Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Agricultural Information Technology in Asia and Oceania 1998 119-119

© 1998 by The Asian Federation for Information Technology in Agriculture

An Information Delivery System for Implementation of A


Forecasting System for Rice Blast Development Based on
Real-Time Weather Data
Eun Woo Park*1), Kyu Rang Kim1), Sung Kee Kim2), Soon Sung Hong2), and Jang
Souck Yang2)
1) Department of Agricultural Biology, College of Agriculture and Life Sciences, Seoul National University, Suwon
441-744, Korea
2) Department of Plant Environment, Kyonggi Provincial Rural Development Administration, Hwasong 445-970,
Korea

Abstract have received only limited acceptance by growers due to


scientific and psychological constraints. Among the
An information delivery system for implementation potential constraints, inaccuracy of disease forecasting
of a forecasting system for rice blast development based information and lack of enough time-lead in the
on real-time weather data is under development on a PC information delivery are probably the major scientific
server with the Linux OS. The system is composed of constraints. Recent progress in weather monitoring and
four servers or programs: weather data management information technologies provides a great opportunity to
server, database server, program server, and web server. improve accuracy of forecasting information and to
The weather data management server regularly retrieves increase time-lead in the information delivery.
hourly weather data from eight automated weather Monitoring of weather conditions within the plant
stations installed at the experimental rice paddy fields of canopy is possible by automated weather stations. Real-
eight Provincial Rural Development Administrations time weather data can be transmitted from automated
through telephone modems. The weather data weather stations at remote sites to personal computers
management server transmits the new data to the via telephone modems or by radio telemetry. The disease
database server. The database server provides weather forecasting information based on the real-time weather
and disease data to other servers. The database server is data can be delivered immediately to the public via the
interfaced with the weather data management server and World Wide Web. This paper is to describe an
the program server by a C language API and with the information delivery system that was developed to
web server by a Perl language interface module. In the implement a forecasting system for rice blast
program server, a simulation model (BLAST) for development in Korea. The information delivery system
forecasting rice blast development calculates daily leaf was designed to produce and distribute site-specific
blast severity and panicle blast incidence based on hourly disease forecasting information on the real-time basis.
weather data provided by the database server. The web
server communicates with DBS and PS by the CGI System Structure and Programming Tools
scripts in Perl, and retrieves the weather and disease
forecasting data stored in DBS upon request by users The information delivery system is composed of four
through the web browser. The web server also provides a servers or programs: weather data management server
simple request form for users to specify the run-time (WDMS), database server (DBS), program server (PRS),
parameters on the web browser. The weather and disease and web server (WS) (Fig. 1). The whole system is
forecasting data are displayed on the web browser in text installed on a personal computer with a 200MHz
or graphic form using the JAVA applet. Pentium Pro CPU, 128MB RAM, three hard drives with
5GB storage space, two 28,800BPS telephone modems,
and a 10MBPS Ethernet interface card. The computer is
connected to the campus local area network (LAN) of
Introduction Seoul National University, which provides the Internet
access. The operating system of the computer is
Weather-based forecasting systems have been primarily based on Red Hat Linux
developed for a number of plant diseases (Campbell and (http://www.redhat.com/) version 4.2 with a
Madden, 1990). In general, however, existing forecasters modification to support Korean language.
*
To whom correspondence should be addressed.
E-mail: ewpark@plaza.snu.ac.kr

- 119 -
Park, et. al.

Automated
Telep
Weather Station hone
mode
(CR10X) m
Weather Data
Management Server
Data input (Perl/SQL) (Linux)
hourly weather data
retrieval

Database Server
(Linux) 4-2. Disease information
update (Perl/SQL)
Program Server
(Linux)
3. Data search 4. Requested
the BLAST model
(Perl/SQL) data (text)
execution
4-1. Script for BLAST
execution (CGI)

Web Server 1. Data requesting form


(Linux)
2. Data request by users WWW Browser
information delivery 5. Data (graph and text)

Fig. 1. An overview of the information delivery system for rice blast forecasting based on real-time weather data.

Weather Data Management Server


Sensors
Weather data Weather data
in ASCII format retrieval
module
Datalogger MEMORY
CPU
Weather
database Weather data
update module

Modem Modem
Update Data Location 1
Execution

Location 2
Hourly

Database Server
Location 3

Cron daemon ...


Weather Tables
in the OS

Fig. 2. The weather data management server (WDMS). WDMS retrieves weather data from remote weather
stations and update corresponding weather tables in the database server.
                         

- 120 -
Park, et. al.

WDMS consists of a module for retrieval of name, phone number, and memory address of the last
weatherdata from remote sites and a module for updating data retrieval are specified to telecommunicate with an
the weather data in DBS (Fig. 2). The data retrieval automated weather station (Anonymous, 1997). The
module, which is executed every hour by the crontab Riccardo Facchetti’s programming library, Libmodem
command of Linux, searches for a station file in the version 1.0, is used to control the computer modems.
script directory to access an automated weather station Once the automated weather station is accessed, the
via telephone modem. In the station file, the station newly stored data after the last access are retrieved and

Database Server
Disease
Weather data tables forecasting Disease forecasting
Station
index table data tables
index table Station n
Parameter A Result X
Station 1 Hourly data
DailyStation
data 2 Parameter B Forecasting
Station 2
... Hourly data ... Information
Result B
...
Station 1 Parameter X ... Forecasting
Station n Daily data Result A
... Hourly data Information
Daily data ... Forecasting
... Information
...

C language application programming interface library Perl language interface module

Weather Data Management


Program Server Web Server
Server

Fig. 3. The database server (DBS). DBS contains tables of weather data and disease forecasting data. The data tables
are indexed for easy access to requested data by the other servers.

Program Server Database Server


(BLAST) Latent Period
Spore Liberation air temperature
solar radiation
wetness period Weather Tables

Panicles
Spore Deposition no ? yes
air temperature Disease Tables
wind speed
Lesion Growth Neck Blast Incidence
air temperature
Spore Infection relative humidity

air temperature
wetness period Leaf Blast Severity Rice Growth On-demand Web
rainfall Execution Server
leaf area air temperature

Fig. 4. The program server (PS). PS is executed by a CGI program of the web server when a web user requests new
forecasting information.

- 121 -
Park, et. al.

stored in the computer in ASCII text format. The produce the disease forecasting information based on the
weather data update module processes the stored data specifications in a CGI script from WS. WS also
and updates the weather database in DBS. provides a simple request form for users to specify the
DBS, which is developed using the Mini SQL run-time parameters on the web browser. WS
database system (Hughes Technologies Pty. Ltd., communicates with DBS and PS by the CGI scripts in
Australia: http://www.hughes.com.au/), stores a database Perl. The weather and disease forecasting data are
with a number of tables containing weather and disease displayed on the web browser in text or graphic form
forecasting data (Fig. 3). DBS maintains not only tables using the JAVA applet.
for weather data and disease forecasting data but also
several index tables to identify stations and run-time Results and Discussion
parameters like rice cultivars, transplanting dates, and
nitrogen fertilizer levels. The database server is In Korea, the Committee for Forecasting Crop
interfaced with the weather data management server and Diseases and Insect Pests (CFCDIP) of the Rural
the program server by a C language API and with the Development Administration has been organized every
web server by a Perl language interface module. year to forecast possible outbreaks of major plant
Using the hourly weather data from DBS, PS diseases and insect pests. The forecasting information is
produces daily leaf blast severity and panicle blast issued at one or two-week intervals from April to
incidence estimated by a simulation model (BLAST) and September. The rice blast forecasting information from
stores them in DBS (Fig. 4). The hourly weather data CFCDIP is usually based on weather forecasting at
required by BLAST include air temperature, relative meso- or macro-scales and the number of air-borne
humidity, rainfall, solar radiation, wind speed, and leaf spores of the pathogen collected at more than 100
wetness. BLAST estimates the number of spores experimental rice paddy fields over the country. The
deposited on leaves, the percentage of spores penetrated information from CFCDIP is quite useful to understand
leaves successfully, the latent period in days, the number current situations of rice blast occurrence at the national
and size of lesions, the area of leaves per hill, and the scale. However, the information is neither site-specific
number of diseased panicles (Kim, 1994, Park et al., nor accurate enough for local growers to use it in
1997). practice for their disease management. In addition, there
WS retrieves the weather and disease forecasting data is some time-delay in delivering the forecasting
stored in DBS upon request by users through the web information to rice growers, so that it is difficult to apply
browser (Fig. 5). If the requested information on disease appropriate disease management technologies in time.
forecasting is not available in DBS, PS is driven to The present study is to establish an information delivery

Database Server
4-2. Disease information update

3. Data search 4. Requested Program Server


(Perl/SQL) data (text)
4-1. BLAST execution with
appropriate parameters

CGI program

1. Data requesting form for web users


WWW
HTTP Daemon 2. Weather and disease data request
Browser
5. Data & JAVA applet (graph / text)

Web Server

Fig. 5. The web server (WS). WS retrieves the weather and disease forecasting data stored in the database server upon
request by users.

- 122 -
Park, et. al.

20 0.8
1993 1994
15 0.6

10 0.4
Disease Severity (%)

5 0.2

0 0.0
7/9 7/19 7/29 8/8 8/18 7/15 7/22 7/29
1.0 0.8
1995 1996
0.8 0.6
BLAST + APR
0.6
Observed 0.4
0.4
0.2 0.2

0.0 0.0
7/19 7/29 8/8 8/18 7/19 7/29 8/8 8/18
Date (1993 - 1996)

Fig. 6. Results of BLAST model execution at the Kyonggi Provincial Rural Development

system that can support the Real-time Plant Disease server, respectively. If the number of automated weather
Tracking System (RPDTS) covering the whole country. stations and the concurrent users are significantly
RPDTS is presently under development with objectives increased in the future, all servers of the system are to be
to predict the current and future development of major overloaded. In order to avoid the problems of server
diseases at remote sites and to give early warnings of overloading, TCP/IP is used in the communication
possible outbreaks of diseases. between the servers. Therefore, the servers can be
By incorporating the automated weather monitoring installed separately on two or more computers if
system, the disease forecasting model for rice blast, and necessary.
the Internet, the information delivery system described in Another major obstacle in the rice blast forecasting is
this paper has many advantages as compare with the the validity of the disease forecasting model. Although
forecasting system currently used in Korea. The accuracy of the forecasting information could be
advantages include (1) increased accuracy and site- increased by using weather data measured in the rice
specificity of the forecasting information, (2) real-time paddy fields instead of the macro- or meso-scale weather
delivery of the forecasting information, (3) effectiveness data, the model itself may cause inaccurate forecasting
in sharing information on the Internet by multiple users, due to its limitation of validity. The results in Fig. 6
and (4) cost efficiency in the system implementation by indicate that BLAST has some limitations in the
using public domain software and programming tools. In accuracy of forecasting, especially when the disease
addition, the information delivery system can be development is very low. In 1993 when leaf blast
decentralized so that local extension offices can maintain occurred severely, the prediction by BLAST appeared
the system for their own interest. accurate over the season. However, when the disease
All server programs are currently installed in a single severity was less than 1% in 1994-1996, BLAST tended
personal computer with a 200MHz Pentium Pro CPU to overestimate the severity. BLAST in the program
and 128MB RAM. With this capacity of the computer, server needs to be improved in the future to increase its
approximately 15 and 100 simultaneous accesses are accuracy in forecasting.
possible with and without executions of the program The information delivery system in this study can be

- 123 -
Park, et. al.

applied to forecast other diseases and insects pests just by


changing forecasting models in the program server. Also,
if mesh climatic data are available at a fine scale in the
future, the system can produce the disease forecasting
information at the same scale without automated weather
stations at every site. In conjunction with geographical
information systems, this system can be used to generate
maps to illustrate disease development over the country.

References

Anonymous. 1997. CR10X Measurement and Control


Module Operator’s Manual. Campbell Scientific, Inc.
Campbell, C. L. and Madden, L. V. 1990. Forecasting
plant diseases. In Introduction to Plant Disease
Epidemiology. pp. 423-452. John Wiley & Sons, Inc.
Kim, K. R. 1994. Development of a rice blast forecasting
system based on near real-time microclimatic data.
M.S. Thesis. Seoul National University 100 pp.
Park, E. W., Kim, K. R., and Hwang, C. I. 1997.
Quantification of adult plant resistance of rice
cultivars to leaf blast. (Abstract) Phytopathology
87:S75.

- 124 -

You might also like