In order to solve the problem of the land agroecological estimation (natural resources potential) automation and artificial information system development, it is necessary to make the domain knowledge (DK) conceptualization, or conceptual modelling. The unified modelling language (UML) was chosen as a descriptive system. Three abstract objects (class, attribute and relationship) were selected to describe 33 concepts for land plot basic natural characteristics and 13 significant nature process aspects regulating changes of those characteristics. For 6 DK concepts abstract object “class” was chosen, for 27 DK concepts – “attribute”, for 13 nature process aspects – “relationship”. Class “land plot” is a central one interacting with the other 5 classes: “relief”, “agrometeorological resource”, “soil”, “erosion”, “vegetation”. All classes and attributes interdependencies are described by relationship classification of 3 types. The first type is dependency relationship showing on UML diagrams a directed connection between two classes towards the main class, which means that changing the main class properties implies changing the dependant class properties; the second type is association relationship, which is any relationship showing connection characterized by almost any verb of the Russian language; the third type is composition relationship showing connection between composite and its part and is always directed to the composite, where deletion of the composite class implies deletion of all parts. Optimization of the DK conceptual model described by means of UML diagram is a permanent process, thus new classes and concepts can be added to the model throughout the time.