Landscape Interactions With Thunderstorms in Interior Alaska

Dorte Dissing
Email: fsdd@aurora.alaska.edu
Dave Verbyla
Email: dverbyla@lter.uaf.edu

Preliminary results of this study was highlighted in the December 12th, 1998 issue of the New Scientist Magazine, and can be viewed at:

http://www.newscientist.com/ns/981212/newsstory12.html

Introduction

The majority of the circumpolar region is covered by boreal forest. Boreal forest is the second largest biome on Earth (Whittaker, 1975), and therefore an important factor in Earth system studies. Global climate models predict that the largest future changes will happen in the polar regions. This has lead to speculations on the influence of warmer, potentially drier summers on the fire activity in boreal forest regions (Ryan, 1991). Fire frequency and severity changes forest stand composition, which impacts regional carbon storage, albedo, surface roughness and the partitioning of latent and sensible heat fluxes (Kasischke et al, 1995). Annually, thunderstorms start fires that burn large areas in the boreal forest. For example, most of the area burned in interior Alaska is by lightning-caused wildfires. In the 1990-1996 period, over 2.7 million hectares burned in interior Alaska, and 93 percent of this area was due to lightning-caused wildfires (Boles, 1998). Thunderstorm development is a very important factor in the continental boreal forest fire regime through its influence as a fire starting mechanism. Understanding the forcing factors that determine where these storms develop is a vital part of understanding the atmosphere - ecosystems interactions that control the boreal environment. The issues addressed here were originally developed in modeling studies, however, we propose to use remote sensing and geographical information system techniques to validate the ideas. If this study indeed reveals a picture of strong, local factors forcing and controlling thunderstorm development, new perspectives on future fire regimes will be created. Additionally, if landscape heterogeneity and the presence of fire scars are part of a feedback mechanism that promotes convective activity, a strong, enforced fire regime may be a possible result of the predicted warm, dry climate. This study is an example of how local factors can influence regional scale circulation patterns for one of the worlds largest biomes, thus having a global influence. A recent study by Skinner et al. (1999) examined the correlation between the 500 mb anomaly and the fire severity in the Canadian boreal forest region, and concluded that the accurate prediction of the future Canadian fire regimes requires identification of the relationships between regional scale climate and fires.

Background

In general, the prerequisites for thunderstorm formation are: a) Sufficient amounts of moisture in the lower (< 3km) air mass. b) An unstable airmass. c) A triggering mechanism to set off the convective activity. The moisture provides the thunderstorm with energy by the release of latent heat through condensation (Grice and Comisky, 1976). The triggering mechanism acts to release the instability and is often caused by orographic or frontal lifting, low-level convergence or heating from below (Schroeder and Buch, 1970). In the Northern Hemisphere, airflow around a surface low or trough is counterclockwise and spins inwards toward the center. This is a region of convergence, where the air must rise to conserve mass. Thus the wind pattern close to the surface is a relevant factor for determining areas favorable for thunderstorm development. Reap (1991) attempted to determine which factors were adequate for predicting CG lightning occurrence and found the highest correlated parameter to be the 850-500 mb lapse rate. Anderson (1991) examined patterns in lightning strikes in Alberta, Canada, and concluded that while convection was the best factor for determining occurrence, it was not possible to predict frequency. Surface moisture, vegetation, and topography also influence the distribution of thunderstorm occurrences. Thunderstorm development, and therefore lightning strikes and potential wildfire starts, are often closely linked to regional climate.

Biswas and Jayaweera (1976) used NOAA Advanced Very High Resolution Radiometer (AVHRR) data to study the predominant patterns of thunderstorm meteorology in Alaska. They found that most of the lightning-caused fires in the state are a result of Air-mass thunderstorms, which are associated with sloping topography and the effect of differential heating. These are formed in the absence of large-scale synoptic effects (Biswas and Jayaweera, 1976; Henry, 1978). Hence, continental boreal forest thunderstorms, in the absence of strong synoptic forcing, are probably tied very closely to at least elevation / topography, but could additionally be very sensitive to vegetation feedbacks through differential surface heating. Variations and magnitudes in sensible and latent heat fluxes influence local and regional climate and thus conceivably the formation of thunderstorms. The energy partitioning and the magnitude of the respective heat fluxes vary with vegetation type. In general, latent heat fluxes are greater over tundra than over boreal forest, whereas sensible heat fluxes are larger over boreal forest than over tundra (Chapin et al, 1998; Lafleur and Rouse, 1995; Pielke and Vidale, 1995; Eugster et al., 1998). Pielke and Vidale (1995) report the difference in sensible heat flux to be ~ 50 Wm-2, calculated from a daily temperature difference. Eugster et al.(1998s) measured over 250 Wm-2 difference in summer afternoon sensible heat fluxes between lakes and the surrounding vegetation. The large sensible heat fluxes over the boreal forests result in increased air temperature, supporting thermal convection and producing deep boundary layers (Pielke and Vidale, 1995; Eugster et al., 1998).

Alaska covers both Arctic, Subarctic, and Temperate climatic zones. The majority of Interior Alaska is covered by boreal forest / tundra, and the climate is continental and characterized by dry, relatively warm summers and extremely cold winters. The region is dominated by intermittent permafrost and is susceptible to significant changes in regional carbon fluxes with climatic warming and change in wildfire disturbance regime.

Data

The Bureau of Land Management (BLM) Alaska Fire Service (AFS) operates an automated network of cloud-to-ground (CG) lightning sensors. A lightning flash is reported only if detected by more than one of the sensors. Positional accuracy of estimated lightning locations vary with number of detectors sensing a strike and with detector geometry. Each strike report contains the computed location, time of the flash, plus measures of the positional accuracy. In Alaska, the network has been in operation since 1976 and consists of 9 stations within the state and 3 in the Yukon Territory. After the site error corrections were made in May 1995, based on the 1994 fire season data, the Alaskan system is assumed to have a positional accuracy of 2-4 km (Personal communication, T.Weatherby, AFS).

Results

AFS has located on average more than 26,000 CG lightning strikes per year during the period from 1986-97. The most lightning activity in Interior Alaska takes place between 4 to 6pm during late June/early July. Unlike the conterminous United States, thunderstorms in Interior Alaska are rarely nocturnal; this is probably due to the long summer daylight period at high latitudes. Figure 1 shows that the annual variability of CG Strikes is large; It varies form about 40,800 recorded strikes in 1994, to about 12,800 in 1989. The severe storms in Alaska typically yield between 2000 and 5000 strikes. How many strikes these severe events actually produce appears to be related to the total amount of CG strikes for that particular warm season.

Figure 1. Variability of annual lightning strikes

Mountains generally enhance thunderstorm development due to differential heating and orographic triggers that promotes strong, convective activity. Reap (1991), using Alaska lightning data from 1987-89, found a generally positive relationship between lightning density and elevation from sea level to about 800 meters and a negative correlation above 800 meters of elevation. Similarly to a figure from Reap (1991), we plotted lightning strike density as a function of elevation, with elevation zones of 100m and a horizontal grid size of 1km x 1km . Figure 2 shows the spatial distribution of the 100 m elevation zones across Alaska. Figure 3 shows lightning strikes per unit area, distributed over the 100 m elevation zones, to examine the relationship between mountainous areas and the likelihood of lightning strikes.

Figure 2. 100-m Elevation Zones of Interior Alaska 

 

Figure 3. Lightning strike density by elevation zone

We found peak at the 500-1000m interval and a negative correlation between lightning density and elevation above this altitude. This is similar to Reap's findings, despite the fact that Reap analyzed 2 years of data at 48 km grid cell size while we analyzed 12 years of data at 1km grid cell size. This relationship varies across Alaska's physiographic regions, and does not appear to be closely related to the location of mountains or river flats.

Reap (1991) found a clear pattern showing most lightning strikes at higher elevations earlier in the day, moving to lower elevations as the day progresses. We divided the interior Alaska region into three broad elevation zones of 0-500m, 500-1000m, and >1000m. Within each of these zones, we analyzed the hour of maximum lightning activity. The patterns for the individual elevation zones for the fire seasons from 1986-97 were examined for changes over the season and inter-elevational changes. We found most activity to be centered around 4pm local time, with very little change over the season. The highest elevation zone had the maximum activity earlier in the day relative to the lower zones for 9 of 12 time periods.

Regional patterns in the lightning data

We divided interior Alaska into regions, based primarily on physical landscape features such as topography and the major rivers, for example the Yukon, Tanana, and Copper River Flats. The zones have different areal extent, thus figure 4 is based on the density of the cloud-to-ground lightning strikes and not the total strike counts in the regions. To make the density numbers easier to comprehend, they have been multiplied by 1000.
 

Figure 4. Lightning strike density by physiographic region 

The figure shows the most dense regions to be located in the central Interior of the state, south of the Brooks Range and north of the Alaska Range. This zone is characterized by comparatively hot summers which have been suggested to have enough available moisture, due to large-scale advection, to fuel thunderstorms (Reap, 1991; Sullivan, 1963). The coastal areas have milder climate with more stable air-masses, less surface heating, and warmer air aloft, prohibiting extensive convective activity. In addition, some of the common triggering mechanisms in the Interior region, the thermal trough or the high pressure ridges, does not occur frequently at the coastal regions. A few zones south of the Alaska Range; the Copper River Flats and the Talkeetna Mountains, still report more strikes than their surrounding regions. The red zone mark the densest region, which received 40-80*10-3 lightning strikes km-2 and consists of the Yukon-Tanana Uplands and Flats, the Nowitna, Tetlin, and Kantishna River Flats and the Ray Mountains. The lowest densities are found along coastal Alaska (as explained above). The White Mountains and the Kuskokwim Mountains are the regions with the highest densities recorded; on average in the analyzed period from 1986-97 each region was hit by lightning more than 3000 times during the warm season, corresponding to 23% and 10% of the total strikes per year, respectively. The Yukon Uplands, southern part of the Brooks Range and the Ray Mountain regions all record more than 200 strikes per year on average.

We compared the lightning strike distribution to the land surface cover, combining the vegetation types into three major types; "tundra", "shrub" and "forest".

Figure 5. Forest/Shrub/Tundra Zones of Interior Alaska.

Figure 6 shows the three vegetation classes and the corresponding lightning strike densities. The data reveal an overwhelming preference for lightning strikes to hit within forested regions, followed by tundra, and least favorable is shrubs. The data suggests a very marked difference between the two groups of forest marked as "closed" and "open". The "open" forest category gets hit significantly more than the "closed". This is due to the regional distribution of the two forest categories; the closed forest is found along coastal Alaska. Other authors have found that forested, high vegetation areas are a preferred area for lightning strikes (Wells and McKinsey, 1993; Gisborne, 1927). Pielke and Vidale(1995) suggest a larger sensible heating over the boreal forest as opposed to tundra by about 50 Wm-2 as an average daily figure over a season, or 0.8 Cday-1 , which could cause the observed pattern. The question that arises from these results are: Does boreal forest really promote thunderstorm development or is it simply due to the occurrence of this biome within a climatic region which sustains the convective activity ? Which is one of the questions we attempt to address with this research.

Figure 6. Lightning strike density by tundra/shrub/forest vegetation class .

Spatial and Temporal Scales

What are the appropriate temporal and spatial scales of this analysis? We are analyzing daily lightning strike data from 1986 to present for Interior Alaska (Alaska Range to Brooks Range). The positional uncertainty of the lightning strike locations is approximately 5-10 km for most of the period (1986-1994) and 2-4 km after 1994. Our elevation data are at a grain size of 1-km grid cells and our satellite data vary in grain size from 30-m for Landsat TM to 1-km (nadir) for AVHRR.

Additional Proposed Research :

Regional properties, such as landscape heterogeneity, topography, and vegetation, control the spatial and temporal distribution of thunderstorm development and lightning strike density in continental high latitude ecosystems. Surface properties such as albedo, moisture, surface roughness, and landscape heterogeneity may play a role in enhancing sensible heat fluxes and convective activity and influencing lightning strike patterns (Rabin and Martin ,1996; Lopez and, Holle, 1986; O'Neal, 1996). My preliminary results show significantly more lightning strikes within areas of boreal forest than within areas of shrubs, tundra and water (Mulvaney, 1998). The tundra regions with high lightning strike density appear to be adjacent to forested areas. Variations in sensible heat fluxes as a result of heterogeneous landscape types (lakes, boreal forest, tundra) are capable of producing mesoscale circulations and thus influence not only the local, but also the regional climate (Eugster et al., 1998). Figure 1 is a flow chart illustration of the main concepts in this proposal. The hypothesis tested by the box or link in question is explained in the text outside the boxes, and the hypothesis number is referenced.

Figure 7. Flow Chart showing Concepts of Thunderstorm Development in High Latitudes.

Hypothesis1: Surface temperature differ significantly across vegetation types - boreal forest, tundra, shrubs and burn scars (*).

Hypothesis 2: In absence of strong atmospheric forcing, Cumulus clouds in continental boreal forests develop as a result of a) strong surface heating or b) locally initiated mesoscale circulation systems.

Hypothesis 3: Airmass Cumulonimbus (Cb) clouds (thunderheads) form mainly over highly heterogeneous areas, as opposed to over high temperature, homogeneous surfaces.

* See link to preliminary results of radiant temperatures of burn scars: ./dorte/burns.html

Literature List

Anderson, K.R., 1991 : Models to Predict Lightning Occurrence and Frequency Over Alberta. M.S.Thesis, University of Alberta, Dep. of Geography.

Boles, S.H. and Verbyla, D.L., 1998: Comparison of three remote fire detection methods using AVHRR in Interior Alaska. Remote Sensing of Environment, submitted.

Bonan, G.B., Chapin, F.S. and Thompson, S.L., 1995 : Boreal Forest and Tundra Ecosystems as Components of the Climate System. Climatic Change, Vol 29, pp 145-167.

Brutsaert, W; Hsu, A.Y.; and Schmugge, T.J., 1993: Parameterization of Surface Heat Fluxes above Forest with Satellite Thermal Sensing and Boundary-Layer Soundings. Journal of Applied Meteorology, Vol. 37, pp.909-917.

Chapin, F.S.; Eugster, W.; McFadden, J.P.; Lynch, A.H.; and Walker, D.A., 1998 Summer differences among arctic ecosystems in regional climate forcing. (Unpublished manuscript)

Eugster, W.; Rouse, W.R.; Pielke, R.A.; McFadden, J.P.; Baldocchi, D.D.; Kittel, T.G.F.; Chapin, F.S.; Liston, G.L.; Vidale, P.L.; Vaganov, E.; and Chambers, S., 1998: Northern ecosystems and land-atmosphere energy exchange in arctic tundra and boreal forest: available data and feedbacks to climate. (Unpublished manuscript).

Gisborne, H.T, 1927 : The objectives of forest fire-weather research. Journal of Forestry, pp 452-465. Referenced in Kessler, 1992.

Grice, G.K. and Comisky, A.L., 1976. Thunderstorm climatology of Alaska. NOAA technical memorandum NWS AR-14.

Henry, D.M. 1978: Fire occurrence using 500 mb map correlation. NOAA Technical Memorandum, NWS AR-21, 31 pp.

Kasischke, E.S. Christiansen, N.L.,and Stocks, 1995: Fire, Global Warming, and Carbon Balance of Boreal Forests. Ecological Applications, Vol 5(2), pp. 437-451.

Kessler, E. (Ed.), 1992 : Thunderstorm Morphology and Dynamics. University of Oklahoma Press.

Knowles, Captain J.B., 1993; M.S. Thesis: The influence of forest fire induced albedo differences on the generation of mesoscale circulations. Department of Atmospheric Science, Colorado State University, 86 pp.

Lafleur, P.M. and Rouse, W.R, 1995: Energy partitioning at treeline forest and tundra sites and its sensitivity to climate change. Atmosphere-Ocean, Vol.33, No.1, pp 121-133.

Lopez, R.E. and Holle, R.L., 1986: Diurnal and spatial variability of lightning activity in Northeastern Colorado and Central Florida during the summer. Monthly Weather Review, Vol.114, pp 1288-1312.

Mulvaney, K., 1998: Bolts from the blue. New Scientist, Vol.160, No.2164, 12. December, p.26.

Nemani, R., Pierce, L., and Running, S., 1993: Developing Satellite-derived Estimates of Surfae Moisture Status. Journal of Applied Meteorology, Vol 32, pp.548-557.

O'Neal, 1996: Interactions between land cover and convective cloud cover over midwestern North America detected from GOES satellite data. International Journal of Remote Sensing, Vol.17,No.6, pp1149-1181.

Pielke, R.A. and Vidale, P.L., 1995 : The Boreal Forest and the Polar Front. Journal of Geophysical Research, Vol 100 No d12, pp 25,755-25,758.

Price, C. and Rind, D., 1994 : Modeling Global Lightning Distributions in a General Circulation Model. Monthly Weather Review, Vol 122, pp 1930-1939.

Rabin, R.M. and Martin, D.W., 1996: Satellite observations of shallow cumulus coverage over the Central United States: An exploration of land use impact on cloud cover. Journal of Geophysical Research, Vol 101 No d3, pp 7149-7155.

Rabin, R.M., Stadler, S., Wetzel, P.J., Stensrud, D.J., and Gregory, M., 1990: Observed effects of landscape variability on convective clouds. Bulletin of American Meteorological Association, Vol.71, No.3, pp272-280.

Reap, R.M., 1991 : Climatological Characteristics and Objective Prediction of Thunderstorms in Alaska. Weather and Forecasting, Vol 6, No 3, pp 309-319.

Ryan, K.C.,1991: Vegetation and Wildfire Fire: Implications of Global Climate Change. Environmental International, Vol.17, pp.169-178.

Schroeder, M.J. and Buch, C.C., 1970 : Fire Weather. Agriculture Handbook 360, U.S.Department of Agriculture, Forest Service, May 1970.

Skinner, W.R., Stocks, B.J., Martell, D.L., Bonsal, B., and Shabbar, A., 1999 (in press): The Association Between Circulation Anomalies in the Mid-Troposphere and Area Burned by Wildland Fire in Canada. Theoretical and Applied Climatology, Vol xx, pp xxx-xxx.

Vazquez, D.P., Reyes, F.J.O., and Arboledas, L.A., 1997: A Comparative Study of Algorithms for Estimating Land Surface Temperature from AVHRR Data. Remote Sensing of Environment, Vol.62, pp.215-222.

Wells, M.L.and McKinsey, D.E., 1993 : The Spatial and Temporal Distribution of Lightning Strikes in San Diego County, California. In GIS / LIS Proceedings, 2-4 November, Minneapolis Convention Center, Volume 2, 1993.

Whittaker, R.H., 1975: Communities and Ecosystems. Macmillan, New York.