CLIMA: A weather generator framework

Weather generators (WG) can be defined as collections of models to estimate site specific weather data and derived variables. Their use spans from providing inputs to a variety of biophysical models to deriving weather indices. Also, using either global circulation models or local area models inputs, sets of parameters calculated from long term weather series specific to a site can be modified to reproduce via WG synthetic series representing climate change scenarios. Finally, models implemented in WG are used for estimating missing data and to perform quality control on data collected from sensors in weather stations. The models implemented in WG vary from purely empirical to physically based. There are several models to either estimate or to generate each weather variable, with different input requirements. New models are continuously being proposed, and, whether some models to estimate specific variables are commonly accepted as reference methods, the lack of some inputs requires at times using alternate approaches. Currently available WG are applications which implement a predefined set of modelling options, in software implementations which do not allow for independent extensions by third parties. The CLIMA weather generator is a component based application which consist of a set of reusable graphical user interface (GUI) components, and of a set of extensible model components. The latter are subdivided into six namespaces to estimate variables related to air temperature, rainfall, solar radiation, evapotranspiration, wind, and leaf wetness. The time characteristic of the variables estimated varies from a day to ten minutes. Another library allows estimating climatic indices from one year of daily data at the time. The current implementation consists of a total of more than 300 models. Components are usable either via the CLIMA GUI, or via custom developed applications in a client-server architecture. The architecture of components is based on the composite and strategy as keystone design patterns. Models are implemented as single approaches (simple strategies), and as composite models (composite strategies) which are associated to models of finer granularity. Another type of model unit is represented by context strategies, which implement logic to select within associated models. Finally, the GUI allows building composite models which can be saved as libraries, to be reused both within CLIMA for weather series generation, or independently by other applications. The components are implemented as .NET libraries. They implement the test of pre- and post-conditions, and a scalable tracing via .NET listeners. All variables and parameters are documented via a description, units, default, maximum, and minimum values. Components are extensible: new models can be added independently by third parties and detected by the CLIMA application, which can also use them for data generation via building new composite libraries. Each component is made available via a software development kit which includes the code of two sample projects, either to extend or to reuse the component. CLIMA and its model components are freely available for reuse in no-profit applications.
JRC.DG.G.3-Monitoring agricultural resources