santarém where I Acevedo, O. C., O. Moraes, R. da Silva, teach an undergraduate course D. Fitzjarrald, R. Sakai, R. Staebler, and in environmental physics. In 2008 M. Czikowsky (2004), Inferring noctur

Figure 2: Rodrigo da Silva and Prince Charles Page 14

CO2 fluxes of tropical ecosystems with different land-use in Panama Sebastian Wolf, Werner Eugster and Nina Buchmann

Land-use change has significant sources of Panamanian ecosys- has a mean temperature of 26.5 pasture (Wilsey et al. 2002). In impacts on the carbon cycling of tems is available up to now. °C and receives 2350 mm pre- 2001, parts of the site were terrestrial ecosystems. In par- Within our project, we thus aim cipitation annually, with a pro- turned into an improved affore- ticular tropical ecosystems are to quantify the CO2 and water nounced dry season from Janu- station (as a tree diversity ex- affected by ongoing land-use vapour fluxes of two tropical ary to April (less than 50 mm periment; Potvin et al. 2004) change, primarily driven by the ecosystems with different land- rain per month). One tower has while grazing continued on the demand for timber and arable use in Panama ( and been installed in an improved adjacent pasture. land. Since biophysical and bio- pasture), to assess potential afforestation (i.e., a plantation geochemical feedbacks influence differences in the driving factors using native tree species only), Instrumentation & Experi- the global climate, an improved of net ecosystem fluxes, and to and the second one in an adja- ences at the site understanding how different land estimate the carbon sequestra- cent, traditionally grazed pasture Our flux measurement systems -use types affect carbon cycling tion potentials for both land-use (Fig. 2). Like other countries in consist of open path infrared gas in tropical ecosystems is needed. types. Central America, Panama experi- analyzers (Licor-7500) and However, continuous measure- Based on collaborations with enced considerable land-use CSAT3 sonic Catherine Potvin (McGill Univer- change in the last 60 years (Campbell), both hooked up to ments of ecosystem scale CO2 fluxes are still scarce in tropical sity) and the Smithsonian Tropi- (Wright & Samaniego 2008). The an industrial PC running a regions, with only few localities cal Research Institute (STRI), we site, part of the “Sardinilla Pro- LINUX system (Fig. 3). Addi- in Central America. Although have been running two flux tow- ject” (www.gl.ipw.agrl.ethz.ch/ tional measurements include soil carbon accounting within the ers in Sardinilla, Central Panama infrastructure/research_sites/ climate profiles, soil respiration Clean Development Mechanism (Fig. 1) since February 2007. international/panama), was fluxes, leaf area index (LAI), (Kyoto Protocol) might also be Sardinilla is located about 40 km logged in 1952/1953 and used biomass production as well as an option for Panama, no infor- north of Panama City, at 9.3° N, for agriculture for two years, grazing intensity. After solving mation on carbon sinks and 79.6° W at 70 m a.s.l.. The site before being converted into some problems in the beginning

Figure 1: Location of Sardinilla, Panama Page 15

CO2 fluxes of tropical ecosystems in Panama

Figure 2: Seasonal changes in phenology at the two Sardinilla flux tower sites. Dry season pictures were taken in March 2007, wet season pictures in June 2007 (360° panoramas). due to heavy rainfall during the an annual time scale. In contrast, became carbon sources in April son (Fig. 5). In addition, carbon rainy season, we have been ac- the afforestation system was a and May. The pasture ecosystem loss of the 9 ha pasture seemed quiring continuous flux data carbon sink on an annual time seemed to be more susceptible also to be related to grazing since June 2007. Particularly with scale, although mean midday than the afforestation system intensity. During most of the the help of our local technician assimilation was lower than that (due to soil water limitations), as year, grazing intensity was rela- in Panama, we were able to cope of the pasture during the rainy seen in the pasture assimilation tively low (8-12 cattle per 9 ha), with further challenges we ex- season. An interesting period flux being reduced gradually to but increased up to 70 cattle for perienced at our site, including was the prolonged dry season in zero during this drought period short periods. Such high stocking insects in and around our instal- 2008, when both ecosystems until the onset of the rainy sea- rates reduced standing biomass lations (e.g., ants and spiders) as well as some security issues (e.g. equipment theft).

First results Analyzing Net Ecosystem Ex- change (NEE) of both ecosys- tems from June 2007 to March 2009, we observed considerable differences in diurnal and sea- sonal NEE between both land- use types (Fig. 4). High midday assimilation rates of the pasture ecosystem were likely related to the high productivity of dominat- ing C4 grasses. However, respi- ration losses in the pasture ex- ceeded photosynthetic inputs at daily and longer time scales, resulting in a carbon source on Figure 3: Fieldwork in Sardinilla. (a) Sebastian Wolf and José Monteza (technician; right). (b) José Monteza cleaning sonic . (c) Curious visitor on pasture site. (d) Spider webs on sonic anemometer. (e) Sebastian Wolf taking soil respiration measurements. (f) José Mon- teza working at bottom of afforestation tower. Page 16

CO2 fluxes of tropical ecosystems in Panama

in Panama, at least during the establishment phase. Besides the seasonally constrained availability of water, grazing intensity seems to play a major role in the pas- ture ecosystem, leading to un- sustained carbon losses.

Outlook Measurements at the afforesta- tion tower were discontinued in July 2009 while flux measure- ments over the pasture ecosys- tem are continued until at least January 2010. With more than two consecutive years of con- tinuous data, we now have a small but unique dataset of two

Figure 4: Seasonal mean diurnal NEE from June 2007 to March 2009. April and October 2008 show selected months with low and high common land-use types in Pa- ecosystem productivity. nama. Because of the sparse flux coverage in Central America, these two datasets might also be of interest for the growing FLUXNET modelling community.

contact: Sebastian Wolf Grassland Sciences Group ETH Zurich, Switzerland www.gl.ipw.agrl.ethz.ch/people/ sewolf

[email protected]

References Potvin, C., E. Whidden & T. Moore (2004) A case study of carbon pools under three different land-uses in Panama. Climatic Change 67: 291-307. Wilsey, B.J., G. Parent, N.T. Roulet, T.R. Moore & C. Potvin (2002) Tropical pasture carbon cycling: relationships between C source/sink strength, above- ground biomass and grazing. Ecology Letters 2: 367-276. Figure 5: Fingerprints of gap-filled NEE from June 2007 to March 2009 Wright, S.J., & M.J. Samaniego (2008) and thus also assimilation fluxes, lost about 230g C/m² from June C/m² (2008: 181g C/m²). Historical, demographic, and economic leading to carbon losses from 2007 to March 2009 (2008: 114g In summary, our results indicate correlates of land-use change in the Republic of Panama. Ecology and Society the pasture system. Conse- C/m²), while the 8-yr-old affore- a carbon storage potential for 13(2): 17. quently, the pasture ecosystem station system gained about 315g the native tree species plantation Page 17

Between deforestation and climate impact: the Bariri Flux tower site in the primary montane rainforest of Central Sulawesi, Indonesia Panferov, O., Ibrom, I., Kreilein, H., Oltchev, A., Rauf, A., June, T., Gravenhorst G. and A. Knohl

Tropical rain forests in Indo- Weber et al, 2007 estimated the Continent, is on the one hand forest relicts in the area. These nesia are challenged by land average rate of anthropogenic particularly exposed to the ir- facts make the forests a unique use and climate changes deforestation on Sulawesi of regularities of global circulation object for medium- to long-term Tropical forest ecosystems play about 0.6% yr-1 for the period of patterns (Walker Circulation, studies of rainforest ecosystem an important role in the interac- 1971-2001. Additionally to an- Hadley Cell) like ENSO due to functioning and responses to tion between climate and bio- thropogenic deforestation some its location in Western Equato- climate variability and climate sphere. They are vulnerable to losses of forests can be also rial Pacific. On the other hand it anomalies. Therefore, over the external forcing, especially along caused by natural factors, e.g. is characterized by a highly het- last nine years a large multi- and their boundaries (in transient , extreme weather erogeneous mixture of moun- trans-disciplinary research pro- zones between land use types) events including El Niño South- tainous islands and ocean surface gram Stability of Rainforest Mar- and, when affected, may exercise ern Oscillation (ENSO) caused which in some parts amplify and gins (STORMA) funded by the important controls on the global droughts. Deforestation of in some parts inhibit the impact German Science Foundation Earth system. Recent reports tropical rain forest areas results of global circulation. Our analysis (DFG) was carried out in Cen- showed alarming data on net in changes of energy, H2O- and of remote sensing data has tral Sulawesi in order to assess loss in tropical forest areas CO2-budgets of the land surface, shown (Erasmi et al., 2009) that the natural and socio-economic (FAO, 2000). A particular rapid and as a consequence, in changes the montane rainforests of Su- factors leading to deforestation decrease of tropical forests is of local and regional climate lawesi are particularly resilient to (www.storma.de). observed in South-East Asian (Lawton et al., 2001). The Lore the drought anomalies caused countries, where areas covered Lindu National Park (LLNP) in even by strongest ENSO-warm The Bariri Flux Site (BFS) by forest decreased from 53.9% Sulawesi plays an essential role in events (El Niño), but its future The investigation area is located in 1990 to 48.6% in 2000 (UN, conservation of endemic flora sensitivity to the changing fre- on Central Sulawesi, Indonesia 2005). In many regions including and fauna. Therefore, deforesta- quency and extent of climatic (Fig. 1). The study site near the Indonesia such changes are con- tion, either natural or anthropo- variability is unknown. The cur- village Bariri (1°39.476’S, 120° nected with a rapidly growing genic, results in serious threats rent trends in land use change 10.409’E) is a part of the LLNP human population and non- for remaining rainforest ecosys- indicate that these mountainous and lies on a small plane that is sustainable land-use practices for tems and in loss of biodiversity. rain forests might soon become surrounded by mountain chains food and timber production. Indonesia, the so called Maritime the only remaining tropical rain surmounting the plane by 300 to

Figure 1. Map of Indonesia (left) and a geographical location of Lore Lindu National Park in the island of Sulawesi (right). Yellow star at the right panel shows a position of the flux tower. Page 18

Between deforestation and climate impact

Figure 2: A view on the main 70 m tall flux tower at BFS (left) and one of the two additional towers for measuring horizontal advection (right) Photos H. Kreilein.

400 m. Within a 500 m radius 7500, Li-Cor, Lincoln, Nebraska, A strong carbon sink? implementation of the common around the tower, elevation is USA) which are installed on the The long-term eddy covariance light use efficiency approach in 1416 m a. s. l. with a standard 70 m tall steel tower (Fig 2). data starting in 2003 indicate tropical forests. The analysis deviation < 20 m. The plane is Additionally short and long wave that the forest at BFS is an unex- showed that the irradiance level, slightly inclined (< 5%) towards radiation, air temperature and pectedly strong sink for atmos- at which photosynthesis satu-

West. The mean annual tem- , horizontal wind speed pheric CO2 (Fig. 4). The data rated, decreased with increasing perature is 19.1°C, relative hu- and precipitation rate are meas- show considerable seasonal vapour saturation deficit, midity 83.3% and precipitation is ured at various heights. The sets variation both of the net flux and thereby explaining the effect of 2700 mm yr-1 (Rauf, 2009). measuring the incoming and its components, partly reflecting dryer seasons on carbon uptake.

outgoing short and long wave wetter and drier periods. Using The overall annual CO2 budget Measuring fluxes in a tall radiation are mounted above and these data, Ibrom et al. (2008) estimate based on the single montane tropical forest below the tree canopy. Due to suggested a novel approach tower eddy covariance measure- Measurements at BFS began in the complex topography of the describing the dependence of ments amounts to a high net- October 2003 and still continue. site (Fig. 3) we expected vertical GPP on absorbed photosyntheti- uptake of 970 g m-2 yr-1. Like in The eddy covariance system and horizontal advection to play cally active radiation. The other tropical sites the carbon consists of a three-dimensional a relevant role and installed two method allows overcoming the budget is relatively uncertain, sonic anemometer (USA-1, Me- additional towers with 6 sonic problem of saturation of canopy because of the high proportion tek, Elmshorn, Germany) and an anemometers to quantify these photosynthesis under the high of calm nights. The complex open-path CO2–H2O sensor (LI- effects (Fig 2). irradiance which complicates the terrain complicates the correct Page 19

Between deforestation and climate impact

with biometric assessment of the and social constraints. Eds. T. carbon pools and their changes Tscharntke, C. Leuschner, M. Zeller, E. Guhardja and A. Bidin, Springer Verlag, and with biophysical modelling Berlin, pp. 463–492. using field measurements of Ibrom, A., A. Oltchev, T. June, H. photosynthesis at leaf level and Kreilein, G. Rakkibu, T. Ross, O. Pan- canopy structure parameters ferov and G. Gravenhorst 2008. Varia- (Rakkibu, 2008). tion in photosynthetic light-use effi- ciency in a mountainous tropical rain Current analyses showed that forest in Indonesia. Tree Physiology. measuring carbon fluxes at this 28:499–508. unique site provide excellent Lawton, R.O., Nair, U.S., Pielke Sr., R.A. means to observe ecosystem and Welch, R.M., 2001. Climatic impact of tropical lowland deforestation on responses to climatic variability nearby montane cloud forests. Science and to investigate carbon fluxes 294, 584-587 in tall vegetation and complex Olchev, A., Ibrom, A., Priess, J., Eras- terrain. It is essential that flux mie, S., Leemhuis, C., Twele, A., Radler, Figure 3: A view on the 70 m tall flux tower at BFS within the landscape. investigations at the Bariri Flux K., Kreilein, H., Panferov, O. and G. Gravenhorst, 2008. Effects of land use Site continue, because (i) it is changes on evapotranspiration of tropi- one of only two sites worldwide cal rain forest margin area in Central where primary mountainous rain Sulawesi (Indonesia): modelling study forest can be studied in situ and with a regional SVAT model. Ecol. Mod., (ii) it is located in South East 212, 131–137. Rakkibu, G., 2008. Photosynthesis in a Asia with its unique tropical tropical montane rainforest of South- climate, a region of tight land- east Asia: field measurements and atmosphere-ocean interaction, model analysis, PhD Thesis, pp 114. but where coverage with flux Rauf A., 2009. Rainfall interception and effect of energy and mass transfer on sites is extremely low. tropical rainforest, case study in Lore Lindu National Park. PhD Thesis, IPB contact: Bogor, pp. 128 Oleg Panferov Ross, T., 2007. Eddy-Kovarianz Figure 4: Annual variation of measured net ecosystem exchange flux Fc, of extrapolated Messungen über einem tropischen respiration flux RE, and the sum of both fluxes gross photosynthesis (Pg) at the Bariri Tower Department of Bioclimatology, -2 -1 Regenwald in komplexem Gelände, PhD site (g (C) m 10 days ). Values are not corrected for advection. Georg-August University, Buesgen- Thesis, Göttingen, pp. 134 interpretation of the measured factors, the measured data were weg 2, Goettingen, Germany Sogachev, A. and Panferov, O., 2006. signal and the contribution of the classified according to the condi- Modification of two-equation models to surrounding forest especially tion of atmospheric stratification [email protected] account for plant drag. Bound. Lay. during night time. Therefore we (stable, neutral, unstable) and the Meteo. 121, 229–266. Weber, R., Faust, H., Schippers, B., used the 3D atmospheric bound- separate corrections was applied References Mamar, S. Sutarto, E., & W. Kreisel, ary layer model SCADIS based for each situation. The correc- Erasmi, S., Propastin, P, Kappas M. and 2007. Migration and Ethnicity as Cul- on E-ω closure (Sogachev and tion reduced the net uptake by O. Panferov, 2009, Patterns of NDVI tural Impact Factors on Land Use variation over Indonesia and its rela- Panferov, 2006) to evaluate the 152 g (C) m-2 yr-1 under condi- Change in the Rainforest Margins of tionship to ENSO during the period Central Sulawesi, Indonesia. In: The influence of heterogeneous ter- tions of neutral stratification and 1982-2003, J Climate, accepted stability of tropical rainforest margins: -2 -1 rain on CO2 and H2O fluxes by 204 g (C) m yr at stable Ibrom, A., Oltchev, A., June, T., Ross, Linking ecological, economic and social between rainforest and the at- atmospheric stratification result- T., Kreilein, H., Falk, U., Merklein, J., constraints. Eds. T. Tscharntke, C. mosphere. The estimated cor- ing in an about 20% lower esti- Twele, A., Rakkibu, G., Grote, S., Rauf, Leuschner, M. Zeller, E. Guhardja and A., Gravenhorst, G., 2007. Effects of A. Bidin, Springer Verlag, Berlin, pp. 463 rection coefficient varied mation of annual CO2 uptake land-use change on matter and energy –492 (depending on wind direction) (Ross, 2007). As an independent exchange between ecosystems in the from 0.6 – east wind to 1.5 west measure, we try to estimate the rain forest margin and the atmosphere. wind. To apply the correction carbon uptake of the system In: The stability of tropical rainforest margins: Linking ecological, economic Page 20

Looking back and looking forward at eddy flux studies in forests David R. Fitzjarrald

[“Do not confuse the star you of using the eddy covariance These efforts feed interesting measure heat, moisture and CO2 steer by for your destination.” approach finally won over the data-model fusion activities. Do budgets through the forest can- John Barth, Lecture, Univ. of ecological community. However, we risk introducing a bias with opy. In the room were veterans early eddy CO flux results indi- our well-behaved data? What of BOMEX and GATE, large- Illinois, 1968.] 2 cated that nocturnal that made information is discarded? scale tropical meteorological little ‘ecological sense’—too campaigns (e.g., Esbensen 1975). When eddy covariance fluxes much carbon was accumulating Horizontal subcanopy mo- It seemed natural to apply the were finally accepted by the given growth rates estimated tions and scalar budgets same techniques to estimate using other approaches. The water and energy budgets on the ecological community, the need [“It's one of those things a to account for ecosystem respi- Goulden et al. ‘u correction’, microscale in forests. * person has to do; sometimes a ration reinvigorated meteoro- the solved this issue for many— To the students, the nocturnal logical interest in nocturnal tur- simply replace measured fluxes person has to go a very long canopy air density profile resem- bulent fluxes. When water vapor on calm nights with ‘reliable’ distance out of his way to bled the tropical convective and heat fluxes were paramount, ones measured on well-mixed come back a short distance boundary layer (CBL), so Wood- micrometeorologists concen- nights. Also, discard rainy peri- correctly.” Edward Albee, The well & Dykeman (1966) estimat- trated on the more tractable ods, and finesse the lack of en- ing the forest canopy respiration Zoo Story.] daytime convective cases. It ergy balance closure with artful rate by monitoring subcanopy takes some audacity to apply the adjustments. Overlook transient Consider the issue of the calm CO2 accumulation seemed eddy flux approach to get long- events (e.g., nocturnal CO2 vent- night eddy CO2 fluxes. If we do straightforward. We admired term flux averages in forests. ing). Now many labor to pro- not measure CO2 emerging Odum & Jordan's (1970) quixotic After years of seeking places duce regular flux time series, vertically from a forest on calm effort to enclose the rain forest with no obstacles to measure happily supplanting observations nights, does it go out the sides of in a large plastic cylinder at the fluxes, researchers feared the with corrected numbers, filling the box? What observational Luquillo Forest. We saw no subcanopy (Fig. 1). Many other gaps by inserting fictional output arrays are needed to estimate mystery in ‘countergradient difficulties flowed directly from from a site-specific model. The subcanopy trace gas divergence transport’ in the canopy— the nature of Reynolds averag- resulting continuous record then in forest canopies? The answers buoyant advection accomplishes ing. In stable conditions, effects ‘validates’ (such a self-serving have been trickling in over the the same in the CBL, where the of temporal or spatial inhomoge- term!) another model or re- last few years, but they might scale of mixing eddies is larger neities are exaggerated, as turbu- motely sensed data. It is consid- have come much earlier. In 1978, than that of the local gradient. lent intermittency, breaking ered acceptable to discard 60- a mixed crew of ecology and How small a budget box would waves, and drainage flows all 90% of nocturnal data at some meteorology graduate students be viable? In the end the project confound the situation. sites with surrogates and still sat with Prof. Garstang in Char- was scuttled: Somehow we had In the end, the clear advantages refer to the product as data. lottesville to design a project to overlooked Desjardin’s work

Figure 1.:Harvard Forest, Sept. 12, 2002. Rangefinder scan with panoramic photo. A 2D sonic anemometer is visible at 340° (sonic anemometer, arrow). A clearly resolved twin tree at 540° is also marked. (from Staebler 2003). Page 21

Looking back and looking forward...

with eddy CO2 fluxes over canopy produce mean net trans- crops, our topic fell between port of CO2 into or out of the disciplines, professors squabbled, control volume; and 4) A sub- and no funding ensued (related canopy network had to include events). We could have been 20-30 cases of light winds and contenders! These ideas were clear skies to capture the rele- left to incubate for years. It was vant gradients and transport eight years before my seminar processes. Few did. We consid- on Amazon forest eddy fluxes ered it unfeasible to make advec- with the immortal title (no one tion corrections on the 1-hour recalls anything else): time scale, arguing for clear night “Intermittent nocturnal coupling composites. We retained a gen- in the Amazon jungle.” eral uneasiness about how to

When the ‘u* filter’ issue bub- deal with quantitative estimates bled up, we looked at subcanopy of canopy structure. The elusive

CO2 advection at Harvard For- goal continues to be to estimate Figure2: Harvard Forest. The average canopy area density (CAD) distribution, its standard est (Staebler and Fitzjarrald the likely effect of horizontal error (dotted) and the winter profile of a similar nearby mixed deciduous stand (dashed; Staebler and Fitzjarrald, 2004) 2004, 2005), the ‘DRAINO’ advection at a given site knowing projects. These started in the local topography, estimates of ments were placed to monitor Ruminations & future direc- growing season of 1999 and, as the respiration rates, the mean slopes on either side of one tions. resources became available and vertical vegetation profile (e.g., airshed gully adjacent to the flux [“A good teacher teaches well, we understood instrumentation Fig. 2), and the wind aloft. tower. Tóta found that colder regardless of the theory he issues better, an observation Our team made similar observa- air could move CO2 laden air strategy emerged over the fol- tions at the LBA-ECO tower site uphill at intervals, suggesting the suffers from.” John Barth, lowing three summers. In mod- in the Tapajós National Forest, presence of atmospheric seiches Lecture,Univ. of Illinois, 1968.] erately complex terrain, we Brazil (Tóta et al. 2009). This site in the microbasin. learned to measure the direction was ideally ‘flat’ except that it In these studies, we always pos- What is our goal? Modelers and that the drainage flow follows was close to a cliff. We found tulated a similar vertical CO2 field observationalists alike seek (not necessarily down the local that the forest floor was so profile shape, forced on us by generalizations. As we rush so terrain gradient), using a ‘slope decoupled from the flow aloft financial circumstance (e.g., Fig. confidently to make those lovely flow rose’. We learned that that drainage flow toward a local 3). In 2008, we returned to Har- global maps of NEE, ET, appar- stations 50 m apart could give microbasin was observed day vard Forest to look at subcanopy ently ‘solved’ problems should reasonable results, but that 20- and night, casting doubt on the flows around another flux tower, perhaps be revisited at intervals.

40 cases were needed to tease relevance of the nocturnal u* one in simpler topography with Data storage is no longer an out the small signal. Several correction alone. We found that regular subcanopy anabatic and issue, but finding the resources issues were identified but not the horizontal advection could katabatic flows. Case studies of for mining existing raw data re- understood: What is the source indeed be measured, and it could canopy in-canopy scalars are mains a major challenge. Re- of the cold air that motivates the account for most of the ‘missing planned to refine our profile cently a conference speaker drainage flow, and how does it flux’. shape factors. Linking new sodar confidently announced, “The era get to the forest floor? Tóta is now documenting a observations upwind and in the of curiosity-based science is We hoped that others would second series of observations at lee of local topography with the over…” Will committees be notice our methodology for the ZF-2 tower site near earlier work at Harvard Forest adequate to map out the future quantifying horizontal subcanopy Manaus, an area of dissected will help us evaluate whether the for us all? advection: 1) Show that system- terrain that forms an airshed above-canopy flow can signifi- Observers regularly come upon atic subcanopy flows are measur- that promotes complex sub- cantly alter subcanopy motions unexpected and puzzling phe- able; 2) Such flows are related to canopy motions. At this site the on calm nights. nomena. This inclines me to a physical driving mechanism; 3) original DRAINO approach consider observational evidence

Observed CO2 gradients and could not work given our limited in ‘eco-micrometeorology’ as transport processes in the sub- equipment. Subcanopy instru- primary to theory regardless of Page 22

Looking back and looking forward...

contact: FluxLetter David R. Fitzjarrald Atmospheric Sciences Research The Newsletter of Center, University at Albany, SUNY FLUXNET [email protected] Vol.2 No.3 October, 2009

References FluxLetter is produced Czikowsky, M. J. & Fitzjarrald, D. R. quarterly at the FLUXNET (2009) Detecting rainfall interception in Office with support from the an Amazonian rain forest with eddy flux National Science Foundation. measurements. Journal of Hydrology 377:

92-105. Esbensen, S. (1975) An Analysis of Subcloud-Layer Heat and Moisture Budgets in the Western Atlantic Trades. Journal of the Atmospheric Sci- ences 32: 1921–1933. Odum, H. T. & Jordan, C. F. (1970) Metabolism and evapotranspiration of the lower forest in a giant plastic cylin- der. in Odum, H.T., Pigeon, R.F. (Eds.): A Figure 3: Tapajós National Forest, Brazil.Typical nighttime normalized median profiles of CO2, wind speed and their product (uc) horizontal transport (from Tóta et al., 2008). Tropical Rain Forest. pp. I165-I189. Oak Ridge, TN: U.S. Atomic Energy Com- mission, Division of Technical Informa- theoretical elegance. Our under- approach: tion. standing of turbulence—e.g., the • To what extent are parame- Staebler, R. M. & Fitzjarrald, D. R. characteristic time and length terizations that link current (2004) Observing subcanopy CO2 advection. Agricultural and Forest Meteor- scales in canopies—is incom- atmospheric ‘forcings’ to plant ology 122: 139-156. plete. Since we often report response compromised by cir- Staebler, R. & Fitzjarrald, D. (2005) fluxes without commensurately cadian rhythms, shown in some Measuring canopy structure and the detailed description of canopy studies to march along regard- kinematics of subcanopy flows in two structure, rules to generalize less of local conditions? forests. Journal of Applied Meteorology 44: 1161-1179. momentum transport in cano- • Is the observed enhanced car- Staebler, R. M. (2003) Forest Sub- pies are drawn from intuition in bon uptake on cloudy days only canopy Flows and Micro-Scale Advec- This issue of FluxLetter was models or from a small number the result of the diffuse radiation tion of Carbon Dioxide, Ph. D. disserta- edited, designed and produced by: of intensive field observations. fraction, or does it matter what tion, University at Albany, SUNY. : 179 pp. Another way to extract more light-dark cycles the clouds im- Tóta, J., Fitzjarrald, D., Staebler, R., Rodrigo Vargas information from the accumulat- pose on the canopy? Sakai, R., Moraes, O., Acevedo, O., Dennis Baldocchi ing Ameriflux records is to con- Wofsy, S. & Manzi, A. (2009) Amazon sider alternate definitions of the rain forest subcanopy flow and the FLUXNET Office, 137 Mulford carbon budget: Santarém LBA-ECO Hall, University of California, ensemble mean in the Reynolds Berkeley, CA 94720 site. Journal of Geophysical Research G: average to form event-based ["How more satisfying when ph: 1-(510)-642-2421 Biogeosciences 114. Fax: 1-(510)-643-5098 composites. We recently used the voyage, not the port, was Woodwell, G. M. & Dykeman, W. R. this approach to estimate rainfall (1966) Respiration of a Forest Meas- our destination. If life's a jour- We plan to make the FLUXNET interception at the Tapajós site ured by Carbon Dioxide Accumulation ney and the grave its goal, newsletter a powerful information, reanalyzing existing eddy flux during Temperature Inversions. Science networking, and communication 154: 1031-1034. data (Czikowsky & Fitzjarrald getting there is all the fun. -- resource for the community. If you John Barth, Sabbatical"] want to contribute to any section or 2009). propose a new one please contact the Answers to some questions that FLUXNET Office. THANKS!! vex me may come from this