Reimplementation and reuse of the Canegro model: From sugarcane to giant reed


2015 - Computers and Electronics in Agriculture, 113, 193-202
Stella, T., Francone, C., Yamaç, S.S., Ceotto, E., Pagani, V., Pilu, R., Confalonieri, R.

Abstract:

Model reuse can be limited by software design, which often forces third parties to completely rewrite
new versions of existing models before adapting them to new needs. This tendency removes resources
from the improvement of models and from the extension of their domain, leading to the proliferation
of software tools representing a variety of different implementations of the same algorithms. The
component-oriented paradigm allows these limitations to be overcome, facilitating model reuse and
extension. This study presents the application of component-oriented principles to the reimplementation
of the sugarcane (Saccharum officinarum L.) model Canegro (DSSAT v4.5) in a framework-independent
component following the BioMA architecture. The potential for reuse and extension of the component
(UNIMI.Cassandra.CaneML.Canegro) is here demonstrated by its straightforward adaptation to the simulation
of giant reed (Arundo donax L.), a promising energy crop that shares several morphological and
physiological features with sugarcane. The new component, named UNIMI.Cassandra.CaneML.Arungro,
extends UNIMI.Cassandra.CaneML.Canegro, and was effectively developed by inheriting about 70% of
the sugarcane model code. The development, calibration and evaluation of the giant reed model were
performed using field data collected in two experimental sites in Northern Italy between 2009 and
2012. Model performances were satisfactory, with average relative root mean square error and modelling
efficiency for aboveground biomass simulation of 34.33% and 0.57, respectively. The Canegro component
is distributed via a Software Development Kit that includes documentation of code and algorithms, and
the source code of sample applications illustrating how to use it.


Keywords: Model reuse, BioMA, CaneML, Giant reed, Arungro
DOI: DOI:10.1016/j.compag.2015.02.009


CaneML
A framework-independent component implementing different approaches for growth of cane species