A spatio-temporal methodology for greenhouse microclimatic mapping

Abstract

Greenhouse internal microclimate has been proven to be non-homogeneous in many aspects. However, this variability is only sometimes considered by greenhouse models, which often calculate climatic variables without any spatial reference. Farmers, on the other hand, may wish to have these differences highlighted as they could lead to aimed actions only for a specific area of the greenhouse, while at the same time, they are not willing to invest in sensors to be installed everywhere. This paper presents a data-driven methodology to generate a virtual 2D map of a greenhouse, which allows farmers to control any critical parameter they desire with minimum investment, as monitoring is done via soft sensing with only a few actual sensors. The proposed flow starts with a set of temporary sensors placed in the points of interest; then, a model for each of them is developed via linear regression and, finally, a map of the entire area can be derived by interpolating values from these models. This allows the generation of accurate models at a reduced cost as temporary sensors can be reused at other locations. The methodology has been tested on adjacent greenhouses and in two farms, where temperature and other climatic variables have been monitored. Experimental results show that the proposed methodology can reach an adjusted R2 value of 98% for predicting values in different greenhouse locations.

Fig 1. Scheme for the proposed methodology.

Proposed methodology for monitoring multiple points of interest based on a single sensor.

Fig 2. The Evja station.

The Evja station hosting persistent sensors.

Fig 3. Weather station and its position.

Aerial photo of the greenhouse with the positions of the weather station and of the Evja sensors.

Fig 4. Indoor sensors positioning.

Position of the Evja persistent and temporary sensors across the two greenhouses. Sensors denoted by letters were added in a second moment for validation purpose.

Fig 5. Soft sensor’s hierarchy.

Hierarchy of Soft Sensors and their relationships with available data sources.

Fig 6. Creation flow of Sparse Soft Sensors starting from acquired data.

Fig 7. Runtime usare of Sparse Soft Sensors at time t.

Fig 8. Runtime detection of anomalies.

Runtime exploitation of models to estimate the correctness of data read from a persistent sensor.

Fig 9. Validation phase.

Plot for validation phase of regression for sensor #9.

Fig 10. Comparison between timed and un-timed models.

Accuracy comparison for un-timed and timed models when a local recurring event occurs.

Fig 11. Validation of forecast model.

Validation data for the forecast model against the original forecast and the values sensed by the weather station.

Fig 12. Comparison between Sparse Soft Sensors and 2D-Sensor for two positions.

Comparison between estimated values with Sparse Soft Sensors and interpolation for two locations in the map. The first plot refers to a central location where interpolation estimates temperature precisely, while the second one refers to a corner of the working area where interpolation is less accurate than Sparse Soft Sensors.

Fig 13. Temperature map obtained by combining estimation from Sparse Soft Sensors and the 2D-Sensor.

Fig 14. Comparison between the faulted sensors and model’s predicted values.

Comparison between temperature readings from the Evja sensor and the same variable modeled through the weather station. The regression model detects the fault injected in the sensor.

Fig 15. Greenhouse heat map.

2D temperature map of the greenhouse obtained by interpolating temperature values sensed in specific positions (blue dots).