Forecasting frost risk in forest plantations by the combination of spatial data and machine learning algorithms

Diniz, Écio Souza; Lorenzon, Alexandre Simões; Castro, Nero Lemos Martins de; Marcatti, Gustavo Eduardo; Santos, Osmarino Pires dos; Júnior, José Carlos de Deus; Cavalcante, Rosane Barbosa Lopes; Filho, Elpídio Inácio Fernandes; Amaral, Cibele Hummel do

doi:10.1016/j.agrformet.2021.108450

Cited by 17 publications

(3 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another approach to calculating frost risk is represented by the methodology of [28], which defines frost resistance of fruit trees based on the lethal temperature for 50% of the trees contained in the sample (LT50). Other authors also use more advanced and unconventional approaches in their risk assessment, such as machine learning [29] and a coupled stochastic model [30], where the first phase of the process is parameterized by the Ornstein-Uhlenbeck process with linearly increasing spring temperature and the second phase with the thermal time model.…”

Section: Introductionmentioning

confidence: 99%

Frost Risk Assessment in Slovenia in the Period of 1981–2020

et al. 2023

View full text Add to dashboard Cite

As spring frost proves to be an increasing risk throughout Slovenia and Europe, a better assessment of frost risk is needed. The statistical approach presented in this article consists of the conditional probability that the last spring frost occurs before budburst or flowering. The analysis was conducted using two separate phenological models and phenological data of various grapevine (Vitis vinifera L.), apple (Malus domestica), and sweet cherry (Prunus avium L.) varieties in locations across Slovenia. The increase in risk of spring frost for grapevine ranged from 1 to 1980, from 0.06 to 12 for apple, and from 1 to 180 for sweet cherry. Overall, the varieties most prone to frost proved to be Refošk (Teran) and Merlot grapevine varieties as well as the Germersdorf sweet cherry variety. We have identified the location in the hilly region with moderate climate where the Bobovec apple variety is grown as the least exposed to frost. Although counterintuitive, the GDD generally proved somewhat more efficient than the two-phase phenological model BRIN, although not in all cases. For the purpose of the study, the phenological models were calibrated, and the model parameters can serve as invaluable information for further research of this topic.

show abstract

Section: Introductionmentioning

confidence: 99%

Frost Risk Assessment in Slovenia in the Period of 1981–2020

et al. 2023

View full text Add to dashboard Cite

show abstract

“…The development of control methods for the system of wind and solar power plants is a priority to reduce the intermittent mode of green energy [5], which is the main limiting factor for its development [6]. The explosive interest in machine learning algorithms for improving climate and meteorological forecasts is already leading to important results for forestry [7], agriculture [1], telecommunication systems [8], in addition to fundamental studies of climate dynamics [3,9,10].…”

Section: Introductionmentioning

confidence: 99%

Estimates of the accuracy and rate of convergence of short-term meteorological forecasts using the regional climate model RegCM4

Tertychnaya,

Khoperskov

2023

E3S Web Conf.

View full text Add to dashboard Cite

We use the regional climate model RegCM 4.9 for the territory of Southern Russia to estimate the accuracy of meteorological forecasts. The errors of meteorological and climatic forecasts is due to complex non-equilibrium physicochemical processes in the components of the Earth's geosystem, which are characterized by gasdynamic and thermal instabilities at various spatial scales. A method for assessing the convergence of forecasts based on time series analysis is discussed for various localities. The reliability of forecasts for temperatures up to 1 degree can lie within 5-10 days for the south of the European part of Russia, depending on the predominance of cyclonic or anticyclonic atmospheric movements. The interval of acceptable precipitation forecast, as a rule, does not exceed 2-4 days for an area of about 104km2 and greatly decreases with decreasing area of the region under consideration. The annual volume of precipitation correlates in the cities of Rostov-on-Don and Volgograd at the level of 70 percent with each other. The correlation between the Rostov-on-Don City or the Volgograd City on the one hand, and the cities Astrakhan, Stavropol, Elista on the other hand, turns out to be significantly less, ranging from 30 to 40 percent.

show abstract

“…e R 2 of MLP exhibited higher performance in summer and winter at 0.9549 and 0.9590, respectively. In addition, Diniz et al [9] generated possible local-scale predictors of frost occurrence, which included longitude, latitude, elevation, relative altitude, relief orientation, and Euclidean distance from hydrography. ree machine learning classifiers (random forest (RF), support vector machine (SVM), and MLP) were compared in order to determine which would most accurately predict frost occurrence, and RF has been found to be the most proficient algorithm.…”

Section: Introductionmentioning

confidence: 99%

Frost Forecasting considering Geographical Characteristics

Kim

2022

Advances in Meteorology

View full text Add to dashboard Cite

Regional accuracy was examined using extreme gradient boosting (XGBoost) to improve frost prediction accuracy, and accuracy differences by region were found. When the points were divided into two groups with weather variables, Group 1 had a coastal climate with a high minimum temperature, humidity, and wind speed and Group 2 exhibited relatively inland climate characteristics. We calculated the accuracy in the two groups and found that the precision and recall scores in coastal areas (Group 1) were significantly lower than those in the inland areas (Group 2). Geographic elements (distance from the nearest coast and height) were added as variables to improve accuracy. In addition, considering the continuity of frost occurrence, the method of reflecting the frost occurrence of the previous day as a variable and the synthetic minority oversampling technique (SMOTE) pretreatment were used to increase the learning ability.

show abstract

Forecasting frost risk in forest plantations by the combination of spatial data and machine learning algorithms

Cited by 17 publications

References 42 publications

Frost Risk Assessment in Slovenia in the Period of 1981–2020

Frost Risk Assessment in Slovenia in the Period of 1981–2020

Estimates of the accuracy and rate of convergence of short-term meteorological forecasts using the regional climate model RegCM4

Frost Forecasting considering Geographical Characteristics

Contact Info

Product

Resources

About