Frost Occurrence Research Articles

The study of frost occurrence and risk analysis in Indo-Gangetic Plains of India during recent decades

Frost, an important meteorological extreme event has discernible impacts on agricultural crops specifically in Indo-Gangetic plains. The information on time of occurrence of frost and its frequency can help in reducing the risks and avoidable losses due to frost. The study investigated the occurrence of frost indices viz., onset, cessation and frost-free duration at nine locations in six States in Gangetic Plains of India during November to March (151 Days) using daily minimum temperature from 1980 to 2015. The multiple risk levels (90%, 70% and 50%) were used to determine the occurrence of frost probability for all frost indices. Trends in the frost indices were analyzed using the non- parametric Mann-Kendall test. The results showed the onset of frost mostly in December and January, whereas cessation of frost was observed during January to March with variations in date of onset and cessation from one location to another. The mean value of frost duration varied between 42 days at Amritsar to 2 days at Delhi during past 36 years. Number of frost days showed negative trends reflecting reduction in duration of frost. The rate of decrease in frost duration ranges from 0.5 day to 1.5 days per decade depending on the location. Trends for onset, cessation and frost-free duration were inconsistent at different stations.

Earlier and more variable spring phenology projected for eastern Canadian boreal and temperate forests with climate warming

Climate change affects timing and variability of spring phenology, as well as occurrence of spring frosts, and therefore influences forest structure, function, and management practices. In this study, we evaluated changes in budburst timing, sequential order of budburst, budburst temperatures, and frosts, from 1981–2010 baseline to different future climates (+2, +4, and +6 °C for moderate, high, and extremely high warming) for selected boreal and temperate species in eastern Canada. We used species-specific budburst models that are developed from large-scale forcing experiments.Budburst averaged 10–15 days earlier per 2 °C increase of temperatures, except for temperate yellow birch and eastern white pine that slowed down to 5–7 days in budburst advancement from +4 to +6 °C. Earlier budburst was associated with greater annual and interspecific variations in budburst timing, lower budburst temperatures, more frosts, and less annual variations in sequential order of budburst. Compared to temperate trees, boreal trees had greater budburst advances and annual variations, but less interspecific variations in timing and sequential order of budburst. Early flushing species had greater phenological changes and annual variations, as well as more frosts. Our results suggest that budburst advance that has occurred will continue with greater among years/species variability but is unlikely to converge between boreal and temperate regions in eastern Canada under anticipated levels of climate warming.

Agricultural zoning and recommendations for the seeding of wheat (Triticum species) in the Central-Southern Mesoregion of Parana State in Brazil

Wheat is one of the most cultivated cereals and is essential for human and animal feeding. Since it is a sensible species to the meteorologic conditions, the climate changes can alter all its cultivation areas and practices known in the present agriculture. This paper aims to make the zoning of climatic risk for agriculture for the wheat crop (Triticum species) and recommend the planting window for the central-southern Mesoregion of Parana State. The crop's climate risk was evaluated by several variables: Temperature (minimum and average), precipitation, water deficit, and frost. The entire Mesoregion presents aptness for wheat cultivation. The water demand for the species is supplied in all tested scenarios. The risk is also not identified for precipitation. The average temperature in a section of the region is lower than the recommended, but the seeding can be shifted to seasons with increasing temperatures. The entire region presents a high probability of frost occurrence. Because wheat is a winter crop, seeding during this period could be beneficial if late frosts are avoided, mainly during August and September. Thus, seeding is recommended in small intervals in the region, varying from May 21 to September. Key words: Climate aptness, climate risk, agricultural planning, climate change.

Assessing temporal response to biomass removal: A framework for investigating evolving constraints on boreal stand development

Impacts of harvest intensity and soil disturbance on site productivity are likely to vary with stand development stage and interactive processes related to ecosystem function. On upland boreal sites, stand productivity may initially be constrained by microclimatic conditions. However, increasing nutrient demands and overstory light interception associated with canopy development may place progressively greater emphasis on nutrient availability, profile water storage capacity and asymmetric competition. We illustrate these effects using 20-yr post-harvest data from fourteen jack pine (Pinus banksiana Lamb.) and black spruce (Picea mariana (Mill.) BSP) Long-term Soil Productivity (LTSP) installations in northern Ontario.Initial growth and ten-year survival were as great or greater without forest floors (NFF), but by year 15 dominant height increments were usually greater with intact forest floors (FF). Growing season frost occurrence and high vapor pressure deficits occurred more frequently with FF than NFF treatments during the initial stand establishment phase. Similarly, mean growing season soil temperatures were 2-5˚C warmer with NFF during this phase, with stand Leaf Area Index (LAI) values < 0.5. During the following Accelerated Growth phase, treatment differences in microclimate dissipated as LAI increased. Foliar N concentrations and N:P ratios, and soil net N mineralization were usually greater from an early age with FF, and often declined as canopy closure approached. Likewise, modelled site-level constraints related to soil water availability increased with stand development, peaking at LAI values associated with the stem exclusion stage. Placing post-harvest response (i.e., impacts of harvest intensity) in the context of relationships among stand structural development, changes in microclimate, and resource supply and demand should provide greater insights into longer-term effects of establishment practices.

Climate Response in Tree-Rings of Sawara Cypress [Chamaecyparis pisifera (Siebold &amp; Zucc.) Endl.] in Poland

Sawara cypress [Chamaecyparis pisifera (Siebold &amp; Zucc.) Endl.] is originally from Japan. It was introduced to Europe in the latter half of the 19th century (in England and Holland in 1861, and in Poland in 1864). The aim of this study was to examine the influence of climatic conditions on tree-ring width among Sawara cypress populations growing in Poland. Additionally, other indicators determining the growth-climate relationship for the studied tree species were investigated such as false rings, missing rings, or frost rings. Five stands of Sawara cypress from northwestern and central Poland were selected for study. Samples were taken from 97 trees, using Pressler borers at breast-height. Tree-ring widths were measured down to 0.01 mm. Climatic data came from weather stations located the nearest to the study plots. Tree-ring width in the studied populations of Sawara cypress varies (from 1.94 to 4.47 mm). The oldest Sawara cypresses grow in Glinna Arboretum and are nearly 130 years old. The youngest ones grow in Rogów Arboretum (67 years old) and Wirty Arboretum (58 years old). Ten regional pointer years, including six negative and four positive ones, were determined for local chronologies. Negative pointer years were associated with the occurrence of cold winters and water shortages in summer. Positive pointer years are mostly periods with a warm winter season, early and warm spring, and with high precipitation totals during summer months. Correlation and response function analysis corroborates the results yielded by pointer year analysis. False rings carry an additional information on pluvial conditions in the summer period, and frost rings are an aid in dating dendrochronological series and indicate the occurrence of both very cold winters and persistent ground frost occurrences in the spring period. In comparison to native conifers, the Sawara cypress can be regarded as a fast-growing species. The knowledge of acclimatization, growth rate, and growth-climate relationship may be useful, especially in the time of a rapidly changing climate, increasing human impact, and highly intensified invasion of insect and fungal species attacking native forest-forming taxa.

Analysis of an Occurrence of High Frosts During the Growing Season in Central-East Poland in 2001-2018

Analysis of an Occurrence of High Frosts During the Growing Season in Central-East Poland in 2001-2018

Machine Learning-Based Hourly Frost-Prediction System Optimized for Orchards Using Automatic Weather Station and Digital Camera Image Data

Spring frosts damage crops that have weakened freezing resistance after germination. We developed a machine learning (ML)-based frost-classification model and optimized it for orchard farming environments. First, logistic regression, decision tree, random forest, and support vector machine models were trained using balanced Korea Meteorological Administration (KMA) Automated Synoptic Observing System (ASOS) frost observation data for March from the last 10 years (2008–2017). Random forest and support vector machine models showed good classification performance and were selected as the main techniques, which were optimized for orchard fields based on initial frost occurrence times. The training period was then extended to March–April for 20 years (2000–2019). Finally, the model was applied to the KMA ASOS frost observation data from March to April 2020, which were not used in the previous steps, and RGB data were extracted by digital cameras installed in an orchard in Gyeonggi-do. The developed model successfully classified 117 of 139 frost observation cases from the domestic ASOS data and 35 of 37 orchard camera observations. The assumption of the initial frost occurrence time for training helped the most in improving the frost-classification model. These results clearly indicate that the frost-classification model using ML has applicable accuracy in orchard farming.

Phenological advance of blossoming over the past century in one of the world’s largest urban forests, Gauteng City-Region, South Africa

The Gauteng City-Region in the northern interior of South Africa hosts one of the world’s largest and most densely vegetated urban forests. The tree species, distributed across pavements, parks and suburban gardens, comprise a range of indigenous and alien species. The most aesthetically distinct of these is Jacaranda mimosifolia (Bignoniaceae), a purple-blossoming tree introduced from Brazil in the 1800s to beautify the cities. The distinct appearance during flowering and their abundance in the Gauteng City-Region has resulting in reporting of peak flowering events in local newspapers throughout the past century. This provides a valuable phenological record, particularly in southern Africa where phenology is seldom recorded. Analysing these reports of Jacaranda mimosifolia flowering, an advance of 2.1 days per decade is calculated for the period 1927–2019. This occurs against a backdrop of statistically significant annual and monthly temperature increases of ∼0.1−0.2 °C/decade for Tmax and ∼0.2−0.4 °C/decade for Tmin, and non-uniform change in rainfall. This phenological advance is most significantly related to winter climatic conditions, including Tmax, rainfall and frost occurrence. The strongest phenological driver is June Tmax, at a rate of 4.3–5.3d/°C across the City-Region. This advance reflects the response of the tree to regional climate warming, which poses threats to the species and the urban forest in the long term when thresholds for adaptation are surpassed.

Ground Frosts in Poland in the Growing Season

The ongoing climate warming affects, among others, and the variability of thermal conditions in spring and autumn are resulting in earlier dates of the beginning and end of the growing season. The present paper provides detailed characteristics of the phenomenon of ground frosts, addressing the question of whether the risk of frost-related damage in the extending growing season is still present. The assessment of temporal and spatial distribution of ground frosts (5 cm above ground level-AGL) in Poland in the thermal growing season (AT &gt; 5 °C) was conducted on the basis of the results of air temperature measurements at 5 and 200 cm a.g.l. obtained from 52 station of the Institute of Meteorology and Water Management—National Research Institute for the period 1971–2020. The thermal growing season was calculated using the method by Gumiński. The following were calculated: the dates of the occurrence of ground frosts in spring and autumn, duration of the frost-free period, the number of days with frosts of various intensity, according to years, ten-day period and days in a year together with trends of change. The conducted analysis demonstrated that the number of days with frosts in the thermal growing season in the multiannual period under analysis (1971–2020) does not show any statistically significant changes. It was found that in the thermal growing season, the average number of days with ground frosts in Poland amounts to 28 and ranges in spring from 15 to 22, and in autumn from 8 to 12. Most frequently, approx. 49%, slight frosts are recorded, followed by moderate (29%), severe (15%) and very severe (7%). A positive effect of water reservoirs on decreasing the frequency and intensity of frosts, as well as sporadic occurrence of the phenomenon in the second half of June were demonstrated. Statistically significant earlier dates of disappearance of frosts in spring, later dates of occurrence in autumn and the lengthening of the frost-free period from approximately 2 days in the north-east of Poland to approximately 8.0 days over 10 years in the Pomerania region, create increasingly more favourable conditions for the cultivation of plants with high thermal requirements in Poland.

Impact of successive spring frosts on leaf phenology and radial growth in three deciduous tree species with contrasting climate requirements in central Spain.

Rear-edge tree populations forming the equatorward limit of distribution of temperate species are assumed to be more adapted to climate variability than central (core) populations. However, climate is expected to become more variable and the frequency of climate extremes is forecasted to increase. Climatic extreme events such as heat waves, dry spells and spring frosts could become more frequent, and negatively impact and jeopardize rear-edge stands. To evaluate these ideas, we analyzed the growth response of trees to successive spring frosts in a mixed forest, where two temperate deciduous species, Fagus sylvatica L. (European beech) and Quercus petraea (Matt.) Liebl. (sessile oak), both at their southernmost edge, coexist with the Mediterranean Quercus pyrenaica Willd. (Pyrenean oak). Growth reductions in spring-frost years ranked across species as F. sylvatica > Q. petraea > Q. pyrenaica. Leaf flushing occurred earlier in F. sylvatica and later in Q. pyrenaica, suggesting that leaf phenology was a strong determinant of spring frost damage and stem growth reduction. The frost impact depended on prior climate conditions, since warmer days prior to frost occurrence predisposed to frost damage. Autumn Normalized Difference Vegetation Index data showed delayed leaf senescence in spring-frost years and subsequent years as compared with pre-frost years. In the studied forest, the negative impact of spring frosts on Q. petraea and especially on F. sylvatica growth, was considerably higher than the impacts due to drought. The succession of four spring frosts in the last two decades determined a trend of decreasing resistance of radial growth to frosts in F. sylvatica. The increased frequency of spring frosts might prevent the expansion and persistence of F. sylvatica in this rear-edge Mediterranean population.

AGRICULTURAL CLIMATE RISK ZONING FOR GLADIOLUS IN SANTA CATARINA

This work aims to developing an Agricultural Climate Risk Zoning for gladiolus in the State of Santa Catarina. The bioclimatic requirements and the risk criteria were defined for the crop through a bibliographic review. The probability of frost occurrence, water balance and frequency of maximum air temperatures above 34°C were calculated with meteorological database. A maximum assumed risk of 20% was considered. The period when planting is not recommended was determined for each of the development cycles (early, intermediate I and II and late). It was observed that low temperatures and occurrence of frost in autumn and winter limits gladiolus planting in the State of Santa Catarina. Analysis of the water balance indicated there is risk of water deficiency throughout the year, which varies according to the locations. Maximum air temperatures above 34oC do not occur frequently in the State. In the Coastal regions, there are no restrictions to planting throughout the year. The Agricultural Climate Risk Zoning indicate the potential for gladiolus farming in Santa Catarina, where the greatest planting potential is in the Coastal, West and Extreme West regions.

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

Brazil is one of the leading timber producers in the world. However, in South Brazil, frost events frequently cause damage and reduce yield in forest plantations, a situation predicted to become more common under global change scenarios. This raises the need for low cost and efficient tools, such as machine learning algorithms to improve forecasting of frost risk. This study used machine learning algorithms to create zoning classifications forecasting frost risk for forest plantations located in the south-central region of Rio Grande do Sul State, Brazil. For this, we gathered and processed data from a local geodatabase (i.e. high-spatial-resolution contour lines, hydrography, and forest stands limits) comprising 30 management units with consistent historical data of frost occurrence. Then, we generated possible local-scale predictors of frost occurrence, which included longitude, latitude, elevation, relative altitude, relief orientation, and Euclidean distance from hydrography. We carried out tests of three machine learning classifiers (Random Forest – RF; Support Vector Machine-SVM and Multi-layer Perceptron-MLP) in order to determine which would most accurately predict frost occurrence. We found that RF provided the highest accuracy (> 90%), as well as the smallest percentages of class-specific errors (i.e. commission and omission errors), when compared to SVM and MLP. Latitude was the most important predictor of frost occurrence when using RF. Conversely, MLP performed worst, especially for classifying frost occurrence versus non-occurrence, and therefore had the highest percentage of class-specific errors. Our findings lead us to conclude that RF is the most proficient algorithm for forecasting frost occurrence from local-scale geomorphological data, without the need for high-cost investment in micro-meteorological sensors to monitor climate frost events linking temperature to plant damage. With increasing global climate extreme events, accurate risk zoning is essential for planning strategies of plantation at the landscape scale.

Mimosa scabrella Benth. planted for forest restoration in southern Brazil

Tropical trees planting resurgent in a global warming scenario, and this activity relies on information about forest species ecology and its different functional traits. we evaluated the initial performance of Mimosa scabrella using the Generalized Linear Models (GLM) technique to fit mathematical models for collar diameter growth x height, and for crown projection area x collar diameter in a forest restoration planting. The following variables were obtained in the first four years after planting: collar diameter (mm), total height (m), crown projection area (m²) and crown volume (m³). The models fitting was performed by GLM on Gamma, Normal and Poisson distribution, identity and logarithmical join function, and they were analyzed based on Akaike and Bayesian Information Criteria, standard deviation of the estimate and determination coefficient. M. scabrella showed an excelled performance on the 48 months of age, with average increment of 200.28 mm in collar diameter, 6.88 m in total height, 2.95 m² in crown projection area and 83.41 m³ in crown volume. Its growth reduced only on periods of frost occurrence, but it restarted after these climatic events. The species growth, as well as its high survival rates and resprouting, allow its recommendation as a shady species of fast canopy fulfillment in forest restoration projects.

The enigma of relict large sorted stone stripes in the tropical Ethiopian Highlands

Abstract. Large forms of sorted patterned ground belong to the most prominent geomorphic features of periglacial and permafrost environments of the mid-latitudes and polar regions, but they were hitherto unknown in the tropics. Here, we report on relict large sorted stone stripes (up to 1000 m long, 15 m wide, and 2 m deep) on the ca. 4000 m high central Sanetti Plateau of the tropical Bale Mountains in the southern Ethiopian Highlands. These geomorphic features are enigmatic since forms of patterned ground exceeding several metres are commonly associated with distinct seasonal ground temperatures, oscillating around 0 ∘C. To systematically investigate present frost phenomena and relict periglacial landforms in the Bale Mountains, we conducted extensive geomorphological mapping. The sorted stone stripes were studied in more detail by applying aerial photogrammetry, ground-penetrating radar measurements, and 36Cl surface exposure dating. In addition, we installed ground temperature data loggers between 3877 and 4377 m to analyse present frost occurrence and seasonal ground temperature variations. Superficial nocturnal ground frost was measured at 35–90 d per year, but the ground beneath the upper few centimetres remains unfrozen the entire year. Seasonal frost occurrence would require a mean annual ground temperature depression of about 11 ∘C, corresponding to an air temperature decrease of about 6–8 ∘C (relative to today) as inferred from a simple statistical ground temperature model experiment. Our results suggest the formation of the large sorted stone stripes under past periglacial conditions related to lateral and vertical frost sorting in the course of cyclic freezing and thawing of the ground. It is likely that the stone stripes formed either in proximity to a former ice cap on the Sanetti Plateau over the last glacial period due to seasonal frost heave and sorting or they developed over multiple cold phases during the Pleistocene. Although certain aspects of the genesis of the large sorted stone stripes remain unresolved, the presence of these geomorphic features provides independent evidence besides glacial landforms for unprecedented palaeoclimatic and palaeoenvironmental changes in the tropical Bale Mountains during the (Late) Pleistocene.

Warmer springs have increased the frequency and extension of late-frost defoliations in southern European beech forests

Climate change is increasing the frequency of extreme climate events, causing profound impacts on forest function and composition. Late frost defoliation (LFD) events, the loss of photosynthetic tissues due to low temperatures at the start of the growing season, might become more recurrent under future climate scenarios. Therefore, the detection of changes in late-frost risk in response to global change emerges as a high-priority research topic. Here, we used a tree-ring network from southern European beech (Fagus sylvatica L.) forests comprising Spain, Italy and the Austrian Alps, to assess the incidence of LFD events in the last seven decades. We fitted linear-mixed models of basal area increment using different LFD indicators considering warm spring temperatures and late-spring frosts as fixed factors. We reconstructed major LFD events since 1950, matching extreme values of LFD climatic indicators with sharp tree-ring growth reductions. The last LFD events were validated using remote sensing. Lastly, reconstructed LFD events were climatically and spatially characterized. Warm temperatures before the late-spring frost, defined by high values of growing-degree days, influenced beech growth negatively, particularly in the southernmost populations. The number of LFD events increased towards beech southern distribution edge. Spanish and the southernmost Italian beech forests experienced higher frequency of LFD events since the 1990s. Until then, LFD events were circumscribed to local scales, but since that decade, LFD events became widespread, largely affecting the whole beech southwestern distribution area. Our study, based on in-situ evidence, sheds light on the climatic factors driving LFD occurrence and illustrates how increased occurrence and spatial extension of late-spring frosts might constrain future southern European beech forests' growth and functionality. Observed alterations in the climate-phenology interactions in response to climate change represent a potential threat for temperate deciduous forests persistence in their drier/southern distribution edge.

Shifts in the thermal niche of fruit trees under climate change: The case of peach cultivation in France

Climate influences plant phenological traits, thus playing a key role in defining the geographical range of crops. Foreseeing the impact of climate change on fruit trees is essential to inform policy decisions to guide the adaptation to new climatic conditions. To this end, we propose and use a phenological process-based model to assess the impacts of climate change upon the phenology, the suitability and the distribution of economically important cultivars of peach (Prunus persica), across the entire continental France. The model combines temperature dependent sub-models of dormancy, blooming, fruit survival and ripening, using chilling units, forcing units, frost occurrence and growing degree days, respectively. We find that climate change could have divergent impacts on peach production. On the one hand, blooming would occur earlier, warmer temperatures would decrease spring frost occurrence and fruit ripening would be easily achieved before the start of fall. On the other hand, milder winters would impede the plant buds from breaking endodormancy, with consequent abnormal patterns of fruit development or even blooming failure. This latter impact would dramatically shift the geographic range of sites where peach production will be profitable. This shift would mainly be from the south of France (Languedoc-Roussillon, Rhône-Alpes and Provence-Alpes-Côte d'Azur), to northwestern areas where the winter chilling requirement would still be fulfilled. Our study provides novel insights for understanding and forecasting climate change impacts on peach phenology and it is the first framework that maps the ecological thermal niche of peach at a national level.

Probability of frost occurrence based on meteorological observation in Japan

We investigated meteorological conditions for frost occurrence using routine observation data during 1991–2010 from manned meteorological stations in Japan. The majority of the frost occurrence was observed when the daily minimum air temperature was in the range of − 5 °C to 5 °C. Due to strong radiative cooling of a frosted surface, frost was observed even when the daily minimum air temperature was above the dew-point temperature at a normal meteorological observation height. We proposed an analytical scheme to evaluate the probability of frost occurrence under given meteorological conditions. This scheme was based on a probability distribution of observed frost occurrence in a three-dimensional space of meteorological variables. The obtained distribution centered on − 1.9 °C for the daily minimum air temperature and 73% for the relative humidity, with an exponential decay with wind speed. The probability was halved when the daily minimum air temperature and relative humidity change by 3.3 °C and 24 percent-points, respectively, from the distribution center. Since this scheme gives probability of frost occurrence, it is advantageous for risk assessment under given meteorological conditions. Recent popularization of gridded meteorological datasets will help to expand the potential usage of the scheme.

Developing a satellite-based frost risk model for the Southern African commercial forestry landscape

Frost is a sporadic meteorological event affecting the productivity of commercial forests in South Africa. Severe frost occurrences may cause irreversible damage to forest stands, slowing down tree growth or leading to tree mortality. Using the Moderate Imaging Spectrometer (spatial resolution: 1 km by 1 km, swath: 1 200 km by 1 200 km) night-time land surface temperatures between 2002 and 2011, this study mapped frost risk classes using six satellite-derived variables at the landscape level. These variables included calculated thresholds of minimum temperature, probability of frost occurrence, mean temperature, total number of frost days, frost duration and the frost severity index. Results show that, using an unsupervised random forest approach with partitioning around medoids, clustering was successful in mapping frost risk using eight optimal clusters. The methodology developed in this study contributes to building a robust frost-risk model to manage and mitigate forest frost damage.

Genetic control of productivity and genotypes by environments interaction for Eucalyptus dorrigoensis in southern Brazil

Background: Eucalyptus dorrigoensis withstands cold weather and the occurrence of frost, making it a potential species for improvement programs in Southern Brazil where such conditions occur. However, the use of the species is still uncommon and its genetic variability remains poorly understood.d. Results: For site called Sertão Santana, the narrow sense heritability ranged from 0.46 (H) to 0.64 (MAI) and for Lavras do Sul it ranged from 0.38 (H) to 0.47 (MAI). The coefficient of genetic variation varied from 8.1% (H) to 26.1% (MAI). There is significant genetic correlation between DBH and MAI, reaching a value of 0.97. The GxE interaction was significant and mostly complex (78.7%). The best genetic materials for each environment, along with the most stable genetic materials, were identified. In addition, a thinning simulation was performed and the genetic gains for Sertão Santana and Lavras do Sul were 12.62 and 7.43%, respectively. Conclusion: The studied populations have genetic variability that can be used in breeding programs by selecting among progenies and individuals within progenies. The GxE interaction is complex, and as such, the best genetic material should be selected independently for each site. The results of this study have practical implications for the companies and offer advances in knowledge of the species for breeding programs.

Ensemble causal modelling for frost forecast in vineyard

Being a kind of natural phenomenon, frost occurrence is influenced by environment factors with an accumulated impact. The relation between environment factors and frost event is of cause-effect governed by a process taking place in time. Having cause-affect concerned, frost forecast is a problem of complex cause-effect rather than a complicated association and problem modelling plays a key role for the success of forecast. With limited data and lack of true physical model, a well-trained model by machine learning from data is only an approximation constructed on a sub-space of problem domain. As a continued study of causal modelling in frost forecast developed previously, this paper proposes an ensemble causal modelling to compensate the performance of individual models. Such an ensemble involves models with different length of time-delay so to provide a spectrum of early alarm of frost occurrence. Experiments are done using sensor data collected from a vineyard in Hokkaido, Japan.