SUMMARY: This report provides the overall approach and some
results of the work
of the "Kazakhstan Climate Change Study" project in three main
areas: inventory of
greenhouse gas (GHG) emissions and sinks, mitigation assessment,
and vulnerability
assessment. In the first area, the information on greenhouse gas
sources and some
sinks, estimates of emissions and removals for Republic of
Kazakhstan for 1990, as
well as a brief description of the methodology used to evaluate these
estimates and
the associated uncertainties are given. An estimation of future
CO2
emissions from 1990 through 2000 is also presented. For the
mitigation assessment,
the principal methods and approaches for evaluating mitigation
options for six sectors
and some possible results in the energy sector are provided. The
vulnerability
assessment is addressing the following sectors: agriculture, forestry,
and water
resources. The methodology for this assessment, which includes both
empirical-statistical
and simulation approaches, is described. Also, this article discusses
some
preliminary results and uncertainties of the vulnerability assessment
on the basis of
GCM-based climate change scenarios associated with increasing
CO2
concentrations in the atmosphere.
Unfavorable consequences of climate change in connection with anthropogenic increase of the concentrations of CO2 and other greenhouse gases in the atmosphere generate concern throughout the world and in Kazakhstan, too. The Republic of Kazakhstan is one of the 150 countries that signed the United Nations Framework Convention on Climate Change (UNFCCC). The Kazakhstan Government supports international cooperation on climate change issues.
The Republic of Kazakhstan covers 2,717,000 sq km with a population of over 17 million. The Republic consists of 19 regions, 220 districts, and over 80 cities and towns.
The Republic consumes about 10 million tons of coal, and its coal reserves are estimated at 39 billion tons. Oil production amounts to 26.6 million tons with the reserves estimated at about 2,357 million tons. There are also deposits of natural gas. The production of electric power exceeds 90 billion kWh including over 82 billion kWh produced by thermal stations. Per capita electric power consumption is 6,100 kWh annually.
The sowing areas of Kazakhstan exceed 35 million hectares including about three million hectares of irrigated land. The gross yield of grain is 25 to 30 million tons per year. Part of the grain is exported. The leading branch of animal husbandry is sheep breeding based on desert and semidesert pastures which occupy vast expanses. The number of cattle livestock is 9 million. Pigs, camels, and horses are also bred.
Forests in Kazakhstan occupy a small area, only 3.6 percent of the total territory (9,648,000 ha). Out of these, 1,800,000 ha are covered with coniferous forests, and the rest with leaf-bearing woods and shrubs. The largest portion, 4.7 million ha, is covered with saksaul.
The U.S.-Kazakhstan Project, "Kazakhstan Climate Change Study," was initiated on October 1, 1993. The main objectives of this project are to carry out the following:
Inventory of GHG Emission and Sinks The Intergovernmental Panel on Climate Change (IPCC), together with other international scientific organizations (United Nations Environmental Program, Organization for Economic Cooperation and Development, Global Environment Facility, International Energy Agency, etc.), has prepared the methodology that was used for this work (IPCC Draft Guidelines for National Greenhouse Gas Inventories, 1994). These guidelines provide for comparison and estimation of the authenticity of work obtained in different countries.
Information about yearly fuel consumption of all fuels for 1990 is the basis for the calculation of GHG (greenhouse gases) emission from HPSs (heat power station) and large boiler-houses. Also, the year 1990 conforms to the IPCC recommendation. This information was obtained from the documents of the State Statistical Accounts. The yearly fuel consumption is recorded in tons for every HPS and for their sources of supply (e.g., deposit, oil refineries, etc.). Knowing the percentage content of carbon dioxide in every fuel, we can determine the quantity of carbon dioxide burned by simple multiplication of this percent by the volume of consumed fuel.
However, the methodology of IPCC can be difficult to apply. This is possibly connected with the fact that some countries do not have as good an initial data base as Kazakhstan does. For comparison of our results with data of other countries and to use the IPCC software, we made our calculations with IPCC methods.
The methodology will consist of the calculations of the likely decrease of GHG emissions with fuel switching, the introduction of the new technologies, the increased use of renewable energy sources, more active use of CH4, and other options. The official state statistical data will be used. If these data are not available, the methods of balance accounts will apply. The latter are based on data which were obtained by the Scientific Research Institutes of Kazakhstan as well as on calculation of specific coefficients contained in the IPCC methodology.
The creation of scenarios of economic development will take into account the predicted assessments of the Departments of Trade Ministry, Kazakhstan Ministry of Economics (former State Planning Committee of Kazakhstan) and Kazakhstan Institute of Economics of Academy of Science. To carry out the mitigation analysis in the energy sector the ENPEP model‹ developed by Argonne National Laboratory‹ will be used.
The first step in the vulnerability assessment was the development of future climate change scenarios. Scenarios of the main climate change for Kazakhstan to asses both long-term (doubling CO2 in 2050‹ 2075) and short-term (2010, 2030) impacts were prepared. The climate change under a doubling of CO2 was obtained on the basis of General Circulation Models (GCM) outputs. We used outputs of three GCMs: the model of Canadian Meteorological Center (CCCM) (Boer et al., 1991), the model of Geophysical Fluid Dynamics Laboratory (GFDL) (Manage and Wetherald, 1987), and a transitional version of the same model (GFDL-Transient).
The near-term climate scenarios were obtained using the Probabilistic Forecast Model (PFM) (Gruza and Rankova, 1991) we adapted to Kazakhstan (Pilifosova, 1991) for the year 2010 and the results of GFDL‹T for 2030. Moreover, to evaluate future climate changes for comparison, a baseline climate scenario was used. This scenario represents the current climate for the base period 1951‹ 1980 without a warming trend in the baseline.
The assessment of vulnerability in some sectors was based on models developed in Kazakhstan. Thus, for the estimation of yield of agricultural crops and grasslands we used a nonstationary dynamic model that had been developed by KazNIGMI several years ago (Lebed and Belenkova, 1991). A similar model was employed for estimating the yield of potatoes (Glumova, 1988). An assessment of the vulnerability of sheep-breeding was carried out on the basis of an unfavorable weather conditions criterion for sheep productivity (Gulinova, 1988). A mathematical runoff model was applied in the water resources sector (Golubtsov et al., 1989).
In general, the main principle for application of these models to the vulnerability assessment was defined as follows. Different climate variables (air temperature, insolation, humidity, rainfall, etc.) were used as model input parameters. The simulation first was run under the current climate conditions in accordance with a baseline climate scenario. Then input climate parameters were changed according to the regional climate change scenarios and used in another simulation. The difference between these simulation outputs represented the changes in the yield of agricultural crops, grasslands, livestock (sheep), water resources, and forests, which occurred due to climate change impacts.
Obviously, in a number of cases the application of these models required the transition of climate parameters from one space-time averaging scale into another. For example, monthly means of air temperature and precipitation were obtained from the GCMs. But models of the yield of agricultural crops require the 10-day mean data. After the data adaptation and fitting of model parameters, analysis of the model sensitivity to different changes (e.g. incremental scenarios) of input climate parameters was conducted. For example, the estimation of vulnerability of crops and sheep when air temperature is increased by 0.5°C through 4.0°C was conducted. Input data included observed climatic data from numerous reference books, data on crops productivity of The State Committee on Statistics and some experimental data from agricultural fields.
In addition to models developed in Kazakhstan, we used the DSSAT (Decision Support System for Agrotechnology Transfer) model. DSSAT integrates the models which generally describe the development, growth and yield of crops on homogenous area of soil exposed to certain weather conditions. This system was useful for running and validating the models, for conducting sensitivity analysis, and for evaluating the variability and risks of different management strategies for a range of locations specified by soil and weather data. The CERES-Wheat model (Ritchie, Otter, 1985) from DSSAT developed by IBSNAT (International Benchmark Sites Agrotechnology Transfer) was used for spring and winter wheat productivity vulnerability assessment in Kazakhstan, which was based on the GFDL and CCCM scenarios.
The Holdridge Model was used for the assessment of vulnerability of forestry. We prepared the distribution of Holdridge life zones (Holdridge, 1967) under the current climate conditions as well as the maps of these zones for four climate change scenarios on the basis of GCM outputs for a doubling of CO2 levels in the atmosphere: GISS (Hansen et al., 1983), UKMO (Wilson and Mitchell, 1987), OSU (Shlesinger and Zhao, 1989), and GFDL. These GCM outputs were built into the Holdridge model.
The Holdridge model relates the current spatial distribution of vegetation to features of the climate system. The Holdridge classification is suitable for examining the broad-scale patterns of vegetation as they relate to climate and how changes in climate patterns may influence the suitability of a region to support different vegetation/forest types. However, this approach does not address vegetation processes per se and as such cannot be used to predict the temporal dynamics of species composition and stand productivity, features that are important in evaluating the potential impacts of environmental change on forest resources and conservation. In order to make up the maps of the territorial distribution of forest we chose the additional scheme, which connected the forests distribution with a precipitation and evapotranspiration (PET) model.
The results of calculations of annual CO2,
CH4, CO,
N2O, NOx and nonmethane volatile compounds
emissions are divided
into 13 groups:
As a result, both the summary emissions of all six GHGs for 1990
and the contribution
of
separate sources (or branches of industry) were defined. The
calculation results are
presented in Table 1. The
results are expressed in
gigagrams (Gg) in accordance with the IPCC. More
than 90
percent of all GHG emissions is, as expected, from CO2
(198,729 Gg or nearly
200
million ton/yr). Thus the per capita CO2 emission is over
11 tons/yr.
CO2 absorption from the atmosphere by forests of
Kazakhstan was estimated.
The calculations have shown that forests absorb up to 1,530 Gg/yr or
less than 2.55 of the
total
emissions.
The largest sources of CO2 emission are heat power
stations and district
boiler-houses (48.5 percent), residential boiler-houses and stoves
(17.2 percent),
internal combustion
engines (ICE) (12.9 percent), and enterprises of ferrous and
nonferrous metallurgy (5.2
percent). The
largest sources of NOx emission are ICEs (53.7 percent)
and heat power stations
(36.4
percent). The largest sources of CH4 emission are from
solid waste open dumps
and
wastewater treatment (49.5 percent) and from agriculture (27
percent). The largest sources
of CO
emissions are ICEs (67.8 percent), metallurgy (18.3 percent),
residential boiler-houses
and stoves
(3.2 percent). Data are presented in percentages of the total
emissions of the respective
gas.
In accordance with the IPCC guidelines, the estimation of the
initial data reliability
(uncertainty) was made. The most reliable are the data on heat
power stations, which give
49,5
percent of the total emission of CO2, 39 percent of the
total emission of
NOx, and 19 percent of the total emission of CO. The data
were obtained by the
analyses of CO and NOx contained in waste gases, and
CO2 was
obtained by the balance calculations. The probabilistic errors here do
not exceed 5
percent. In other
branches of industry the power registration data are not highly
accurate so that possible
errors are
within the limits of 20 percent. The most unreliable calculation
results are those
connected with ICE
(13 percent of CO2 emission, 53.7 percent of
NOx emission and 67
percent of CO emission).
As for the estimation of the authenticity of the data, the comparison
of our indices with
the data
reported by the State Statistical Committee showed that variations on
separate gases were
within the
limits of 5ѳ 20 percent. For example, the total NOx
emission from
stationary
sources in 1990 was 330 Gg in the State Statistical Committee data
but it was 314.7 Gg in
our
calculations. The emission indices for residential boiler-houses and
stoves are the most
unreliable,
but these emissions are not high.
In Table 2 the predicted data
of CO2
emission
for the period from 1991 to
2000 taking into account expert assessments and fuel consumption of
main branches of
economy are
shown. These data show that the total CO2 emission
volume for the decade from
Kazakhstan territory will be 1,582,000 Gg. At the same time
CO2 absorption by
forests is estimated at 40,000 Gg. Thus, the difference (without
consideration of
CO2 absorption by other reservoirs) will be 1,542,000 Gg.
The Aktubinsk HPS building that will use a steam-gas cycle and
produce 954 NW of power
and 6 billion kWh annual electric power production will be
completed in 2000. Similar
powerplants
with less capacity are expected to be put into operation in Uralsk and
Atyrau (in 2000-
2005). The
problem of replacing traditional steam turbine engines with gas
turbines in Uralsk
HPS‹ 1,
Atyrau HPS, Shimkent HPS‹ 1, HPS‹ 2, Jambyl HPS‹
3, Jambyl state district
electric power station (SDEPS) is being studied. When energy is
produced by steam-gas HPSs
the
CO2 emission will decrease by 11,988 Gg. The total
decrease of
CO2 emission from energy sources is estimated at 37,860
Gg.
Concerning the use of renewable sources, the most promising
project is the development
of a
wind-electrical station "Jungarskie vorota" with 300 megawatts
power and 900 billion kWh
annual
power production. In addition the wind-electric engines in the Chilick
corridor, Kurday
passage,
Jengiz-To, Derjavinka, and Mugojary are projected to be put into
operation. Also wind-
electric
engines with small capacity are planned for remote locations for
water pumping, heating
and
electricity generation. The decrease of CO2 emission
associated with renewable
energy options will be 4,627.2 Gg for 1996‹ 2020.
Our calculations based on the observed data in Kazakhstan show
that the rise of annual
average air temperature is 1°C/100 years and this is
approximately twice as much as
the mean
global rise of temperature. The analysis of prediction curves of
temperature and
precipitation with
the use of the PFM model has allowed us to conclude that the rise of
CO2
concentration in the atmosphere will cause an average rise of
aridness (the increase of
temperature
and decrease of precipitation) all over the region. The highest rise of
temperature will
be 6°C
in comparison with the mean temperature for 1951‹ 80. It is
expected to occur in the
cooler
half of the year in the North of Kazakhstan. For the rest of the
territory of the
Republic, an increase
in the temperature of 1‹ 3°C in the summer and 3‹
4°C in the winter is
expected by 2010.
There is a significant probability that the increase of
CO2 concentration
may
cause some increase in atmospheric precipitation in the south and
southwest, and an
increase of
aridness in the west and in the northeast Kazakhstan in the winter.
In the summer a
decrease of
precipitation of 20‹ 50 percent is expected for all of
Kazakhstan, except for the
western
regions.
The results of the preliminary analysis on the basis of our models
of crops
vulnerability made
for Western Kazakhstan show that the increase of air temperature
for the period of shoot
ripening of
spring wheat causes significant deterioration of the thermal regime
by 20‹ 50 percent
relative
to optimal conditions. In this case the forecast crops yield is expected
not to be above
0.22‹ 0.44 ton/ha. In comparison the spring wheat yield was
0,82 ton/ha in 1991.
Forestry According to 2xCO2 OSU scenario, the forest area
remains in its present-day
boundaries. The area of forest-steppe zone according to the IET
model is decreased along
the
Southern and Western boundary (50‹ 70 km).
The most optimistic scenario of the forestry is obtained according to
the climate change
scenario
based on the GISS model. It is the only one of four models which
predicts the increase of
suitable
areas for the growth of the forests due to a probable climate
warming. The boundary of
steppe-
forest-steppe is moved by 120‹ 180 km towards the south and
the west. According to
this scenario
the areas suitable for forests growth are increased 1.6‹ 1.8
times.
The impacts of the scenarios of GFDL and the OSU scenarios are
midway between the
impacts of
the GISS and the UKMO scenarios.
The central problem with the mitigation analysis is the cost
assessment of the mitigation options in view of the economical
declines, especially production declines and inflation. It is a
difficult challenge to predict the development of these processes
now.
There are two principal sources of uncertainties in the
vulnerability assessment. The first is associated with uncertainty of
the climate change scenarios particularly at a regional
level. It is
known that increased greenhouse gas emissions will likely raise
global temperatures and
precipitation; however, no reliable suggestions can be made about
their regional effect.
The second source of uncertainties arises from the imperfection of
the models used in
assessment of local conditions. The use of the DSSAT model demands
input parameters which do not correspond to our data. For example,
information about tillage, chemical composition of fertilization are
not available. We often are limited in availability of current
meteorological information for the input parameters of models. In
this case we have to use the Weather Generators, which do not take
into account the local climatic diversity of our regions. The DSSAT
model is also oriented for local fields, while we need to obtain
estimates for the whole Kazakhstan region.
Similar problems are connected with the use of the Holdridge
model. In mountain regions the vertical zonality is formed, which is
simulated poorly where the resolution of the database set (0.5*0.5
degrees) is small. Furthermore, such territories contain areas
(especially hollows and canyons) which exist due to additional water-
flow from the surrounding slopes. The result is that if the
evapotranspiration exceeds the precipitation, the vegetation is still
formed. Although this forest vegetation is of fragmentary character,
the total area of these territories may be considerable. Neither the
Holdridge model nor the PET model consider these specific conditions
which cause errors in vegetation classification.
Both models fail to predict the pine forests propagation due to the
fact that the
ordinary
pine is a
drought-resistant species under the current conditions of Kazakhstan.
It forms forests
when the
deficit of precipitation is 250mm or more. The ordinary pine grows
under such conditions
only on
the sands with good aeration, developing powerful root systems,
which can reach soil
waters. The
soil types are not considered in these models.
Therefore we tried to use the models worked out in KazNIGMI for
the vulnerability
assessment in
agriculture and water resources sectors. But of course, the models we
used have their own
advantages and disadvantages. Advantages of these models are their
good fitting for
geographic,
climate, and other peculiarities of the region of Kazakhstan and the
use of observed data
as inputs of
model. The major disadvantage is that they first were made for near-
term projections
(month, season,
timeframe) and then were modified for this vulnerability
assessment. So we have to
transform data
from one time-scale averaging to another. This introduced additional
uncertainties.
Boer, G.J., N. McFarlane, and M. Lazare. 1991. Greenhouse gas-
induced climatic
change
simulated
with the CCC second-generation GCM. Accepted for publication in the
J.Climate.
Golubtsov, V.V., V.I. Lee, and T.P. Stroeva. 1989. Simulation of
flow formation
processes
when
information is limited.
Proceedings of V All-Union hydrological symposium. 6. P. 374-382.
(in Russian).
Gruza, G.V., and E.Ya. Rankova. 1991. Holdridge, L.R. 1967, Life Zone Ecology, Tropical Science Center,
San Jose,
Costa Rica.
IPCC. 1993. Greenhouse Gas Inventory Reporting Instructions,
Final Draft,
Vol.1. IPCC/OECD
Joint Program.
IPCC. 1993. Greenhouse Gas Inventory Workbook, Final Draft,
Vol.2. IPCC/OECD
Joint
Program.
IPCC. 1993. Greenhouse Gas Inventory Reference Manual, First
Draft, Vol.3.
IPCC/OECD Joint
Program.
Manabe, S., and R.T. Wetherald. 1987. Large-scale changes in soil
wetness induced
by an
increase
in carbon dioxide. April, Atmos. Sci., 44: 1211-1235.
Pilifosova, O.V. 1992. Probabilistic of precipitation in the
Kazakhstan - Middle
Asia
region. Proceeding of KazNIGMI, 111: 64-72 p. (in Russian).
Ritchie, J.T.., and S. Otter. 1985. Description and performance of
CERES-Wheat: A
User-oriented
Wheat Yield Model. In: Willis W.O., ed. ARS Wheat Yield Project.
Washington D.C.: US
DOA,
Agricultural Research Service. Ars-38. p. 159-175.
Schlesinger, M.E., and Z.-C. Zhao. 1989. Seasonal climate changes
induced by doubled CO2 as simulated by the OSU
atmospheric GCM/mixed-layer ocean model. J.Climate.
Wilson, C.A., and J.F.B. Mitchel. 1987. A doubled CO2
climate sensitivity experiment with a global model including a simple
ocean. Journal of Geophys. Res., 92:13315-13343.
Estimates of emissions of nitrous oxide, carbon monoxide and
nonmethane volatile compounds
were
obtained from the records at the State Statistical Accounts. Emissions
of carbon dioxide,
methane,
and nitrous oxide have been determined by balance calculations
taking into account real
fuel
consumption, quantity of cattle, rice area, and other data. The
emission factors
recommended by
IPCC and regional institutes were used in the calculations. Values of
specific GHG
emissions of
internal combustion engines were obtained from the Kazakh
Scientific Research Institute of
Motor
Transport.
Vulnerability Assessment
As a result of the above described approaches, "optimistic" (GFDL)
and "pessimistic"
(CCCM)
scenarios under 2 x CO2 conditions, were defined. GFDL-
Transient outputs give
an
"intermediate" scenario. According to the GFDL model, a minimum
increase of temperature is
expected in summer, when most of territory will be 4‹ 5°
C warmer. The maximum
(about 8°C) is expected to be in the winter. The mean annual
temperature increase is
about
5°C. According to CCCM, the mean annual temperature increase
is 7°C and the
maximum is of 12°C in the spring. In most cases, the relative
changes of
precipitation will be
in the range of 0.8‹ 1.2 or 80‹ 120 percent (i.e., within
the normal limits).
Crops
To estimate the possible impacts of climate change on wheat
production in the main wheat
producing regions of Kazakhstan, the DSSAT model was used. The
DSSAT model combines the
CERES-wheat crop growth model under GFDL and CCCM scenarios. The
GFDL scenario shows
the spring and winter wheat yield increasing in Western and Central
Kazakhstan by
approximately
10 percent and 5 percent, respectively. However, in Northern
Kazakhstan the yield
decreases
approximately by 12 percent. According to the CCCM scenario, the
spring wheat yield would
decrease by 35 percent and the winter yield would not decrease
significantly. Note that
the yields
changes under the baseline scenario are about 2‹ 4 percent.
Grasslands
For the region located north of the Aral Sea the possible increase of
air temperature in
the vegetative
season of 2°C is accompanied by some increase in grassland
productivity (6‹ 20
percent). These increased temperatures may allow for a change in
precipitation in the cool
season of
30° 40 mm resulting in changes in feed productivity ranging
from -18 to + 12 percent.
A
considerable decrease of productivity up to 40‹ 50 percent
from existing level is
estimated
with temperature increases of 2 to 3°C. The possible climate
changes due to a
CO2 doubling scenario (e.g., GFDL model) may cause a
2‹ 3 times decrease in
feed productivity on Priaralie grasslands in the summer-autumn
period with some increase
in its
reserves in the spring.
Potato
The preliminary results on potato productivity were obtained for
five North Kazakhstan
regions. The
calculations of dynamics of dry potato biomass during the vegetation
season show the
potential for
considerable decrease of water storage in soil level. A 5mm decline
in water levels would
decrease
productivity by 5‹ 8 percent and a decrease of water storage of
20 mm causes
productivity
losses of 20‹ 26 percent.
Increasing the air temperature by 0.5°C decreases potato
productivity by 2‹ 3
percent.
An increase of air temperature by 2°C causes a productivity
decline by 6‹ 10
percent.
Sheep-breeding
In estimating vulnerability of sheep-breeding the data from
observations of sheep pasture
conditions
for 1959‹ 1990 and biometeorological parameters defined
earlier by other researchers
were
used. If the number of days in a ten day period with stable hot
weather (SHW) equal 6 or
more, a
decrease of sheep weight is observed. Such unfavorable hot periods
which repeat one after
another
and form a whole period with SHW are being currently observed in
the South and East-South
of
Kazakhstan. An increase of air temperature of 1°C in May and
June causes an increase
of the
average SHW duration of 3 to 6 days. The SHW duration increases
slightly less (by 2 to 4
days) with
rising air temperature in August and September. If temperatures in
both periods increase
at the same
time by 2°C, the average duration of SHW periods will increase
by almost two weeks.
Changes in atmospheric precipitation in the summer months do not
significantly change the
duration
of the SHW period.
Having analyzed the discrepancies obtained by the Holdridge (PET)
model we have calculated
the
forest and forest-steppe zones in correspondence with the 2°
CO2 climate
conditions predicted by 4 models (GISS, UKMO, GFDL, OSU). The most
pessimistic results
were
obtained using the UKMO model. According to this model only the
northern part of the
Republic
(the stripe with the width of 70‹ 150 km located along the
Northern boundary) remains
a forest-
steppe zone. The area suitable for forest growth according to UKMO
model is expected to be
15
percent of that for current climate.
DISCUSSION
Conducting our work on the Kazakhstan Climate Change Study
Project, we came across a
number of
uncertainties and problems. One of them is related to the assessment
of future GHG
emission in our
country (Table). At the present
time it is difficult to predict reliably the volume of GHG emissions for
the period from 1991 to 2000. The state authorities in Kazakhstan
have changed after the USSR was split up, specifically authorities
such as the State Planning Committee have been dismantled. For 60
years from 1927, the planning and development of the national
economy (in which the share of the private sector was minimal) was
led by the government of the USSR in accordance with confirmed
five- or seven-year plans. There are no such plans at the
Departments and the role of private sector productive forces
increases while that of state-owned sector decreases. As a result of
that, it is impossible to receive any information from government
offices. That is why we had to rely on expert evaluations.
CONCLUSIONS
The main conclusions of this work are as follows:
REFERENCES
March 1995
