A SOFTWARE FRAMEWORK TO SUPPORT AGRICULTURE

ACTIVITIES USING REMOTE SENSING AND HIGH

PERFORMANCE COMPUTING

Shamim Akhter and Kento Aida

National Institute of Informatics (NII) and Tokyo Institute of Technology (TITECH), Tokyo, Japan

Keywords: Agriculture, Remote Sensing (RS), High Performance Computing (HPC), Web Portal.

Abstract: Agricultural activity monitoring, enclosed quantifying the irrigation scheduling, tracing the soil hydraulic

properties, generating the crop calendar etc., is very important for ensuring food security. Farmers want to

know these information in a regular basis. Additionally, large scale agricultural activity monitoring requires

to congregate information from Remote Sensing (RS) images and that type of processing takes a huge

amount of computational time. Thus, optimization on the computational time is a vital requirement. In such

cases, High Performance Computing (HPC) can help to reduce the processing time by increasing the

computational resources. Moreover, web based technology can contribute an understandable, efficient and

effective monitoring system. Still, the merging domain researches on RS image processing, agriculture and

HPC are mainly in hypothetical or conjectural theme rather than practical implementation. Thus, this

research contributes a new software system to support agriculture activities in real time using both RS and

HPC. The main purpose of the system is to serve the valuable crop parameters information to the farmers

through a web base system in real time. Additionally, we are going to discuss in details about the

implementation issues of the proposed software system.

1 INTRODUCTION

Agriculture activity monitoring or prediction on crop

parameters such as crop growth in terms of planting

date, date of emergence, extents, acreage, planting

intensity, water stress, biomass, yield and etc. is

important. It can contribute to better policymaking,

timely countermeasures, optimization of water

resources distributions, damage assessment and

finally to supply food security. Particularly, when an

on-going experiment covers large area such as a

country, satellite imagery plays a vital role by

providing useful information. However, some

information, or crop parameters, is not visible

through satellite images, which reflects a practical

problem that we can not generate or observe those

parameters from remote places. Indirect method such

as inverse modelling technique with crop model can

solve the problem. However, processing the inverse

modelling with crop model has a problem in

practicality, that is, they require a huge amount of

processing times. It becomes necessary to introduce

methods for using higher processing power such as

High Performance Computing (HPC) technologies.

Some protocols or tools have been developed

concerning the inverse modelling techniques and

their HPC implementation models. However, the

interoperability protocol between those agriculture

applications and existing remote sensing (RS) image

processing software is also necessary to improve

practicality. Thus, a software framework is required

which will make the successful interconnection

between the inverse modeling techniques, the RS

image processing and the HPC technology. In this

study, we are going to propose such web based

software framework to support the agriculture

activity monitoring system. Additionally, the

framework implementation issues or the difficulties

for such framework implementation will be

discussed in this paper.

210

Akhter S. and Aida K. (2010).

A SOFTWARE FRAMEWORK TO SUPPORT AGRICULTURE ACTIVITIES USING REMOTE SENSING AND HIGH PERFORMANCE COMPUTING.

In Proceedings of the 5th International Conference on Software and Data Technologies, pages 210-213

 SciTePress

2 BACKGROUND

2.1 Inverse Modeling Techniques

Crop models, Soil-Water-Air-Plant (SWAP) (Van

Dam et al., 1997) or Decision Support System for

Agrotechnology Transfer (DSSAT) (Tsuji et al.,

1994), have capacity to simulate soil, water and crop

processes and serve as crop productivity monitoring

tool. Crop Assimilation Model (CAM) predicts

parameters of crop models with satellite images. A

new methodology was developed in (Ines, 2004),

CAM-GA, to analyze the crop model (SWAP)

parameters assimilation with Remote Sensing (RS)

data and that parameters assimilation procedure was

optimized by an evolutionary searching technique

called Genetic Algorithm (GA). CAM with double

layers GA, CAM-DLGA, uses directly visible multi-

resolution RS images (ASTER Image Webpage, 2009)

(MODIS Image Webpage, 2009) and inversely

assimilates to SWAP model data for estimating the

non-visible model parameters. Other similar

functionality models, e.g., CAM-PSO (Kamble and

Chemin, 2006) and CAM-PEST (Dorji, 2003), use

different evolutionary searching techniques.

However, processing the agricultural information

with CAM has a problem in practicality, that is, they

require a huge amount of processing times. It

becomes necessary to introduce methods for using

higher processing power such as High Performance

Computing (HPC) technologies.

2.2 HPC Implementation Issues

Multi computer based distributed systems (Clusters

and Grids) have a large processing capacity for a

lower cost, naturally, choice turns towards

developing HPC applications. However, it is not an

easy job to port CAM in HPC environment. The

application performance is significantly affected by

the data and task distribution methods on the HPC

and developers of agriculture or satellite image

processing applications need to solve the problem of

both data and task distribution, or how to distribute

data and tasks among single or multiple clusters

environment. The workload in HPC, the bandwidth,

the processors speed, parameters of evaluation

methods and data size are additional concerning

factors. Moreover, interoperability between the

agriculture application and existing RS image

processing software is also necessary to improve

practicality. However, users need to manually extract

satellite data from some databases in the existing

CAM works. Thus, agricultural researchers require a

software or tool for the agricultural activity

monitoring so that they do not need to concern about

the implementation issues for agricultural models or

RS image processing on HPC.

3 PROPOSED FRAMEWORK

Figure1 presents the essential components during the

building phrase of the proposed system, e.g.,

combines the agriculture applications with RS image

processing tools, models them for HPC and then

interacts with web tools. The objective of this

research is to develop a complete HPC system or

tool for CAM to identify the crop parameters from

satellite image. The overall realistic architecture of

the proposed tool is presented in Figure2 and the

designing steps consists “processing satellite images

automatically through HPC”, “CAM HPC

implementation with appropriate data and task

distribution schemes”. Individually, each step has

been discussed and implemented with different data

and application domains in (Akhter et al., 2007) and

(Akhter et al., 2008). GRASS GIS (GRASS GIS,

2009) tool has been used to process the satellite

images automatically. CAM has been implemented

as a GRASS module. The interconnectivity between

the GRASS on HPC platform has been successfully

established and then the GRASS CAM module has

been implemented in HPC platform with different

data and task distribution methods. However, their

combined framework for modeling the overall

distributed agriculture monitoring scheme has not

yet been established. Web based portal system on

HPC for processing RS images protocol need to be

established and that will merge those issues together

into a unique platform. An excellent, fast and

flexible open source mapping web software is

UMN/MapServer (GRASSMAP, 2009). WMS (Web

Map Service), is on online access to, or integration

and exploitation of geospatial information, was

successfully implemented in (Ninsawat, 2004) with

FOSS (Free Open Source Software) to provide

satellite image server. GRASSLink (Huse, 1995), a

UNIX shell script based model, is capable to access

spatial information from GRASS datasets and

displayed necessary parameters in web through web-

mapping applications. However, those works

provide the mapping concept on static image

environment. However, in the proposed framework,

the image provider can be from different domains

and then automatic data updating process will be a

vital requirement. Thus, an interconnection between

the domains is required with necessary security

A SOFTWARE FRAMEWORK TO SUPPORT AGRICULTURE ACTIVITIES USING REMOTE SENSING AND HIGH

PERFORMANCE COMPUTING

211

infrastructure. Figure 3 presents the web based

architecture for the proposed framework. Openlayer

(OpenLayer, 2009) or PyWPS (PyWPS, 2009) can

be a solution on that circumstance. In our proposed

methodology, we are going to select PyWPS, as it is

relatively new concept and easy to create connection

with GRASS. Additionally, PyWPS supports to

create dynamic GRASS location for a given input

image during the executable phase, makes the task

easier for sharing input images from another

repository. Two separate modules for client and

server will be provided. In the server module, the

CAM modules with GRASS GIS supporting

environment and the HPC implementation for data

distribution and task distribution schemes will be

available. Whereas, in the client module, the user

interface will be provided. User can select the

specific region from a given image through the

interface. XML based parser system will trace the

user given queries and submit those queries to the

server modules. Server module will execute the

GRASS based CAM module with user given queries

and generate unknown crop information; those are

not directly extractable from RS images.

Figure 1: Essential Components to Build the Framework.

Figure 2: The Overall Realistic Architecture.

Figure 3: The Web based Architecture Framework.

Figure 4: The UMN Mapserver based Static Image

Environment.

Figure 5: PYWPS based Agriculture System in Web.

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

212

4 RESULTS

Figure 4 presents the UMN Mapserver based static

image environment and here user can watch only the

predefined and preprocessed images. The images

will not be possible to change without defining the

new images inside the map files. Figure 5 presents

the expected PyWPS based agriculture system in

web. User can select the specific region to execute

the agriculture model and then the agriculture model

will process the input image and generate the outputs

as images in the user interface. This example is a

PyWPS demo implementation for CAM models. The

entire software framework is in the developing phase.

In near future, the fully automated software will be

provided as a package or tool with the support of the

following implementation issues: a) The RS image

and the crop data will be automatically shared from

the RS and agriculture data repository, b) users don’t

need to bother about the backend processes like the

crop models or HPC or the distribution mechanisms,

c) the expected time for processing image will be

within few hours range, and d) multi-users can

interact in the system at the same time. So, the portal

must be capable to serve multi users processing

requests.

5 CONCLUSIONS

This type of work will merge the RS based

agricultural system with HPC and can be a model

application for distributing the RS and agriculture

field of study. Specially, this is the first effort to

implement CAM-GA, CAM-DLGA, LMF and

GRASS on top of HPC to provide the time

optimization. The successful implementation of this

research can be extended to city or provincial level,

helps the policy makers to monitor the on field

agriculture behavior and take prompt decision and

action regarding any unusual condition. The

extended version of the software with necessary crop

model input data and RS image can capable to

monitor any provincial or country level agricultural

activities.

REFERENCES

Akhter, S., Chemin,Y., Aida, K., 2007, Porting a GRASS

raster module to distributed computing: Examples for

MPI and Ninf-G, OSGeo Journal, Vol2., pp.36-44.

Akhter, S., Osawa, K., Nishimura, M., and Aida,K., 2008,

Experimental Study of Distributed SWAP-GA Models

on the Grid, IPSJ Transactions on Advanced

Computing Systems, Vol.1 No.2, pp.193-206.

ASTER Image Webpage, 2009,

http://asterweb.jpl.nasa.gov/, Internet.

Dorji, M., 2003, Integration of SWAP Model and SEBAL

for Evaluation of on Farm, Irrigation Scheduling with

Minimum Field Data, Enschede, ITC, 100 p.

GRASS GIS, 2009, Geographic Resources Analysis

Support System, http://grass.itc.it/, Internet.

GRASSMAP, 2009, Simple demonstrational

GRASS/UMN/MapServer (Spearfish data),

http://grass.itc.it/start.html , Internet.

Huse, S.M., 1995, GRASSLinks: A New Model for Spatial

Information Access for Environmental Planning, PhD

Thesis, University of California, URL:

http://www.regis.berkeley.edu/sue/phd/, Internet.

Ines, A.V.M., 2004, Improved Crop Production Integrating

GIS and Genetic Algorithms, PhD Thesis, Asian

Institute of Technology (AIT), Khlong Luang,

Bangkok, Thailand, AIT Diss No.WM-02-01.

Kamble, B., Chemin, Y.H., 2006, GIPE in GRASS Raster

Add-ons, http://grass.gdf-hannover.de/wiki/,

GRASSAddOns, RasterAdd-ons, Internet.

MODIS Image Webpage, 2009,

http://modis.gsfc.nasa.gov/, Internet.

Ninsawat, S. and Honda, K., 2004, Development of

NOAA and Landsat Image Server using FOSS,

Proceedings of the FOSS/GRASS Users Conference,

Bangkok, Thailand.

OpenLayer, 2009, OpenLayers: Free Maps for the Web,

http://openlayers.org/, Internet.

PyWPS, 2009, Python Web Processing Service,

http://pywps.wald.intevation.org/, Internet.

Tsuji, G.Y., Uehara, G., and Salas, S., 1994, DSSAT v3.0.,

Honolulu, Hawaii: University of Hawaii.

Van Dam, J.C., et. al., 1997, SWAP Model,

http://www.swap.alterra.nl/, Internet.

A SOFTWARE FRAMEWORK TO SUPPORT AGRICULTURE ACTIVITIES USING REMOTE SENSING AND HIGH

PERFORMANCE COMPUTING

213