Development and evaluation of thermal models for predicting skin temperature of dairy cattle

Heat stress significantly impacts dairy production. Skin temperature (Tskin) is demonstrated to be critical to the thermal comfort of animals. The predictive models for Tskin of dairy cattle have great potential in guiding climate control systems in dairy barns to mitigate the heat stress of dairy cows efficiently. This study aimed to develop thermal models to predict the Tskin of dairy cows. Ten models were constructed based on integrating the existing models and formulas associated with the heat production and loss from the body surface and respiratory systems. The models were categorized into the two-layer and the three-layer models according to whether the core layer is included or not. The models were also categorized into the thermoregulatory and the mechanistic models based on the sub-model outside the skin surface. The prediction performances of the models were quantitively evaluated by comparing the predicted values with the Tskin measured by infrared thermography. 85 Holstein cows were included, and 453 infrared images that contained the mean and maximum Tskin were collected in this study. The root mean squared error (RMSE), mean absolute error (MAE), and coefficient of determination (R2) were adopted as the performance metrics. The results show that the two-layer mechanistic model, including the actual core-body temperature, performed best for the mean Tskin with RMSE, MAE, and R2 of 0.649, 0.516, and 0.891, respectively. The two-layer thermoregulatory model, incorporating the estimated core-body temperature, provided satisfactory prediction performance for the maximum Tskin with RMSE, MAE, and R2 of 0.770 degrees C, 0.580 degrees C, and 0.793, respectively. The models generally performed better under heat stress conditions of the temperaturehumidity index >= 72, with all the performance metrics considered. Besides, in comparison to the mean Tskin, the maximum Tskin is less sensitive to environmental parameters but more correlated with the core-body temperature. It seems that the maximum Tskin is a more appropriate representation of skin temperature in dairy cattle than the mean Tskin.

Automated summary

Heat stress significantly affects dairy production, and predicting skin temperature is crucial for guiding climate control systems. The study developed thermal models to predict Tskin of dairy cows, with the mechanistic model performing best for mean Tskin and the thermoregulatory model providing satisfactory performance for maximum Tskin. Environmental parameters impact the models' prediction performance under heat stress conditions.