Journal of Atmospheric Science Research | Vol.3, Iss.1 January 2020

Editor-in-Chief
Dr. José Francisco Oliveira Júnior
ICAT/ UFAL, Brazil
Editorial Board Members
Fan Ping, China
Marko Ekmedzic, Germany
Xuezhi Tan, China
Hirdan Katarina de Medeiros Costa, Brazil
Chuanfeng Zhao, China
Suleiman Alsweiss, United States
Aditi Singh, India
Boris Denisovich Belan, Russian Federation
Perihan Kurt-Karakus, Turkey
Hongqian Chu, China
Isidro A. Pérez, Spain
Mahboubeh Molavi-Arabshahi, Iran
Tolga Elbir, Turkey
Junyan Zhang, United States
Thi Hien To, Vietnam
Jian Peng, United Kingdom
Zhen Li, United Kingdom
Anjani Kumar, India
Bedir Bedir Yousif, Egypt
assan Hashemi Hassan Hashemi, Iran
Mengqian Lu, Hong Kong
Lichuan Wu, Sweden
Raj Kamal Singh, United States
Zhiyong Ding, China
Elijah Olusayo Olurotimi, South Africa
Jialei Zhu, United States
Xiying Liu, China
Naveen Shahi, South Africa
Netrananda Sahu, India
Luca Aluigi, Università di Modena e Reggio Emilia
Daniel Andrade Schuch, Brazil
Vladislav Vladimirovich Demyanov, Russian Federation
Jingsong Li, China
Priya Murugasen, India
Nathaniel Emeka Urama, Nigeria
Barbara Małgorzata Sensuła, Poland
Service Opare, Canada
Che Abd Rahim Bin Mohamed, Malaysia
Maheswaran Rathinasamy, India
Masoud Rostami, Germany
Osvaldo Luiz Leal De Moraes, Brazil
Ranis Nail Ibragimov, United States
Masoud Masoudi, Iran
Pallav Purohit, Austria
B. Yashwansingh Surnam, Mauritius
Alexander Kokhanovsky, Germany
Lucas Lavo Antonio Jimo Miguel, Mozambique
Nastaran Parsafard, Iran
Sarvan Kumar, India
Abderrahim Lakhouit, Canada
B.T. Venkatesh Murthy, India
Olusegun Folarin Jonah, United States
Amos Apraku, South Africa
Foad Brakhasi, Iran
Debashis Nath, India
Chian-Yi Liu, Taiwan
Mohammad Moghimi Ardekani, South Africa
Yuzhu Wang, China
Zixian Jia, France
Md. Mosarraf Hossain, India
Prabodha Kumar Pradhan, India
Tianxing Wang, China
Bhaskar Rao Venkata Dodla, India
Lingling Xie, China
Kazi Sabiruddin, India
Nicolay Nikolayevich Zavalishin, Russian Federation
Xizheng Ke, China
Alexander Ruzmaikin, United States
Peng Si, China
Zhaowu Yu, Denmark
Manish Kumar Joshi, United Kingdom
Aisulu Tursunova, Kazakhstan
Enio Bueno Pereira, Brazil
Samia Tabassum, Bangladesh
Donglian Sun, United States
Zhengqiang Li, China
Haider Abbas Khwaja, United States
Haikun Zhao, China
Wen Zhou, Hong Kong
Suman Paul, India
Katta Vijaya Kumar, Sri Venkateswara University
Mohammed Adnane Douar, Algeria
Chunju Huang, China
Habibah Lateh, Malaysia
Meng Gao, China
Bo Hu, China
Akhilesh Kumar Yadav, India
Archana Rai, India
Pardeep Pall, Norway
Upaka Sanjeewa Rathnayake, Sri Lanka
Yang Yang, New Zealand
Somenath Dutta, India
Kuang Yu Chang, United States
Sen Chiao, United States
Mohamed El-Amine Slimani, Algeria

Dr. José Francisco Oliveira Júnior
Editor-in-Chief
Journal of
Atmospheric Science
Research
Volume 3 Issue 1· January 2020 · ISSN 2630-5119 (Online)

On the Formation of a Bead Structure of Spark Channels during a Discharge in Air at
Atmospheric Pressure
Victor Tarasenko Dmitry Beloplotov Alexander Burachenko Evgenii Baksht
Rainfall Estimation using Image Processing and Regression Model on DWR Rainfall Prod-
uct for Delhi-NCR Region
Kuldeep Srivastava Ashish Nigam
Behavior of the Cultivable Airborne Mycobiota in air-conditioned environments of three
Havanan archives, Cuba
Sofía Borrego Alian Molina
Thumb Rule for Nowcast of Dust Storm and Strong Squally Winds over Delhi NCR using
DWR Data
Kuldeep Srivastava
Volume 3 | Issue 1 | January 2020 | Page 1-39
Journal of Atmospheric Science Research
Article
Contents
Copyright
Journal of Atmospheric Science Research is licensed under a Creative Commons-Non-Commercial 4.0 International
Copyright (CC BY- NC4.0). Readers shall have the right to copy and distribute articles in this journal in any form in any
medium, and may also modify, convert or create on the basis of articles. In sharing and using articles in this journal, the
user must indicate the author and source, and mark the changes made in articles. Copyright © BILINGUAL PUBLISH-
ING CO. All Rights Reserved.
1
9
16
29

1
Journal of Atmospheric Science Research | Volume 03 | Issue 01 | January 2020
Distributed under creative commons license 4.0 DOI: https://doi.org/10.30564/jasr.v3i1.1858
https://ojs.bilpublishing.com/index.php/jasr
ARTICLE
On the Formation of a Bead Structure of Spark Channels during a
Discharge in Air at Atmospheric Pressure
Victor Tarasenko*
Dmitry Beloplotov Alexander Burachenko Evgenii Baksht
Institute of High Current Electronics, Siberian Branch (SB), Russian Academy of Sciences (RAS), 2/3 Akademicheskii
Ave., Tomsk, 634055, Russia
ARTICLE INFO ABSTRACT
Article history
Received: 6 May 2020
Accepted: 22 May 2020
Published Online: 31 June 2020
The conditions for the formation of spark channels with a bead structure in
an inhomogeneous electric field at different polarities of voltage pulses are
studied. Voltage pulses with an amplitude of up to 150 kV and a rise time of
≈1.5 µs were applied across a 45-mm point-to-plane gap. Under these con-
ditions, spark channels consisting of bright and dim regions (bead structure)
were observed. It is shown that when current is limited, an increase in the
rise time and the gap length does not affect the formation of the bead struc-
ture. It was found that an increase in the amplitude of voltage pulses leads
to an increase in the length of beads. The appearance of the bead structure
is more likely at negative polarity of the pointed electrode. The formation
of spark channels was studied with a four-channel ICCD camera.
Keywords:
Discharge in air
Formation of sparks
Bead structure
ICCD camera
Point-to-plane gap
Bead lightning
*Corresponding Author:
Victor Tarasenko,
Institute of High Current Electronics, Siberian Branch (SB), Russian Academy of Sciences (RAS), 2/3 Akademicheskii Ave., Tomsk,
634055, Russia;
Email: VFT@loi.hcei.tsc.ru
1. Introduction
I
n recent years, interest in studying atmospheric dis-
charges has increased significantly, see, for example,
[1–7]
. Attempts are also being made to reproduce atmo-
spheric discharges under laboratory conditions [8-20]
. This is
facilitated by the improvement of equipment for recording
fast processes and the development of various models of
discharges. Interesting results on the observation of blue
jets and red sprites in the upper atmosphere are presented
in [1,2,5-7]
. So, in [1]
, images obtained from the international
space station are presented. They demonstrate the appear-
ance of blue jets over an area with thunderstorm activity.
Images of red sprites observed at altitudes of up to 100
km are given in [2]
. Lightning and lightning protection was
studied in [3,4]
.
Bead lightning is one of the rarest and insufficiently
studied phenomena [8-10]
. In [8]
, it is noted that many re-
searchers deny the very existence of such types of dis-
charges. However, in recent years, new data on the devel-
opment of bead lightning under conditions close to nature
[11]
and on the observation of its analogue in laboratory
spark discharges[12-16]
were obtained.
In [11]
, bead lightning was observed in an initiated at-
mospheric discharge. The discharge was shot at with an
exposure time of 1 ms. It was found that at the first stage,
an uniform bright channel is observed. Further individ-

2
Distributed under creative commons license 4.0
ual beads are appeared in the channel. Then, when the
brightness of lightning decreased, dim regions between
the beads were clearly visible. The length of one bead was
about 50 cm under these conditions. After a return stroke,
the glow of lightning channel again became uniform and
intense.
In [12]
, bright filaments (unformed spark) in the center of
an 18-mm point-to-plane gap filled with air were observed
on the background of emission of a diffuse discharge. The
amplitude of voltage pulses was ≈200 kV. Such unformed
spark channels were observed in pulse-periodic discharg-
es, when nanosecond voltage pulses with amplitudes of
10–15 kV were applied across a 6-mm point-to-plane gap.
It was assumed that the discharge observed is similar to
bead lightning.
An analogue of bead lightning was also observed in a
spark discharge in a gap of several meters in length [13]
. At
the beginning, loops formed by several thin channels were
observed in the discharge channel. The radiation intensity
of the loops decreased faster than the intensity of a single
channel due to the rapid cooling of the loops. This led to
the formation of the bead structure. It was suggested that
under natural conditions, there are loops in the lightning
channel, and as a result, when lightning fades, the bead
structure is observed. Note that X-ray radiation was re-
corded during such spark discharges [17,18]
.
The results of our studies, begun in [12]
, are presented
in [14-16]
. In new studies, a four-channel ICCD camera was
used. In air and nitrogen, spark channels with the bead
structure were observed. No loops were observed in [12,14-
16]
, and a diffuse discharge preceded the spark stage. In
the prebreakdown stage of these discharges, the current of
a runaway electron beam passing through the foil anode
was measured. Collective monographs [21-23]
are devoted
to the study of runaway electron generation in laboratory
discharges. In [15,16]
, the length of a gap did not exceed 8.5
mm, and the rise time of voltage pulses was 200 ns.
This work presents the results of studies of the condi-
tions for the formation of the bead structure during spark
discharges in a point-to-plane gap with a length of up to
45 mm. Voltage pulses of both polarities with a rise time
of the order of a microsecond were applied across the gap.
2. Experimental Setups and Measurement Technique
Two experimental setups were used. A home-made
generator (Figure 1) based on a pulse transformer that
produces voltage pulses with an amplitude of up to 200
kV with a step front was used in the first setup.
Figure 1. Experimental setup 1
Voltage pulses were applied to a cylindrical conductor
with a diameter of 7.5 cm and a length of 14 cm. A 7-cm
long electrode (needle or cone) with a small radius of cur-
vature were installed at the end of the cylindrical conduc-
tor. A cone electrode had a base diameter of 5 mm, a cone
angle of 68°, and a radius of curvature of ≈0.1 mm. The
second electrode was made of a needle that had the base
diameter of 3 mm, the cone angle of 36°, and the radius
of curvature of ≈0.05 mm. Both electrodes were made of
a stainless steel. The opposite electrode was flat. It was
connected to the grounded case of the generator through
a current shunt made of TVO resistors. The interelectrode
distance d was 45 mm.
The generator produced pulses of both negative and
positive polarity. The rise time of the voltage pulse, which
had two steps, was ≈1.5 μs. Waveforms of negative volt-
age pulses in idle mode and during the formation of a dif-
fuse discharge are shown in Figure 2.
Figure 2. Waveforms of negative voltage pulses in (a) idle
mode and (b) during the formation of a diffuse discharge.
Setup 1
The amplitude of the voltage pulses varied due to a
change in the charging voltage of a capacitor C1 (65 nF)
in a primary circuit of the transformer in the range of
7–10 kV. Voltage was measured using a resistive voltage
divider. Signals from the current shunt and the voltage
divider were recorded on an ТDS-2020 oscilloscope
(300 MHz, 5 GS/s).
In the second setup, a high-voltage generator was also
home-made-produced. It formed voltage pulses with an
amplitude of up to 36 kV and with a rise time of 0.2 μs.
DOI: https://doi.org/10.30564/jasr.v3i1.1858

3
In this work, in contrast to the previous ones [15,16]
, the
experiments were carried out not only with negative po-
larity, but also with positive one. The full width at half
maximum (FWHM) of voltage pulses in the idle mode
was ≈300 ns, and the rise time was ≈200 ns. The generator
was connected to a discharge gap via a 60-cm long coax-
ial cable with a wave impedance of 75 Ω. A high-voltage
electrode was made of a sewing needle. The electrode
length, diameter, and curvature radius of the tip were
5 mm, 1 mm, and 75 μm, respectively. The opposite elec-
trode was flat. The interelectrode distance d was 8.5 mm.
A discharge chamber was equipped with a capacitive
voltage divider and a current shunt. When using the mesh
anode and collector, it was possible to measure a current
of runaway electrons [24]
. A collector could be installed
instead of the current shunt for measuring the current of
runaway electrons that came out of the gap at negative po-
larity of the high-voltage electrode.
On the setup 2, the dynamics of the development of a
discharge was studied using a four-channel ICCD camera.
The block diagram of the setup 2 is presented in Figure 3.
Figure 3. Experimental setup 2
On the setup 2, the current shunt was made of SMD
resistors. Unlike TVO resistors, they are more broadband.
Signals from the capacitive voltage divider, the current
shunt as well as the clock signal from the first channel
of the ICCD camera were recorded on the Tektronix
TDS3054B oscilloscope. As a result, this made it possible
to synchronize the ICCD images with the waveforms of
voltage and discharge current.
Gas discharge chambers in all setups were filled with
air at a pressure of 100 kPa. The generators operated in a
single pulse mode. Images of the discharge plasma emis-
sion were taken with a Sony A100 digital camera.
3. Experimental Results
3.1 Setup 1
In contrast to previous studies on discharges with the bead
structures [12,14-16]
, in this study, experiments were carried
out at applying microsecond voltage pulses (Figure 2).
The gap length was 45 mm instead of 8.5 mm or less in
[12,14-16]
. Breakdown voltage and discharge form varied
from pulse to pulse due to the instability of discharge
initiation. This is typical for many types of discharge, in-
cluding both lightning [2-4]
and sparks in meter-long gaps
[17-18]
. The image of the cone-to-plane gap and the images
of a discharge plasma emission in this gap are presented
in Figures 4 and 5.
Figure 4. Image of the cone-to-plane gap and images of a
discharge plasma emission. Setup 1. Negative polarity
Negative voltage pulses were applied across the gap
with d = 45 mm. Bends of the spark channel are observed
in Figure 4b. Similar bends were described in [25]
, where
nanosecond voltage pulses were applied across a point-
to-plane gap. The spark leader, which has not crossed the
gap, and the spark channel are observed in Figure 4c. At
the same time, the spark leader apparently closed on the
spark channel. This area is characterized by a diffuse glow
(Figure 4c). A break in the spark channel and diffuse glow
are observed in Figure 4d.
The obtained images allow us to make the assumption
that the spark leader can transform into a diffuse channel
at negative polarity. In general, the formation of a diffuse
discharge at atmospheric pressure of various gases is pro-
vided by the preionization of the gas by runaway electrons
generated in a high electric field [21-24,26]
.
A zigzag spark channel and a spark channel with a bead
structure are observed in Figs 5a and 5b, respectively.
Figure 5. Images of the discharge plasma emission in the
cone-to-plane gap. Setup 1. Negative polarity

4
The brightness of the zigzag spark channel is not uni-
form. A pronounced bead structure is observed when two
spark leaders cross the gap. Such discharge implementa-
tions were quite rare. The largest current flows through the
bright channel. The spark channel with the bead structure
is characterized by the presence of diffuse regions.
The image of the needle-to-plane gap with the same d
and the images of a discharge plasma emission in this gap
are presented in Fig 6. A diffuse discharge (Figure 6b) was
observed in a number of implementations when the needle
electrode was used instead of the conical one. The needle
electrode has a larger enhancement of the electric field
strength due to the smaller radius of curvature.
Figure 6. Image of the needle-to-plane gap and images of
a discharge plasma emission. Setup 1. Negative polarity
A bright spark leader that did not cross the gap per pulse
is observed against the background diffuse emission in
Figure 6b. In [12,14–16]
, the formation of a spark channel with
a bead structure followed the diffuse stage of a discharge.
The diffuse discharge stage was observed in experiments on
setup 1 at breakdown delay times of an order of magnitude
more than in [15-16]
. Spark channels of various form (linear
and zigzag) with a bead structure are observed in Figs 6c
and 6d. It is seen that the brightness periodically changes
along the channel length. These images were obtained un-
der conditions when the breakdown occurred earlier than
on average. The length of individual beads is longer than
that observed in [15,16]
at breakdown voltages of tens of kV.
The structure of spark channels was changed when
voltage pulses of positive polarity were applied across the
gap (Figure 7).
Figure 7. Image of a discharge plasma emission in (a)
cone-to-plane and (b) needle-to-plane gaps. Setup 1. Posi-
tive polarity
The spark channel at positive polarity of cone or nee-
dle electrodes was often single, diffuse and had many
bends. The beads could only be observed from the side of
the grounded flat electrode. They had less brightness and
length. Spark channels with bead structure over the entire
length of the discharge gap were not observed in any of
the order of hundreds of implementations. Note that the
bead structures in [15,16]
were observed only at negative po-
larity of an electrode with a small radius of curvature.
In the experiments, the waveforms of voltage and
discharge current were also recorded. The waveforms of
voltage and current in idle mode, as well as during diffuse
discharge (Figure 6b) are presented in Figure 2. The cur-
rent through the gap was absent in idle mode, as it should
be. When diffuse discharge occurred, a current pulse with
an amplitude of ≈3.5 A was observed on the falling edge
of the voltage pulse. In those cases, when a spark dis-
charge with and without bead structure were observed, the
breakdown occurred earlier. The waveforms of voltage
and discharge current at negative and positive polarities
when sparks were observed (Figure 5b and Figure 7a,
respectively) are presented in Figure 8.
Figure 8. Waveforms of voltage and discharge current at
both polarities when sparks are formed. Setup 1
It is seen that the breakdown occurred 1–1.3 µs after
applying the voltage pulse across the gap. In this case,
typical for spark discharges, a rapid voltage drop due to
the high conductivity of a spark channel is observed.
3.2 Setup 2
The development of a discharge with spark channels
having a bead structure was studied on setup 2 using the
four-channel ICCD camera with a minimum exposure
time of 3 ns. Similar studies with negative polarity were
carried out in [15,16]
. It was shown that in a point-to-plane
gap filled with air at a pressure of 100 kPa, a diffuse dis-
charge is first formed, and then a spark channel consist-
ing of separate beads is formed. Channels with the bead
structure were observed in each pulse. Their number and
length varied from pulse to pulse. No studies were carried

5
out at positive polarity.
Figure 9 shows ICCD images of discharge develop-
ment as well as corresponding waveforms of voltage and
current obtained on setup 2 at negative polarity.
Figure 9. (a) ICCD images of a discharge in the point-
to-plane gap filled with air at a pressure of 100 kPa. C
– cathode, A – anode. (b, c) Waveforms of voltage and
discharge current pulses. Negative polarity. Setup 2
It is seen that at the initial stage a diffuse discharge
is formed (Figure 9a). The discharge formation time
did not exceed 1.5 ns. In this case, in order to study the
initial stage of the discharge, the ICCD camera channels
was switched on before the breakdown. The discharge
formation time was determined from the waveforms of
current. As shown in our previous paper [27]
the formation
of a streamer is accompanied by the flow of a current,
which we call the dynamic displacement current (DDC).
The fact is that the formation of a streamer is accompa-
nied by a redistribution of the electric field strength in the
gap. A time-varying electric field induces a displacement
current. The magnitude of DDC depends on the streamer
velocity and therefore has characteristic features that are
easy to find on the waveforms of current. DDC increases
sharply when a streamer starts and when it approaches the
opposite electrode. These features are clearly distinguish-
able in Figure 9c, and the corresponding time interval is
designated as streamer propagation. Such streamers with
a large diameter (Figure 9a) are typical for nanosecond
breakdown of point-to-plane gaps [19,27,28]
.
The spark formation lasted several tens of ns (Figure 9a,
spark formation). Under the experimental conditions, the
length of beads and their number changed from pulse to
pulse. The maximum number of beads reached 8, as in
[15,16]
. The position of beads in space can also vary from
pulse to pulse. In general, these experiments confirmed
the stable formation of bead structures of the spark chan-
nel at negative polarity.
When polarity was changed to positive, the discharge
slightly changed. The corresponding ICCD images and
waveforms of voltage and current are presented in Fig-
ure 10.
Figure 10. (a) ICCD images of a discharge in the point-
to-plane gap filled with air at a pressure of 100 kPa. C
– cathode, A – anode. (b, c) Waveforms of voltage and
discharge current pulses. Positive polarity. Setup 2
It is seen that at the initial stage a diffuse discharge is
also formed (Figure 9a). The change in polarity did not
have a qualitative effect on the formation of a diffuse dis-
charge in air. However, the discharge formation time in-
creased up to 2 ns. This means that the average velocity of
a positive streamer was less than that of the negative one.
4. Discussion
The studies showed that an increase in the gap length
from 8.5 to 45 mm and an increase in the rise time of volt-
age pulses from 0.2 µs to 1 µs did not affect the formation
of bead structures of spark channels in air at atmospheric
pressure in an inhomogeneous electric field. We assume
that the bead structure forms due to changes in the electric

6
field strength in the spark leader head that is caused by
changes in its size.
It is known (see, for example, [21,26]
) that, in air at at-
mospheric pressure, a diffuse discharge is formed in gaps
with an inhomogeneous electric field due to runaway elec-
trons. As was found in [12,14–16]
and confirmed in the pres-
ent work, beads are formed when current decreases in the
diffuse stage of the discharge. With a sufficient pulse du-
ration, beads are “smoothed” when a strong current flows.
The probability of the appearance of beads, their length
and quantity, as well as the dynamics of their formation
vary from pulse to pulse and depend on experimental con-
ditions: the length of beads and the distance between them
on the setup 1 were greater than on the setup 2.
At the beginning, a diffuse discharge is formed (Figure
6b, Figs 9 and 10) due to the development of a streamer
or several streamers (an ionization wave) [19,24,27]
. At high
overvoltages, the diameter of the streamer can be compa-
rable with the distance between the electrodes [19]
. This is
common for nanosecond discharges in an inhomogeneous
electric field. This is ensured by the generation of fast (with
energies of hundreds of eV - units of keV) and runaway
(with energies of tens – hundreds of keV) electrons that
preionize the gas ahead a streamer [28].
As shown in this
work, a discharge forms in diffuse form under conditions
when microsecond voltage pulses are applied across gaps.
There is no data on the formation of a diffuse discharge
during the development of lightning in the Earth’s atmo-
sphere. We assume that during the development of light-
ning a diffuse discharge can form in the vicinity of the
leader due to runaway electrons, as well as due to cosmic
rays, which produce preliminary ionization of air. Ioniza-
tion of air by cosmic rays is a long-established fact, see,
for example, the monograph [29]
. X-ray radiation caused by
runaway electrons in lightning was detected experimental-
ly using sensors mounted on an airplane [30]
.
At the stage of discharge constriction, the appearance
of a clot of plasma with a high concentration of electrons
and ions is necessary to start the bead formation processes.
It can be a cathode spot and a spark leader or a negative
step leader, which is responsible for the formation of the
lightning channel [31]
under natural conditions. The electric
field is redistributed and concentrated in the vicinity of the
leader head (spark leader head). In a high electric field,
some electrons can go into runaway mode. They can en-
sure the formation of a diffuse region in front of the leader
or improve uniformity and increase the diameter of the
channel. The diffuse region ‘screens’ the tip of the leader
(bead) due to the redistribution of the electric field. The
electric field strength at the front of the leader decreases,
the number of high-energy (fast and runaway) electrons
decreases or they disappear completely. The electric field
strength at the front of the diffuse region is also small
because of its relatively large diameter. In addition, the
conductivity of the diffuse channel is generally less than
that of the spark channel or the channel formed by the
leader due to the lower electron concentration. Then, a
narrow channel forms from the front of the diffuse region.
Constriction provides heating of this region. As a result, a
bead is formed. The electric field strength at the front of
this narrow channel increases again due to the geometric
factor. The process is then repeated. A new high-energy
electron generation cycle and the formation of a diffuse
region are taking place. In laboratory conditions, a se-
quence of beads having a weak radiation intensity that do
not reach the opposite electrode is often observed [16]
. A
periodic stop of the leader is observed in spark discharges
in large gaps with a negative rod electrode [3]
, as well as
during lightning development[31]
.
With sufficient duration and magnitude of current, the
brightness of the channel can be aligned and the bead
structure disappears. The bead structure can exist for a
long time if a shunt spark channel appears. In atmospheric
discharges, bead lightning is very rare [8]
. It is likely that
the bead structure of lightning disappears due to return
stroke, during which the main current flows [11,31]
. We as-
sume that bead lightning can be observed under conditions
when several channels develop, as well as at relatively
low magnitudes of current.
5. Conclusions
The spatial structure of discharges formed in an inhomo-
geneous electric field at different polarities and durations
of voltage pulses was studied in air at atmospheric pres-
sure at gap width of up to 4.5 cm. At negative polarity of
the electrode with a small radius of curvature, spark chan-
nels with a bead structure similar to bead lightning were
observed: images taken with a digital camera showed that
there are bright and dim regions along spark channel. The
emission of dim regions was similar to that of a diffuse
discharge, and the emission of bright ones was similar to
that of a spark discharge.
Using a four-channel ICCD camera, it was possible to
observe the development of such structures. It was found
that the formation of the spark channel begins from the re-
gion of the electrode spot, which is characterized by a high
concentration of ions and electrons as well as a high tem-
perature. However, the spark channel is non-uniform. The
dim regions follow the bright ones. The results of this work
confirm the hypothesis expressed in [15,16]
about the effect of
electrons in runaway mode on the formation of inhomoge-
neities in lightning channel during its development.

7
Acknowledgments
The work is performed in the framework of the State task
for HCEI SB RAS, project #13.1.4.
References
[1] Chanrion, O., Neubert, T., Mogensen, A., Yair, Y.,
Stendel, M., Singh, R., & Siingh, D.. Profuse activity
of blue electrical discharges at the tops of thunder-
storms. Geophysical Research Letters, 2017, 44(1):
496-503.
https://doi.org/10.1002/2016GL071311
[2] Füllekrug, M., Mareev, E. A., & Rycroft, M. J. (Eds.)..
Sprites, elves and intense lightning discharges.
Springer Science & Business Media, 2006, 225.
https://www.springer.com/gp/book/9781402046278
[3] R. Zeng, C. Zhuang, X. Zhou, S. Chen, Z. Wang, Z.
Yu, J. He. Survey of recent progress on lightning and
lightning protection research. High Voltage, 2016,
1(1): 2-10.
http://dx.doi.org/10.1049/hve.2016.0004
[4] W. Lu, Q. Qi, Y. Ma, L. Chen, X. Yan, Rakov, V. A.,
Y. Zhang. Two basic leader connection scenarios
observed in negative lightning attachment process.
High voltage, 2016, 1(1): 11-17.
https://doi.org/10.1049/hve.2016.0002
[5] Pasko, V. P., George, J. J.. Three-dimensional model-
ing of blue jets and blue starters. Journal of Geophys-
ical Research: Space Physics, 2002, 107(A12).
https://doi.org/10.1029/2002JA009473
[6] J. K. Chou, Hsu, R. R., H. T. Su, A. B. C. Chen, C. L.
Kuo, S. M. Huang, Y. J. Wu. ISUAL-Observed Blue
Luminous Events: The Associated Sferics. Journal of
Geophysical Research: Space Physics, 2018, 123(4):
3063-3077.
https://doi.org/10.1002/2017JA024793
[7] F. Liu, B. Zhu, G. Lu, Z. Qin, J. Lei, K. M. Peng, M.
Ma. Observations of blue discharges associated with
negative narrow bipolar events in active deep con-
vection. Geophysical Research Letters, 2018, 45(6):
2842-2851.
https://doi.org/10.1002/2017GL076207
[8] Barry J. D. Ball Lightning and Bead Lightning. New
York: Plenum Press, 1980.
https://www.springer.com/gp/book/9780306402722
[9] M.A. Uman, V.A. Rakov, Lightning Physics and Ef-
fects, Cambridge University Press, 2003.
https://doi.org/10.1007/s10712-004-6479-9
[10] Vernon Cooray, An Introduction to Lightning,
Springer, 2015.
https://doi.org/10.1007/978-94-017-8938-7
[11] G.O. Ludwig, M.M.F. Saba. Bead lightning forma-
tion. Phys. Plasmas, 2005, 12: 093509.
http://dx.doi.org/10.1063/1.2048907
[12] V.F. Tarasenko, D.V. Beloplotov, E.H. Baksht, A.G.
Burachenko, M.I. Lomaev. Analogue of bead light-
ning in a pulse discharge iniated by runaway elec-
trons in atmospheric pressure air. Atmospheric and
Oceanic Optics, 2015, 28(591).
https://doi.org/10.1134/S1024856015060160
[13] S.P.A. Vayanganie, V. Cooray, M. Rahman, P. Hetti-
arachchi, O. Diaz, M. Fernando. On the occurrence
of “bead lightning” phenomena in long laboratory
sparks. Phys. Lett. A, 2016, 380: 816.
https://doi.org/10.1016/j.physleta.2015.12.039
[14] V. F. Tarasenko, D.V. Beloplotov. Formation of Min-
iature Analogs of Bead Lightning in Nitrogen and
Air during Pulsed Discharge in Nonuniform Electric
Field. Atmospheric and Oceanic Optics, 2018, 31:
400.
https://doi.org/10.1134/S1024856018040164
[15] Beloplotov, D. V., Tarasenko, V. F.. Formation of
a small ‘bead lightning’in a half-microsecond dis-
charge in air. Physics Letters A, 2019, 383(4): 351-
357.
https://doi.org/10.1016/j.physleta.2018.11.004
[16] D. V. Beloplotov, A. M. Boichenko, V. F. Tarasenko
Beaded Discharges Formed under Pulsed Break-
downsof Air and Nitrogen // Plasma Physics Reports,
2019, 45(4): 387–396.
https://doi.org/10.1134/S1063780X19030012
[17] Kochkin, P. O., Nguyen, C. V., van Deursen, A. P.,
Ebert, U.. Experimental study of hard x-rays emitted
from metre-scale positive discharges in air. Journal of
Physics D: Applied Physics, 2012, 45(42): 425202.
https://doi.org/10.1088/0022-3727/45/42/425202
[18] Kochkin, P., Köhn, C., Ebert, U., van Deursen, L..
Analyzing x-ray emissions from meter-scale negative
discharges in ambient air. Plasma Sources Science
and Technology, 2016, 25(4): 044002.
https://doi.org/10.1088/0963-0252/25/4/044002
[19] V.F. Tarasenko, G.V. Naidis, D.V. Beloplotov, I.D.
Kostyrya, N.Yu. Babaeva. Formation of Wide
Streamers during a Subnanosecond Discharge in
Atmospheric-Pressure Air. Plasma Phys. Rep. 2018,
44(8): 746.
https://doi.org/10.1134/S1063780X18080081
[20] Tarasenko, V. F., Sosnin, E. A., Skakun, V. S.,
Panarin, V. A., Trigub, M. V., Evtushenko, G. S..
Dynamics of apokamp-type atmospheric pressure
plasma jets initiated in air by a repetitive pulsed dis-
charge. Physics of Plasmas, 2017, 24(4): 043514.
https://doi.org/10.1063/1.4981385
[21] Runaway Electrons Preionized Diffuse Discharges. /

8
Editors: V.F. Tarasenko. Published by Nova Science
Publishers, Inc. New York. USA, 2014: 598.
[22] Generation of runaway electron beams and X-rays
in high pressure gases. Editors: V.F. Tarasenko. Pub-
lished by Nova Science Publishers, Inc. New York.
USA, 2016, 1: 405.
[23] Generation of runaway electron beams and X-rays
in high pressure gases. Editors: V.F. Tarasenko. Pub-
lished by Nova Science Publishers, Inc. New York.
USA, 2016, 2: 333.
[24] Victor Tarasenko, Dmitry Beloplotov, Mikhail Lo-
maev, Dmitry Sorokin. E-beam generation in dis-
charges initiated by voltage pulses with a rise time
of 200 ns at an air pressure of 12.5–100 kPa. 2019
Plasma Sci. Tech. 2019, 21(044007): 9.
https://doi.org/10.1088/2058-6272/ab079b
[25] C. Zhang, Tarasenko V. F., T. Shao, Beloplotov D. V.,
Lomaev M. I., R. Wang, P. Yan. Bent paths of a pos-
itive streamer and a cathode-directed spark leader in
diffuse discharges preionized by runaway electrons.
Physics of Plasmas, 2015, 22(3): 033511.
https://doi.org/10.1063/1.4914930
[26] Babich, L. P.. High-energy phenomena in electric
discharges in dense gases: Theory, experiment, and
natural phenomena. Futurepast Incorporated, 2003.
[27] D.V. Beloplotov, M.I. Lomaev, D.A. Sorokin, V.F.
Tarasenko, Displacement current during the forma-
tion of positive streamers in atmospheric pressure
air with a highly inhomogeneous electric field, Phys.
Plasmas, 2018, 25: 083511.
https://doi.org/10.1063/1.5046566
[28] Beloplotov, D. V., Tarasenko, V. F., Sorokin, D. A.,
Lomaev, M. I. Formation of ball streamers at a sub-
nanosecond breakdown of gases at a high pressure
in a nonuniform electric field. JETP Letters, 2017,
106(10): 653-658.
https://doi.org/10.1134/S0021364017220064
[29] Yu. P. Raizer, Gas Discharge Physics, Springer, Ber-
lin, 1991; Intellekt, Dolgoprudnyi, 2009.
[30] Kochkin, P., Van Deursen, A. P., De Boer, A., Bardet,
M., & Boissin, J. F.. In-flight measurements of en-
ergetic radiation from lightning and thunderclouds.
Journal of Physics D: Applied Physics, 2015, 48(42):
425202.
https://doi.org/10.1088/0022-3727/48/42/425202
[31] Bazelyan E. M., Raizer Y. P.. Lightning physics and
lightning protection. CRC Press, 2000.

9
ARTICLE
Rainfall Estimation using Image Processing and Regression Model on
DWR Rainfall Product for Delhi-NCR Region
Kuldeep Srivastava*
Ashish Nigam
Regional Meteorological Centre, India Meteorological Department, New Delhi 110003, India
Article history
Accepted: 22 May 2020
Observed rainfall is a very essential parameter for the analysis of rainfall,
day to day weather forecast and its validation. The observed rainfall data
is only available from five observatories of IMD; while no rainfall data is
available at various important locations in and around Delhi-NCR. How-
ever, the 24-hour rainfall data observed by Doppler Weather Radar (DWR)
for entire Delhi and surrounding region (up to 150 km) is readily available
in a pictorial form. In this paper, efforts have been made to derive/estimate
the rainfall at desired locations using DWR hydrological products. Firstly,
the rainfall at desired locations has been estimated from the precipitation
accumulation product (PAC) of the DWR using image processing in Python
language. After this, a linear regression model using the least square meth-
od has been developed in R language. Estimated and observed rainfall data
of year 2018 (July, August and September) was used to train the model.
After this, the model was tested on rainfall data of year 2019 (July, August
and September) and validated. With the use of linear regression model, the
error in mean rainfall estimation reduced by 46.58% and the error in max
rainfall estimation reduced by 84.53% for the year 2019. The error in mean
rainfall estimation reduced by 81.36% and the error in max rainfall estima-
tion reduced by 33.81% for the year 2018. Thus, the rainfall can be estimat-
ed with a fair degree of accuracy at desired locations within the range of the
Doppler Weather Radar using the radar rainfall products and the developed
linear regression model.
Keywords:
Rainfall estimation
Rainfall analysis
Doppler Weather Radar
Precipitation Accumulation Product
Image processing
Linear regression model
Kuldeep Srivastava,
Regional Meteorological Centre, India Meteorological Department, New Delhi 110003, India;
Email: kuldeep.imd@gmail.com
1. Introduction
R
ADAR is an acronym for Radio Detection and
Ranging. It is a device capable of detecting ob-
jects at far off distances, measuring the distance
or range of the object by using electromagnetic waves.
Radars have assisted weather predictions for over fifty
years but its operational use in hydrologic applications
spans only a decade or so. The hydrological applications
of the radar have gained traction with the introduction of
Doppler Weather Radars.
Doppler weather radar makes use of the Doppler Effect
to measure velocity of moving targets it detects. It works
by detecting the change in the frequency of the transmit-
ted and returned signal arising due to the movement of the
target. The velocity component of a target relative to the
radar beam is known as the “radial velocity”. The radar
used for the purpose of this research is a Dual Polarized
Doppler Weather Radar installed at IMD HQ, New Delhi.
It is a C-Band Radar (Frequency: 4-8 GHz, Wavelength:

10
8-4 cm) with a range of 250 Km.
Dual Polarization or Polarimetric Radar differs from
conventional Doppler radar by producing both a horizon-
tally polarized beam and a vertically polarized beam. A
horizontally polarized beam has its electric field oriented
in the horizontal plane, while a vertically polarized beam
has its electric field oriented in the vertical plane. This
allows the radar to provide information on the shape and
orientation of the hydrometeors and non-meteorological
scatterers that it detects.
Weather radars offer an unprecedented opportunity
to improve our ability of observing extreme storms and
quantifying their associated precipitation. These events
trigger floods and flash-floods, debris flow, and landslides.
However, quantitative estimation of rainfall from radar
observations is a complex process. It involves issues of
engineering design of a complicated and sophisticated
hardware with both electronic and mechanical subsys-
tems, signal processing, propagation and interaction of
electromagnetic waves through the atmosphere and with
the ground, image analysis and quality control, physics of
precipitation processes, optimal estimation and uncertain-
ty analysis, database organization and data visualization,
and hydrologic applications.
Past researches have suggested advances in rainfall
estimation using radar polarimetric observations, estima-
tion of the error structure of rainfall rate estimates, and
validation of radar rainfall algorithms along with high-
lighting the potential of radar-rainfall products for oper-
ational flood forecasting [7]
. Research has also shown that
estimates of precipitation are improved when rain gauge
observations are used to calibrate quantitative radar data
as well as to estimate precipitation in areas without radar
data [11]
. Research has also been done to improve the ac-
curacy of rain intensity estimation using artificial neural
network technique [6]
.
In this paper, the validation of the hydrological prod-
ucts of Delhi DWR has been carried out. The algorithm
used for the generation of the hydrological products
is NSSL2005 (a proprietary algorithm of VAISALA
Company). The hydrological products generated by the
above-mentioned algorithm have been analysed using im-
age processing.
A linear regression model was used to quantitatively
estimate the amount of rainfall in and around Delhi-NCR.
The regression models a relation between the depen-
dent(Y) and an independent(X) variable. The independent
variable is called the predictor variable and the dependent
variable is known as the response variable. The relation-
ship can be expressed as Y = c+ mX +……. nth degree.
Linear regression is a type of statistical analysis that at-
tempts to show the relationship between two variables. It
creates a predictive model on any data, showing trends in
data. In our case, the rainfall obtained from PAC product
is the independent variable and the final rainfall that we
want to estimate is the dependent variable.
The rainfall estimate obtained from hydrological prod-
ucts and the data obtained from surface observatories for
the year 2018 was used to train the linear regression mod-
el. This model was then tested on the rainfall data collect-
ed in 2019.
In this study, monsoon period has been considered,
since it is the most crucial period in which we require
accurate rainfall data from as many locations as possible.
This not only helps us in determining the stage of the
monsoon (onset or retreat) but also whether the monsoon
is below normal, normal or above normal. The study will
also help us in determining the rainfall amount at those
stations/areas where no observatory or rainfall measuring
equipment is present.
The second section of this paper describes the study
domain while the third sections lists the various data
sources for the research. The fourth section elaborates the
methodology followed for the purpose of the study. The
fifth and sixth section discuss the results and conclusions
respectively, that have been derived at the end of the re-
search.
2. Study Domain
In this study, rainfall data from Delhi-NCR, Haryana,
Rajasthan and Uttar Pradesh is considered. We have con-
sidered a total of 29 stations (Figure 1 and Figure 2) of the
above-mentioned states that come within 150 km from the
position of Doppler Weather Radar stationed at IMD HQ,
Lodhi Road, New Delhi.
Figure 1. The location of various districts covered as seen
in a PAC product

11
Figure 2. The district wise map of different regions cov-
ered
3. Data Sources
Data used for this study are:
(1) Surface Rainfall Intensity (SRI): The SRI generates
an image of the rainfall intensity in a user selectable sur-
face layer with constant height above ground. A user de-
finable topographical map is used to find the co-ordinates
of this surface layer relative to the position of the radar.
This map is also used to check for regions, where the user
selected surface layer is not accessible to the radar. These
parts of the image will be filled with the NO DATA value.
The product provides instantaneous values of rainfall in-
tensity. The estimated values of reflectivity are converted
to SRI by using Z=ARb [8]
where R is the rainfall intensity,
A and b are constants. The values of A & b vary from sea-
son to season and place to place.
(2) Precipitation Accumulation (PAC): The PAC prod-
uct is a second level product. It takes SRI products of the
same type as input and accumulates the rainfall rates in a
user-definable time period (look back time). Every time a
new SRI product is generated, the PAC generation starts
again. The display shows the colour coded rainfall amount
in [mm] for the defined time period. Precipitation Accu-
mulation (PAC) Products generated by the DWR have
been downloaded (in .gif format) from radar section of
the official website of India Meteorological Department
(https://mausam.imd.gov.in/imd_latest/contents/index_
radar.php) that are updated on a daily basis. A bash script
was written to automate this process. At a fixed time, each
day, the script downloads the PAC product gif images
from the official website and saves them in the repository
with that day’s date.
(3) Those PAC products that were not available on the
website have been generated in the Radar Lab from the
RAW data of the concerned date (Figure 3). Raw Radar
data of the concerned date was sent to the input folder of
the IRIS software. This raw data was then ingested by the
IRIS software. After the ingestion, Surface Rainfall Inten-
sity (SRI) product was created for every 10-min interval.
Once all the SRI products were created, RAIN-1 product
was created for each hour. Thus, a total of 24 RAIN-1
products were created for the whole day. After that, a sin-
gle RAIN-N (PAC) product was created for the whole 24-
hour duration by overlapping all the RAIN-1 products (N
can be programmed to have any value from 1 to 24).
Figure 3. PAC product generation procedure
(4) Realised Rainfall data has been taken from the
Month end rainfall reports submitted to Regional Meteo-
rological Centre (RMC) Delhi by Meteorological Centre
(MC) Lucknow and Meteorological Centre Chandigarh.
These reports are submitted in the form of excel sheets.
These sheets show date wise distribution of rainfall as
well as cumulative amount of rainfall for each month at
various observatories, automatic weather stations and au-
tomatic rain gauges.
4. Methodology
Visual Studio Code Editor was used to code the script in

12
Python 3 which would read the image and carry out im-
age processing. Pixel coordinates corresponding to each
concerned station were determined using Microsoft Paint
and then coded in the script. The script first reads the PAC
image (in .gif format) using PIL/Pillow library and then
converts it into OpenCV image format. Image processing
using OpenCV library was carried out to determine the
amount of rainfall. The algorithm for calculating the rain-
fall amount takes into account the value of the pixel at the
desired coordinate (Figure 4). The RGB combination cor-
responding to each colour in the Precipitation accumula-
tion scale (present on the right side of the PAC image) and
the associated rainfall range were stored in an array. The
pixel coordinates for each of concerned station were also
stored in an array. For each pixel coordinate, the RGB val-
ue was calculated. Based on the RGB value, the rainfall
was calculated. This step was repeated for all the pixels
lying in a 3*3 grid surrounding the station coordinates.
The average rainfall value of all the nine pixels was allot-
ted to the concerned station. This was done to eliminate/
reduce the effect of noise.
Figure 4. Image processing methodology
Data, thus calculated, was stored in csv format date
wise, corresponding to PAC product of each date. We re-
fer to this data as “Estimated Rainfall”. This was done for
both 2018 and 2019. The above calculated data and the
data from various ground stations (We refer to this data as
“Observed Rainfall”) for the year 2018 was then used to
train a linear regression model in R. The model was then
tested on the estimated and observed rainfall data of 2019
and the results were studied and analysed (Figure 5). R
Studio Integrated Development Environment (IDE) was
used for the purpose of the modelling the linear regression
model and testing it out.
Figure 5. Research methodology
5. Results and Discussion
5.1 Analysis of Observed and DWR Estimated
Rainfall for 2018
On comparing the observed and DWR estimated rainfall
for 2018, it is found that there is a high positive cor-
relation between the two (Figure 6). This comparison
was done without use of the developed linear regression
model. The minimum rainfall values shown by both the
methods were identical while the mean rainfall value esti-
mated using RADAR is 35.8% less than the mean rainfall
observed during the concerned months in 2018. The max-
imum rainfall value estimated using RADAR is 57.33%
less than the maximum rainfall observed during the con-
cerned months in 2018 (Table 1).

13
Figure 6. Scatter plot between observed and estimated
rainfall values for July, August and September 2018
Table 1. Estimated and observed rainfall comparison for
2018
Estimated Rainfall Observed Rainfall
Min. : 0.000 Min. : 0.000
Mean : 4.419 Mean : 6.888
Max. : 96.500 Max. : 226.200
5.2 Analysis of Observed and DWR Estimated
Rainfall for 2019
On comparing the observed and DWR estimated rainfall
for 2019, it is found that there is a high positive correla-
tion between the two (Figure 7). This comparison was
done without the use of the developed linear regression
model. The minimum rainfall values shown by both were
identical while the mean rainfall value estimated using
RADAR is 22.85% less than the mean rainfall observed
during the concerned months in 2019. The maximum rain-
fall value estimated using RADAR is 28.24% less than the
maximum rainfall observed during the concerned months
in 2019 (Table 2).
Figure 7. Scatter plot showing positive correlation be-
tween observed and estimated rainfall values for July,
August and September 2019
Table 2. Estimated and observed rainfall comparison for
2019
Estimated Rainfall Observed Rainfall
Min.: 0.000 Min.: 0.000
Mean: 3.285 Mean: 4.258
Max.: 85.100 Max.: 118.600
5.3 Development of Model using Least Square
Method
The model is found using the least square method using
the rainfall data from 2018. The minimum and maxi-
mum residuals associated with the model are -66.388 and
130.167 respectively (Table 3). The residuals, which are
vertical distance between data points and the regression
line, are both positive and negative, in this case which
shows that data points are evenly distributed on both sides
of the regression line.
Table 3. Model residuals
Residuals:
Min 1Q Median 3Q Max
-66.388 -0.461 -0.461 -0.461 130.167
The equation derived from the model is:
Actual Rainfall = 1.45454*(Rainfall estimated from PAC
product) + 0.4607 (1)
The standard error associated the coefficient of the
independent variable (i.e. rainfall estimated using PAC
product) is 0.02557 (Table 4). Such a low value indicates
that there is only a small error in the coefficient. The ratio
of coefficient value and its associated standard error gives
t-value and its high value (in our case: 56.879) indicates
high degree of accuracy of the coefficient of the indepen-
dent variable.
Table 4. Statistical parameters of the model
Coefficients:
Estimate Std. Error t value Pr(|t|)
(Intercept) 0.46071 0.29202 1.578 0.115
Estimated 1.45454 0.02557 56.879 2e-16
The value of R-squared is 0.6895 which indicates that
68.95% of the total variation in the dependent variable
is being explained by our equation/model. The p-value
is less than 2.2e-16 and thus it is less than 10%, 5% and
even 1% level of significance. Therefore, we can safely
reject the null hypothesis and conclude that our indepen-
dent variable is significant.
For the scope of this paper, the intercept value in Eq.
(1) has been ignored, since it is very small (i.e. less than
1mm) and on the assumption that when the radar indicates
that there was no rainfall, there was no observed rainfall.
5.4 Verification of Model Estimated and Observed
Rainfall
After training the linear regression linear model with 2018
rainfall data, the model was tested on 2019 rainfall data

14
and the results have been shown in Table 5.
Table 5. Image estimated rainfall, Model estimated rain-
fall and Observed rainfall comparison for 2019
Rainfall estimated direct-
ly from image
Rainfall estimated by
model
Rainfall Ob-
served
Min.: 0.000 Min.: 0.000 Min.: 0.000
Mean: 3.285 Mean: 4.7778 Mean: 4.258
Max.: 85.100 Max.: 123.7811 Max.: 118.600
With the use of the linear regression model, it can be
seen that the mean rainfall value estimated using model
is 12.2% more than the mean rainfall observed during
the concerned months in 2019 and the maximum rainfall
value estimated using model is 4.3% more than the maxi-
mum rainfall observed.
The linear regression linear model was re-run on the
training data set (2018 rainfall data) and the results have
been shown in Table 6.
Table 6. Image estimated rainfall, Model estimated rain-
fall and Observed rainfall comparison for 2018
Rainfall estimated direct-
ly from image
Rainfall estimated by
model
Rainfall Ob-
served
Min.: 0.000 Min.: 0.000 Min.: 0.000
Mean: 4.419 Mean: 6.4278 Mean: 6.888
Max.: 96.500 Max.: 140.3628 Max.: 226.200
With the use of the linear regression model, it can be
seen that the mean rainfall value estimated using model
is 6.68% less than the mean rainfall observed during the
concerned months in 2018 and the maximum rainfall
value estimated using model is 37.94% less than the max-
imum rainfall observed.
5.5 Discussion
With the use of linear regression model, the error in
mean rainfall estimation reduced by 46.58% and the error
in max rainfall estimation reduced by 84.53% for the year
2019. The error in mean rainfall estimation reduced by
81.36% and the error in max rainfall estimation reduced
by 33.81% for the year 2018.
The root mean square error between the model estimat-
ed and observed rainfall value was found to be as low as
0.9807644 mm (i.e. less than 1mm) which is evident in
Figure 8, Figure 9 and Figure 10
Figure 8. Plot of Observed value corresponding to differ-
ent index values
Figure 9. Plot of Model Estimated values corresponding
to different index values
Figure 10. Comparison of Model Estimated and Observed
values
6. Conclusions
Rainfall data was successfully extracted and actual rainfall
was estimated from the hydrological products of the RA-
DAR using image processing and regression for the mon-
soon period for the years 2018 and 2019. From the above
results and discussion, it can be concluded that the rainfall
at any desired location can be estimated using hydrologi-
cal products generated by the Doppler Weather Radar.
The developed linear regression model can be used to
estimate the rainfall amount to a great degree of accuracy
over a range of rainfall values. The equation obtained by
regression is:
Actual Rainfall = 1.45454*(Rainfall estimated from
PAC product)
However, it was also observed that when the rainfall is
very heavy (115.6 – 204.4 mm) to extremely heavy (=
204.5 mm) the difference between the estimated and ob-
served rainfall is large as compared to when the rainfall in
light to moderate.
This usage of hydrological products generated by
weather radars can be extended to radars located at other
locations across India and used to estimate rainfall where
no observatories/AWS/ARGs are available.
Acknowledgments
The authors are thankful to Director General of Meteo-
rology (DGM), India Meteorological Department (IMD),
New Delhi and Deputy Director General of Meteorology
(DDGM), RMC New Delhi for rendering all the facilities.
The authors also gratefully acknowledge MC Chandigarh,

15
MC Lucknow and RMC Delhi for providing the rainfall
data of Delhi-NCR and adjoining areas, DWR Division
and Radar Lab, I.M.D, New Delhi for providing all the
necessary PAC product images to carry out this study.
References
[1] Anagnostou E.N., Krajewski W.F. Real-Time Radar
Rainfall Estimation. Part I: Algorithm Formulation.
Journal of Atmospheric and Oceanic Technology,
1999a, 16: 189-197.
[2] Anagnostou E.N., Krajewski W.F. Real-Time Radar
Rainfall Estimation. Part II: Case Study. Journal of
Atmospheric and Oceanic Technology, 1999b, 16:
198-205.
[3] Anagnostou E.N., Anagnostou M.N., Krajewski W.F.,
Kruger A., Miriovsky B.J. High-Resolution Rainfall
Estimation from X-Band Polarimetric Radar Mea-
surements. Journal of Hydrometeorology, 2004, 5:
110-128.
[4] Brandes E.A. Optimizing Rainfall Estimates with the
Aid of Radar. Journal of Applied Meteorology. 1975,
14: 1339-1345.
[5] Brandes E.A., Zhang G., Vivekanandan J. Experi-
ments in Rainfall Estimation with a Polarimetric Ra-
dar in a Subtropical Environment. Journal of Applied
Meteorology, 2012, 41: 674-685.
[6] Dutta D., Sharma S., Sen G.K., Kannan B.A.M,
Venketswarlu S., Gairola R.M., Das J., Viswanathan
G. An Artificial Neural Network based approach for
estimation of rain intensity from spectral moments
of a Doppler Weather Radar. Advances in Space Re-
search, 2011, 47: 1949-1957.
[7] Krajewski W. F., Smith J. A. Radar hydrology: rain-
fall estimation. Advances in Water Resources. 2002,
25: 1387-1394.
[8] Marshall J.S., Langille R.C., Palmer W. McK. Mea-
surement of Rainfall by Radar. Journal of Meteorolo-
gy. 1947, 4: 186-192.
[9] Smith J.A., Baeck M.L., Meierdiercks K.L., Miller
A.J., Krajewski W.F. Radar rainfall estimation for
flash flood forecasting in small urban watersheds.
Advances in Water Resources. 2007, 30: 2087-2097.
[10] Sun X., Mein R.G., Keenan T.D., Elliott J.F. Flood
estimation using radar and rain gauge data. Journal
of Hydrology. 2000, 239: 4-18.
[11] Wilson J.W., Brandes E.A. Radar Measurement of
Rainfall - A Summary. Bulletin American Meteoro-
logical Society, 1979, 60: 1048-1058.
[12] Xiao R., Chandrasekar V. Development of a neural
network-based algorithm for rainfall estimation from
radar observations. IEEE Transactions on Geoscience
and Remote Sensing, 1997, 35: 160-171.

16
ARTICLE
Behavior of the Cultivable Airborne Mycobiota in air-conditioned
environments of three Havanan archives, Cuba
Sofía Borrego*
Alian Molina
Preventive Conservation Laboratory, National Archive of the Republic of Cuba, Compostela 906 esquina a calle San
Isidro, PO Box: 10100, La Habana Vieja, Havana, Cuba
Article history
Accepted: 23 June 2020
High concentrations of environmental fungi in the archives repositories are
dangerous for the documents preserved in those places and for the workers'
health. The aims of this work were to evaluate the behavior of the fungal
concentration and diversity in the indoor air of repositories of 3 archives
located in Havana, Cuba, and to demonstrate the potential risk that these
taxa represent for the documentary heritage preserved in these institutions.
The indoor and outdoor environments were sampled with a biocollector.
From the I/O ratios, it was evident that two of the studied archives were not
contaminated, while one of them did show contamination despite having
temperature and relative humidity values very similar to the other two.
Aspergillus, Penicillium and Cladosporium were the predominant genera
in the indoor environments. New finds for archival environments were the
genera Harposporium and Scolecobasidium. The principal species classified
ecologically as abundant were C. cladosporioides and P. citrinum. They are
known as opportunistic pathogenic fungi. All the analyzed taxa excreted
acids, the most of them degraded cellulose, starch and gelatin while about
48% excreted different pigments. But 33% of them showed the highest
biodeteriogenic potential, evidencing that they are the most dangerous for
the documentary collections.
Keywords:
Archives
Environmental fungi
Indoor environments
Microbial quality of archive environments
Quality of indoor environments
Documentary biodeterioration
Sofía Borrego,
Preventive Conservation Laboratory, National Archive of the Republic of Cuba, Compostela 906 esquina a calle San Isidro, PO Box:
10100, La Habana Vieja, Havana, Cuba;
Email: sofy.borrego@gmail.com; sofy.borrego@rediffmail.com
1. Introduction
A
t present the continuous knowledge and
control of the environmental conditions in
archive, libraries and museums constitutes of
the most important elements to take into account in the
preventive conservation of the Documentary Heritage of
a Nation. The prevalence of inadequate environmental
conditions together with the presence of high microbial
concentrations in the air of the repositories of archives
and library where this heritage is conserved, has been
attracting the attention of researchers and specialists in
the area of the conservation of heritage property, due to
the risk that this implies for both for the integrity of the
preserved heritage and for the health of the staff who work

17
in these institutions or who receive systematic services
in them [1-3]
. Specifically, fungal contamination is one of
the main objects of study, since fungal spores constitute
one of the most numerous bioaerosols of all the biological
material that is transported by air, in addition to possessing
a high biodeteriogenic and pathogenic potential [1,4-8]
.
The existence of high values of temperature and
relative humidity in countries with a tropical climate, such
as Cuba, favors the increase of dust and the concentration
of fungal spores and propagules in the air, as well as
their deposition over different materials, facilitating the
development and proliferation of fungi. They have a
powerful, versatile and adaptable metabolic machinery,
which allows them to degrade a wide variety of substrates,
both organic and inorganic, promoting biodeterioration of
the different supports that make up artworks of heritage
value [9-12]
. Also, fungi are characterized by having
different structures and pathogenic mechanisms, which
cause specific diseases in humans [1,13,14]
.
Numerous studies have established a close relationship
between environmental conditions, the presence of
viable or non-viable propagules fungal and their
incidence in triggering respiratory affectations [1,3,13]
,
achieving associate their presence with the development
of symptoms belonging to these types of pathologies
and others [13,16]
. For this reason, multiple research
groups recommend the need to increase the frequency
of systematic studies of environmental conditions in
premises to assess the quality of the environments, in
order of guaranteeing an environmental characterization
of the same to solve problems associated with the
development of pests and/or affections to the health of the
personnel precociously.
Taking these aspects into account, the National Archive
of the Republic of Cuba (NARC) has been investigating
the environmental quality of the documental repositories
not only of the institution itself but also of other archives
in the country. For this reason, the aims of this work were:
(1) to evaluate the behavior of the fungal concentration
of the indoor air in repositories of 3 archives located in
Havana, Cuba, (2) to determine the density and relative
frequency of the isolated taxa in order to know their
ecological and environmental impact, and (3) demonstrate
the potential risk that these taxa represent for the
documentary heritage preserved in these institutions.
2. Materials and Methods
2.1 Characteristics of Repositories
The study was carried out in air-conditioned repositories
of three institutions that preserve documents with heritage
value. They were the Map library (ML) in the National
Archive of the Republic of Cuba (NARC), two premises
of the same repository in the Cuban Industrial Property
Office (CIPO) and two repositories of the Library of
Standard (LS) belonging to the National Center for
Management and Development of the Quality.
Both the CIPO and the NARC are located in the
Habana Vieja municipality a few streets away from each
other, while the LS is located in an adjoining municipality
(municipality of Centro Habana) about 2.3 Km away from
the NARC and CIPO approximately.
The ML is a large repository measuring 15.2 x 6.2 x 5
m (length x width x height) and is located on the first floor
and south side of the building, has several air conditioners
that maintain an annual average temperature between 23
and 26°C. This repository preserves a total of 195 lineal
meters of maps, elaborated mostly in different types of
papers.
The premises of the CIPO are located on the ground
floor of the building and are arranged one below the other
(A and B) in the form of a mezzanine built with steel
and concrete beams, their dimensions are 17 x 8 x 5 m
and they share the same air conditioning system with an
average annual temperature that ranges between 22°C
and 24°C. This institution conserves documentary funds
of great value from the 18th century to the present and
has a total of 1265136 documents in paper format mainly
(inventions, industrial models, scientific discoveries,
trademarks and other distinctive signs).
The repositories of the LS are smaller and measure 6 x
7 x 2.5 m approximately, they conserve the national norms
of quality on paper support. These repositories are located
on the ground floor of the building and are acclimated
through a centralized climate system that works only
during work hours. This repositories do not have windows
and only communicate with the building itself through its
access door.
2.2 Sampling and Mycological Analysis of the Air
For sampling, 11 points were selected in the CIPO (A: 6
in the premises below and B: 5 in the premises above).
Also, outdoor air sample was taken from the courtyard
located in the central area of the building. In the ML, 5
points were analyzed and on the roof of the building the
outdoor air was analyzed while in the LS 6 points were
selected in total (3 in each repository) and one outside
the building (entrance) (Figure 1) . These sampling
points were determined according to Sánchis (2002) [17]
.
All samples were taken between 10:00 am and 1:00 pm,
considering the possibility of the highest concentrations of
fungal propagules in the city’s atmosphere [18]
.

18
Map library (ML) of NARC
Cuban Industrial Property Of-
fice (CIPO). Left: Side view of
the repositories in mezzanine
form. Right: Front view of
right side of the superior repos-
itory
Library of Standard (LS)
Top view of the ML
Top view of the superior floor
of the CIPO repositories
Top view of the LS
Figure 1. Representative images of the studied repositories and microbiological
sampling points in each analyzed repository. In Map library (ML) 5 points were
analyzed, in CIPO 6 points on the floor below and 5 points on the superior floor
were sampled (total = 11 points) and in LS 3 points were sampled
Culturable airborne fungi were sampled at each point
by triplicate using a Super 100 SAS collector (Italy) and
flow rate analyzed was 100 L/min at 1 hour intervals
between replicates. The culture medium used for the
isolation was Malt Agar Extract (BIOCEN, Cuba)
supplemented with NaCl (7.5%) [19,20]
. Once the sampling
was completed, the Petri dishes were incubated at 30°C
for 7 days and the isolation of the different colonies was
carried out. Then, the colony count was performed and
the necessary calculations of air were made in order to de-
termine the microbial concentration expressed in colony
forming units per cubic meter (CFU/m3
).
In parallel, temperature (T) and relative humidity (RH)
at each sampling point were measured in situ during
sampling.
2.3 Identification of the Fungal Isolates
Cultural and morphological characteristics of fungal
colonies as well as conidiophores and conidia fungal
structures were observed under a trinocular microscope
optic with an attached digital camera (Samsung, Korea)
and the identification was performed according to
different manuals [21-30]
.
2.4 Ecological Criteria to the Environmental Taxa
Isolated from the Repositories
Relative density (RD) of the fungal genera or species
isolated from indoor air of each repository was conducted
according to Smith (1980) [31]
where:
RD = (Number of colonies of the genus or species/
Total number of colonies of all genera or species) x 100
The relative frequency (RF) determination was made
according to Esquivel et al. (2003) [32]
to determine
the ecological category of each fungal genus or specie
isolated. It was necessary to use the following formula:
RF = (Times a genus or specie is detected/Total number
of sampling realized) x 100
The ecological categories are: Abundant (A) with RF =
100 – 81%; Common (C) with RF = 80 – 61%; Frequent
(F) with RF = 60 – 41%; Occasional (O) with RF = 40 –
21%; Rare (R) with RF = 20 – 0. %.
2.5 Determination Semi-quantitative of the
Biodegradation Potential of the Isolated Taxa
2.5.1 Determination of Enzymatic Index (EI)
To quantify the cellulolytic, amylolytic and proteolytic
enzymatic index (EI), the following formula was used [5,33]
:
EI = 1- Dc / Dca
Where Dc is the colony diameter and Dca is the sum of
Dc and the diameter of the hydrolysis zone. Values between
0.5 and 0.59 were classified as low EI, between 0.6 and 0.69
as moderate EI, and above 0.7 as high. Each determination
was made in triplicate and averages are reported.
2.5.2 Cellulolytic Enzymatic Index (CEI)
The strains were inoculated in Petri dishes containing
an agar medium, the saline composition of which for
one liter was: sodium nitrate 2g, potassium phosphate
1g, magnesium sulfate 0.5g, ferrous sulfate 0.01 g,
chloride potassium 0.5g, yeast extract 0.5g and 20g of
agar technical No. 1. As a carbon source, carboxymethyl
cellulose (CMC) at 1% was added and incubated at 30°C.
After seven days, a solution of Congo Red (0.05g/L) was
added to each dish and was maintained by one hour, then
that solution was decanted and NaCl at 1 mol/L was added
for 10 min. Cellulolytic activity was evidenced by the
formation of a white halo around the colony [34, 35]
.
2.5.3 Amylolytic Enzymatic Index (AEI)
An agar medium of saline composition similar to that
used in the previous test was prepared in Petri dishes and
was inoculated with each strain. Starch (1%) was added
as a carbon source. After incubating for 7 days at 30°C,
a Lugol reagent solution was added into each culture
dish. The presence of a colorless halo around the colonies
evidenced the starch hydrolysis [9, 36]
.
2.5.4 Proteolytic Enzymatic Index (PEI)
The strains were inoculated in dishes containing an
agarized culture medium with a saline composition

19
similar to that used previously, with gelatin as the carbon
source (1%). The dishes were incubated at 30ºC; the
test reading was performed at 7 days of incubation with
the addition of the Frazier reagent. A white precipitate
around the colony (halo) is indicative of the presence of
non-hydrolyzed gelatin but the colorless halo revealing
the gelatin hydrolysis [37]
.
2.5.5 Determination of the Acid Excretion
0.1 ml of a conidia suspension of each strain was
inoculated into a culture broth with a saline composition
similar to the medium used to determine cellulite
activity. Glucose (1%) was used as the carbon source;
the pH was adjusted to 7 and 0.03% of phenol red was
added as indicator. The cultures were incubated at 30°C
for 3 days and the pH of the broth was subsequently
measured with a pH meter (Pacitronic MV 870, USA),
whose precision is ± 0.2 units. The positive result was
corroborated by the change in the color of the phenol red
indicator (from red to yellow) and the detection of pH
values less than 7 [20,36]
.
2.5.6 Determination of Extracellular Pigments
Excretion
The strains were inoculated in tubes with slants containing
an agarized culture medium with a saline composition
similar to CMC medium but with dextrose as the carbon
source (1%). The tubes were incubated at 30ºC during 7
days and excretion of diffusible pigments was observed in
the culture medium of each tube. This determination is a
modification of those reported by Borrego et al. (2010) [36]
.
Also, the pigments excretion in the medium with CMC
was taken into account.
2.6 Statistical Analysis
The ANOVA-1 and Duncan tests were used to compare
the fungal concentration obtained on the indoor of the
three archives environments as well as to compare the
enzymatic activities among strains. A P value smaller or
equal to 0.05 was considered statistically significant.
3. Results
3.1 Fungal Concentration and Diversity Detected
on Indoor Air of the Repositories
When analyzing the fungal concentrations in the indoor
air of the different archives (Table 1), the significantly
highest fungal concentration was detected in the LS (133.9
CFU/m3
) despite having values of T and RH similar
to those obtained in the other two archives. The other
repositories showed similar concentrations (CIPO with
42.7 CFU/m3
and ML of NARC with 40.8 CFU/m3
).
Table 1. Fungal concentrations detected on the indoor and outdoor environments of the three studied archives located in
Havana, Cuba
Concentrations
CIPO Library of Standard (LS) Map Library (ML) of NARC
Fungi indoor
(CFU/m3
)
T
(0
C)
HR
(%)
Fungi out-
door
(CFU/m3
)
Fungi
indoor
(CFU/m3
)
T
(0
C)
HR
(%)
Fungi
outdoor
(CFU/m3
)
Fungi indoor
(CFU/m3
)
T
(0
C)
HR
(%)
Fungi outdoor
(CFU/m3
)
Maximum 80 26.2 59.9 150 280 27.0 57.9 90 70 22.9 51.4 290
Minimum 55 25.3 54.4 90 130 25.8 52.8 45 20 24.4 49.6 150
Average ± SD 42.7±26.0 a 25.7±0.3 56.5±1.8 103.3±40.4 133.9±72.0 b 26.2±0.3 56.1±1.7 53.0±24.4 40.8±20.6 a 23.5±0.5 50.3±0.7 208.0±51.0
I/O ratio 0.4 2.5 0.2
Notes:
SD: Standard deviation. The determinations in CIPO were made in 11 points, in LS were made in 6 points and in the ML 5 points were analyzed
by triplicate, respectively; hence the data averaged were: n = 33 (CIPO), n = 18 (LS), n = 15 (ML). a, b: Indicates significant differences according
to the Duncan test (P ≤ 0.05) on comparing the fungal concentration obtained in indoor air of the archival environments studied. I/O ratio = Indoor
concentration/Outdoor concentration.
Simultaneous external air determinations in the outdoor
of each archive were made with the intention to estimate the
I/O ratio and to define the air quality in their environments.
The I/O ratios obtained were 0.4 for CIPO, 2.5 for LS and
0.2 for ML. In this case the I/O ratio of LS was markedly
higher indicative of a contaminated environment in spite
of having values of T and RH similar to those that have the
other two archives.
In this study a total of 12 genera of filamentous fungi and
2 non-sporing mycelia (WNSM: White Non-sporulating
Septated Mycelia, PNSM: Pigmented Non-sporulating
Septated Mycelia) were detected on indoor environments
whilst a total of 13 genera and 2 non-sporulating mycelia
were also detected in outdoor environments (Figure 2).

20
From indoor environments a total of 6 taxa were isolated
by the CIPO, 4 taxa by LS and 8 taxa by ML, but in all of
them Aspergillus, Cladosporium and Penicillium genera
as well as a white non-sporulating septated mycelium
(WNSM) were detected and for these reasons were
ecologically classified as abundant. Penicillium spp.
prevailed in CIPO and in LS environments but in ML of
NARC Cladosporium was the genus predominant. On
the other hand, other genera isolated from the CIPO were
Acrodontium and Cylindrocarpon; from LS the other
genus was Trichophyton and from ML other 5 genera were
isolated too (Chrysosporium, Harposporium, Neurospora,
Nigrospora, Scolecobasidium).
A
B
Figure 2. Relative density (RD) of the taxa detected on
the indoor (A) and outdoor (B) environments of the three
studied archives located in Havana, Cuba.
Note:
Mycelium, WNSM: White Non-sporulating Septated PNSM: Pigmented
Non-sporulating Septated Mycelium.
The biggest diversity of taxa was detected in the indoor
environment of ML with 10 of them, but 45.5% of the
taxa detected were part of the repository’s environment
itself while the other 54.5% appear to come from abroad;
in this case the highest incidence was the Cladosporium
spp. Although in LS the number of taxa detected was
markedly lower, 60% of them to come from the outdoor
environment with a high incidence of the Aspergillus
spp. while in CIPO the 100% of taxa detected indoor
environments to come from outdoor with a high impact
of the Cladosporium spp. too, but contrary to the external
impact the prevalence on indoor was the Penicillium spp.
Although a great diversity of species was detected
in general only 3 taxa were ecologically abundant
(Cladosporium cladosporioides, Penicillium citrinum
and WNSM), 9 were common taxa (Aspergillus
ochraceus, Aspergillus flavus, Aspergillus oryzae,
Aspergillus versicolor, Nigrospora sphaerica,
Penicillium griseofulvum, Penicillium oxalicum,
Penicillium simplicissimum and PNSM), and 31 were
classified as occasional taxa (Table 2).
Table 2. Relative density (RD) of the fungal taxa detected
on the indoor air of the three studied repositories as well
as their relative frequency (RF) and ecological category
(EC)
Taxa
CIPO LS ML RF
(%)
EC
RD (%)
Acrodontium simplex (Mangenot) de Hoog 2 0 0 33.3 O
Aspergillus athecius Raper Fennell 0 0 2.8 33.3 O
Aspergillus candidus Link 2.7 0 0 33.3 O
Aspergillus chevalieri L. Mangin 1.0 0 0 33.3 O
Aspergillus flavipes (Bain Sart) Thom
Church
1.0 0 0 33.3 O
Aspergillus flavus Link 2.8 0 2.8 66.7 C
Aspergillus glaucus Link (complex) 0 10.1 0 33.3 O
Aspergillus niger Tiegh. 0 1.8 0 33.3 O
Aspergillus niveus Blochwitz 1.0 0 0 33.3 O
Aspergillus ochraceus K. Wil. 0 2.6 2.8 66.7 C
Aspergillus oryzae (Ahlb.) Cahn 1.0 2.5 0 66.7 C
Aspergillus parasiticus Speare 0 2.5 0 33.3 O
Aspergillus penicilloides Spegazzini 0 6.3 0 33.3 O
Aspergillus unguis (Emile-Weil Gaudin)
Thom Raper
5.5 0 0 33.3 O
Aspergillus versicolor (Vuill.) Tiraboschi 1.0 5.1 0 66.7 C
Aspergillus wentii Wehmer 1.0 0 0 33.3 O
Cladosporium caryigenum (Ellis Lang) 0 0 8.5 33.3 O
Cladosporium cladosporioides (Fresen) G.A.
de Vries
15.0 3.8 10.0 100 A
Cladosporium coralloides W. Yamamoto 0 0 2.8 33.3 O
Cladosporium gossypiicola Pidoplichko
Deniak
0 0 2.8 33.3 O
Cladosporium herbarum (Pers.: Fr.) Link 0 0 2.8 33.3 O
Cladosporium hillianum Bensch, Crous U.
Braun
10.0 0 0 33.3 O
Cladosporium lignicola Corda 0 0 2.8 33.3 O
Cladosporium sphaerospermum Penz. 0 0 2.8 33.3 O

21
Cladosporium staurophorum (Kendrick) M. B.
Ellis
0 0 5.7 33.3 O
Cladosporium tenuissimum Cooke 0 0 2.8 33.3 O
Chrysosporium sp. Corda 0 0 2.7 33.3 O
Cylindrocarpon lichenicola (C. Massal.) D.
Hawksw.
1.0 0 0 33.3 O
Harposporium sp. Lohde 0 0 2.8 33.3 O
Neurospora crassa Shear B.O. Dodge 0 0 5.7 33.3 O
Nigrospora oryzae Hudson 0 0 2.8 33.3 O
Nigrospora sphaerica (Sacc.) E. W Mason 9.0 0 2.8 66.7 C
Penicillium chrysogenum Thom 0 10.2 0 33.3 O
Penicillium citreonigrum Dierckx 2.0 0 0 33.3 O
Penicillium citrinum Thom 5.0 11.5 8.5 100 A
Penicillium commune Thom 0 3.1 0 33.3 O
Penicillium griseofulvum Dierckx 5.0 4.0 0 66.7 C
Penicillium oxalicum Currie Thom 16.0 3.0 0 66.7 C
Penicillium simplicissimum (Oud.) Thom 0 29.6 1.0 66.7 C
Scolecobasidium sp. E.V. Abbott 0 0 2.8 33.3 O
Trichophyton sp. Malmsten 0 2.6 0 33.3 O
PNSM 4.0 0 8.4 66.7 C
WNSM 14.0 1.3 5.6 100 A
Notes:
WNSM: White Non-sporulating Septated Mycelium. PNSM: Pigmented
Non-sporulating Septated Mycelium. According to Esquivel et al. (2003)
[32]
when RF = 100 - 81% the taxon is considered ecologically Abundant
(A); 80 - 61% is Common (C); 60 - 41% is Frequent (F); 40 - 21% is
Occasional (O); 20 - 0.01% as Rare (R).
In relation to the Aspergillus spp., it was evidenced
a high variety of species, since there were 15 identified
in total. Of these, 10 species were isolated in the indoor
environment of CIPO, 6 in the LS environment and
only 3 in the ML environment. None of them turned out
to be ecologically abundant. However, 4 species were
ecologically common to have been detected in two of the
three archives what represents the 44.4% of all taxa that
were ecologically common.
From the 9 species of Cladosporium spp., only
C. cladosporioides had an important ecological
representation, the rest were occasional species because
they were only detected in a single archive, mainly in ML.
About the 7 species of Penicillium spp. 4 of them
(71.4%) were ecologically important (1 was abundant and
3 were common) for the LS environment fundamentally.
3.2 Biodegradative Assays Evaluation
In relation to the degradative activities (Table 3), the
majority of the taxa (93.6%) degrade in more or smaller
measure the cellulose but it is of emphasizing a group
of 13 taxa that showed the highest CEI (EI ≥ 0.7). They
were A. flavus 1, A. niger, A. ochraceus, Cladosporium
caryigenum, Neurospora crassa, Nigrospora oryzae,
P. chrysogenum, P. citrinum 1 and 2, P. griseofulvum,
P. oxalicum 3, P. simplicissimum and WNSM. It is
worth highlighting in this group the predominance of
Penicillium spp. (46.2%). In a second place for having
a moderate EI, 16 strains (34%) were found with a
predominance of species of the genus Aspergillus with a
37.5% (A. athecius, A. flavipes, A. flavus 2, A. ochraceus
2, A. oryzae, A. versicolor). Of the rest, 15 taxa showed
low cellulose degradative activity (31.9%) and 3 did not
degrade the polymer.
Table 3. Enzymatic index (EI) of the taxa isolated
from the indoor air of the studied archives to assess
their biodeteriogenic potential on several materials that
conform the archives collections
Origin Specie/Mycelium
Cellulolytic
Activity
Amilolytic
Activity
Proteolytic
Activity Acids
production
(pH)
Pigment
Excretion *
CEI AEI PEI
CIPO Acrodontium simplex 0.53 c 0.50 b 0.62 fg 5.02 hijklm -
ML Aspergillus athecius 0.60 ef 0 a 0 a 5.90 qrst -
CIPO Aspergillus candidus 0.51 b 0.62 fg 0.58 de 6.07 qrst -
CIPO Aspergillus chevalieri 0.50 b 0.65 gh 0.68 h 4.46 deg -
CIPO Aspergillus flavipes 0.60 ef 0.63 fg 0.58 de 3.72 bc + (yellow)
CIPO Aspergillus flavus 1 0.73 j 0.74 j 0.74 j 6.22 stuv -
ML Aspergillus flavus 2 0.63 fg 0.71 ij 0.56 cd 4.62 egh -
LS Aspergillus glaucus 0 a 0 a 0 a 6.60 vw -
CIPO Aspergillus niger 0.72 ij 0.71 ij 0.73 j 5.41 nño -
CIPO Aspergillus niveus 0.59 de 0.57 d 0 a 5.30 mn + (yellow)
LS
Aspergillus ochraceus
1
0.72 ij 0.69 hi 0.74 j 5.40 nño + (brown)
ML
Aspergillus ochraceus
2
0.66 gh 0.54 c 0.57 d 6.12 rstu + (brown)
CIPO Aspergillus oryzae 0.65 gh 0.68 h 0.71 ij 4.33 de + (yellow)
LS
Aspergillus
penicilloides
0.55 cd 0.58 de 0.66 gh 6.50 uvw -
CIPO Aspergillus unguis 0.57 d 0.56 cd 0.59 de 5.82 opqrs + (yellow)
LS Aspergillus versicolor 0.60 ef 0.68 h 0.62 fg 4.13 d -
CIPO Aspergillus wentii 0.52 bc 0.55 cd 0.53 c 4.82 ghijk -
ML
Cladosporium
caryigenum
0.70 hij 0 a 0 a 6.25 stuvw
+ (green
olive)
ML
Cladosporium
cladosporioides
0.66 gh 0.58 de 0.70 hij 3.34 ab + (brown)
ML
Cladosporium
coralloides
0.58 de 0.55 cd 0 a 5.85 pqrs + (brown)
ML
Cladosporium
gossypiicola
0.65 gh 0.68 h 0.56 cd 5.72 ñopqr
+ (green
dark)
ML
Cladosporium
herbarum
0.68 h 0.72 ij 0.62 fg 6.50 uvw
+ (green
dark)
CIPO
Cladosporium
hillianum
0.58 de 0.53 c 0.54 c 4.16 d
+ (amber
dark)
ML
Cladosporium
lignicola
0.52 bc 0.58 de 0.60 ef 6.60 vw + (brown)
ML
Cladosporium
sphaerospermum
0.66 gh 0.54 c 0 a 6.30 tuvw
+ (green
dark)
ML
Cladosporium
staurophorum
0.56 cd 0.62 fg 0 a 6.30 tuvw + (brown)
ML
Cladosporium
tenuissimum
0.52 bc 0.63 fg 0.54 c 6.11 qrstu + (brown)
ML Chrysosporium sp. 0.60 ef 0.65 gh 0.56 cd 6.40 tuvw
+ (amber
dark)
CIPO
Cylindrocarpon
lichenicola
0 a 0 a 0.63 fg 3.65 bc -
ML Harposporium sp. 0.69 hi 0.65 gh 0.60 ef 3.52 abc -
ML Neurospora crassa 0.73 j 0.68 h 0.72 ij 5.10 jklmn
+ (orange
clearing)
ML Nigrospora oryzae 0.72 ij 0.73 j 0.71 ij 5.10 jklmn + (brown)
CIPO
Nigrospora sphaerica
1
0.55 cd 0 a 0 a 5.21 lmn -
ML
Nigrospora sphaerica
2
0.64 g 0.56 cd 0.68 h 5.84 pqrs -
LS
Penicillium
chrysogenum
0.70 hij 0.69 hi 0.74 j 4.80 ghijk -
CIPO
Penicillium
citreonigrum 1
0.59 de 0.66 gh 0.71 ij 4.45 deg -
CIPO
Penicillium
citreonigrum 2
0.62 fg 0.54 c 0.64 g 6.01 ghijk -
LS Penicillium citrinum 1 0.72 ij 0 a 0.61 ef 5.27 lmn + (yellow)
ML Penicillium citrinum 2 0.73 j 0.72 ij 0.70 hij 4.36 deg + (yellow)
LS
Penicillium
griseofulvum
0.71 ij 0.69 hi 0.62 fg 5.15 klmn -
CIPO
Penicillium oxalicum
1
0.62 fg 0.57 d 0.72 ij 3.21 a -

22
LS
2
0.63 fg 0.52 bc 0 a 5.05 ijklmn -
LS
3
0.70 hij 0.68 h 0 a 4.72 eghij -
LS
Penicillium
simplicissimum
0.71 ij 0.68 h 0.58 de 6.60 vw -
ML Scolecobasidium sp. 0 a 0.56 cd 0.62 fg 6.64 w -
LS WNSM 0.72 ij 0.65 gh 0.58 de 5.35 mnñ + (yellow)
ML PNSM 0.58 de 0.62 fg 0.67 h 5.19 klmn + (brown)
Notes:
CEI: Cellulolytic Enzymatic Index. AEI: Amilolytic Enzymatic Index.
PEI: Proteolytic Enzymatic Index. Enzymatic index (EI) = 0.5 - 0.59 is
low, EI = 0.6 - 0.69 is moderate, EI 0.7 is high. +: indicates excretion
of pigments. - : Indicates no excretion of pigment. Values of pH 7
are indicative of the acids production. WNSM: White Non-sporulating
Septated Mycelium. PNSM: Pigmented Non-sporulating Septated
Mycelium. a - w: Different letters indicate significant differences
according to Duncan test among strains in the same column (P ≤ 0.05).
*: These pigments were detected in CMC medium and a culture medium
with similar composition to CMC but with dextrose as the carbon source
(1%).
Regarding starch, 41 taxa (87.2%) degraded this
polymer, only they did not do it with the same intensity.
Six species (12.8%) showed a high AEI (A. flavus 1
and 2, A. niger, Cladosporium herbarum, Nigrospora
oryzae, P. citrinum 2) while 19 taxa revealed moderate
activity (40.2%), 16 showed a low degradation (34%)
and 6 species did not degrade this nutrient. Likewise,
35 taxa degraded gelatin (74.5%), but 11 species stood
out for showing a high PEI (A. flavus 1, A. niger, A.
ochraceus 1, A. oryzae, Cladosporium cladosporioides,
Neurospora crassa, Nigrospora oryzae, P. chrysogenum,
P. citreonigrum 1, P. citrinum 2, P. oxalicum 1), which
represents 23.4% of the total of taxa, while 14 strains
(29.8%) degraded it moderately, 12 strains revealed low
degradative power (25.5 %) and 10 did not degrade it
(21.3%).
Although the acid was excreted by all the taxa, it is
necessary to highlight that 14 of them (29.8%) were
those that more lowered the pH of the culture medium (A.
chevalieri, A. flavipes, A. oryzae, A. versicolor, A. wentii,
Cladosporium cladosporioides, Cladosporium hillianum,
Cylindrocarpon lichenicola, Harposporium sp., P.
chrysogenum, P. citreonigrum 1, P. citrinum 2, P. oxalicum
1 and 3) while 21 taxa excreted different pigments (47.7%)
with prevalence of the yellow, amber and brown colors.
Among these taxa 4 species were very important for
documentary biodeterioration because they evidenced
the highest enzymatic index related to the degradation of
cellulose, starch and gelatin.
It is important accentuating that when a strain has
several degradative potentialities more dangerous is for
the conservation of documents; because it can use the
paper components as nutritious in a vigorous way if the
T the RH is already appropriate for its growth. The figure
3 shows the results in this sense. It can appreciate that
27% of taxa revealed 4 biodeteriogenic attributes while
33% of them exhibited 5 attributes; these represent a total
of 60% the strains with high potentialities to degrade the
majority of the paper components indicative of their high
biodeteriogenic power.
Figure 3. Behavior of the combination of different
biodegradative attributes related to the enzymatic
characteristics of the analyzed fungal strains detected on
the indoor air of the studied archives
4. Discussion
The influence of T or RH or even of the two parameters
together, on the behavior of indoor fungal concentration
and its diversity has been reported by several previous
studies [5-7, 38-40]
, however, this behavior has not been
evidenced in this study where the evaluated repositories
are air-conditioned and have similar average values of
T and RH. Therefore, this study has shown that the high
degree of air stagnation, the lack of air exchange with the
outside and the existence of a high content of dust inside
in some repositories were the factors that had a marked
impact on the behavior of the quality of the studied
environments and not the thermohygrometric values.
The environmental study of the three archives showed
some differences in the obtained concentrations; in
particular it was found that the LS value was significantly
higher despite the values of T and RH were similar
(Duncan test, p ≤ 0.05). Despite this, the concentrations in
all cases were lower than 150 CFU/m3
which is indicative
that the environments have low fungal loads according
to the criteria of Roussel et al. (2012) [15]
. However, the
concentrations of fungi in the outdoor environments
were higher than the indoor ones in the cases of CIPO
and ML while for LS the opposite occurred, the outdoor
concentration was lower.
Since there is still no standard in Cuba to evaluate
the microbiological quality of indoor environments in
archives, libraries and museums, comparisons were made
with the report of French authors’ mainly [15]
. The results

23
indicate in all cases that the environments had a low
concentration (less than 170 CFU/m3
). The comparisons
made with the value given by American Conference of
Governmental Industrial Hygienists Guidelines (100
CFU/m3
) indicate that only the LS environment was
contaminated, but the comparison with the World Health
Organization Guidelines (500 CFU/m3
) [41]
evidence that
all environments were not contaminated. However, the
climatic conditions of Cuba differ from those of France
or the United States or other European countries where
the environmental studies have been carried out with
greater frequency, for having a humid and very warm
climate; so we consider that the best way to classify the
quality of an indoor environment was by analyzing the
relationship between indoor and outdoor concentrations
(I/O ratio) according to the recommendations made by
other authors [6, 42, 43]
. The obtained results show that in
the case of LS the I/O ratio was markedly higher to 1 (I/
O = 2.5), indicative of a contaminated environment, with
little circulation of the air indoor the repositories and poor
environmental quality [6, 44, 45]
. On the contrary, in the case
of CIPO and ML the I/O ratios were less than 1, indicating
that there has been a good exchange with the outdoor
environment despite the fact that the repositories are air-
conditioned. In these archives, most of the fungi detected
come from outdoor sources.
This environmental behavior in LS can be due to the
fact that these repositories have never had air exchange
with the outdoor, therefore there is air stagnation and
had a high level of dust. It is very probable that the
contamination existing in the dust, on the documents and
on other surfaces of the repositories were kept in a process
of continuous resuspension and with time those fungal
propagules remain in a high concentration in the indoor
environment of the repositories. However, the other two
archives, while also air-conditioned, do exchange air with
the outdoor environment at times, either through doors
when opened or windows when facilities are cleaned
which is the time when windows open to facilitate the
renewal of the air inside the repositories.
It is noteworthy that in the literature refers that in
the outdoor environment there must be a higher fungal
concentration than the indoor one [42]
. This behavior was
detected in the case of CIPO and ML; however for LS the
opposite happened. It is believed that this can be attributed
to the fact that on that day the outdoor sampling carried
out showed a high mobility of the fungal propagules due
to the high existing vehicular movement that favored
the formation of air and dust turbulences, preventing
the propagules from sediment easily or were not readily
captured by the biocollector.
In relation to environmental fungi, the most of the
isolates were anamorphs of ascomycetes which is
indicative of their prevalence in the indoor micobiota [19,46]
.
It is important to highlight that this result is characteristic
of the sampling method used, since the use of culture
media favors development of anamorphic phases in
the fungi. Similar results were previously reported
in environmental studies carried out in the NARC in
acclimatized and natural ventilated repositories [19,20,38,46,47]
.
Regarding the predominance of the genera Aspergillus,
Cladosporium, Penicillium and WNSM, it coincides
with previous reports of results obtained in Cuban and
other countries’ libraries and archives [3,5,8,10,19,20,40,41,48-
51]
. It is reported that these genera can produce numerous
conidia that can be easily dispersed by air for this reason
are common on indoor environments[50]
. However,
other genera were also detected to a lesser extent, such
as Acrodontium, Chrysosporium, Cylindrocarpon,
Nigrospora, Neurospora, Trichophyton, Harposporium
and Scolecobasidium, these last two genera being new
finding for Cuban archive environments.
Piontelli [27]
reported that Aspergillus genus is widely
distributed in the environment throughout the world,
especially in tropical and subtropical areas. Also, Leite-
Jr. et al. [40]
informed that Penicillium is a genus common
in cold climates while Aspergillus is most common in the
tropic climates and warm locations. However, according
to our results, the behavior of Penicillium does not agree
with the previous report, since it was precisely this genus
that predominated in the indoor environments of CIPO
and LS, an aspect that is not the first time that it occurs in
environments of Cuban archives [20,47]
. On the other hand,
Harkawy et al. [42]
and Molina and Borrego [38]
indicated
that is common that Aspergillus spp. and Penicillium
spp. predominate in archive and library environments
due to the presence of objects and documents on paper,
parchment and textiles that are materials that can be
biodeteriorated by species of these genera, in addition to
the fact that they can be present in sedimented dust. Also,
these genera are considered the first colonizers of the
surfaces [3,4,8,45]
.
It is reported that airborne fungi detected on indoor
environment usually enter a building through the
ventilation, air conditioning system, doors and windows,
together with the dust or they are part of the contaminants
that are present on building materials [1,19,49,52]
. This is
one more reason that indicates the need to compare
the indoor environment with the outdoor. Hence when
comparing the behavior of the isolated taxa inside the
archives and the outdoor environments, it was found
that for CIPO the coincidence of was 100%, that is, all

Journal of Atmospheric Science Research | Vol.3, Iss.1 January 2020

Journal of Atmospheric Science Research | Vol.3, Iss.1 January 2020

Recommended

Recommended

More Related Content

Similar to Journal of Atmospheric Science Research | Vol.3, Iss.1 January 2020

Similar to Journal of Atmospheric Science Research | Vol.3, Iss.1 January 2020 (20)

More from Bilingual Publishing Group

More from Bilingual Publishing Group (20)

Recently uploaded

Recently uploaded (20)

Journal of Atmospheric Science Research | Vol.3, Iss.1 January 2020