FROST FORECAST
(PREVENTION AND / OR MONITORING)
STATE OF THE ART
EXAMPLES OF DIFFERENT METHODS USED WORLDWIDE
CONAE
Master of Space Applications on Early Warning and Response for Emergencies
Elena Luisa Campusano Ahumada
October, 2014
2
INDEX
Content Page
INDEX.................................................................................................................................................02
ABSTRACT..........................................................................................................................................04
INTRODUCTION
1. FROST ...................................................................................................................................05
2. FORECAST OF FROST (DEFINITIONS AND METHODS) ..........................................................06
FORECASTING TECHNIQUES TO FREEZE ON THE WORLD BY CONTINENT
1. AFRICA
a. Kenya - Nyandarua, Kiambu, Murang’a and Nyeri
Application of GIS Techniques in Frost Risk Mapping for Mitigating Agricultural
Losses...................................................................................................................................08
2. AMERICA...............................................................................................................................11
a. Argentina - Buenos Aires, Córdoba, Santa Fe
Spatial Characterization of Frost and Water Stress in the Argentine Pampa Region using
Satellite and Complementary Information...........................................................................13
b. Argentina – Mendoza
Low-Power Wireless Sensor Network for Frost Monitoring in Agriculture
Research...............................................................................................................................14
c. Argentina - San Juan
Weather Station with GSM Communication for Frost Prediction and Detection of wind
Zonda....................................................................................................................................17
d. Bolivia – Andean Highlands
Downscaling MODIS-derived maps using GIS and boosted regression trees (BRT): The case
of frost occurrence over the arid Andean highlands of Bolivia............................................19
e. Chile - Region of O´Higgins
Frost Prediction Characteristics and Classification using Computational Neural
Networks..............................................................................................................................21
f. Peru – Mantaro Valley - Huayao, Jauja, Santa Ana
Calibration and Validation of Models Forecast frost in the Mantaro Valley........................23
g. Uruguay
Remote Sensing and Geographic Information Systems, some applications in
Uruguay................................................................................................................................24
3
3. ASIA
a. China – North of China
Hyperspectral characteristics of winter wheat under freezing injury stress and LWC
inversion model....................................................................................................................26
4. EUROPE
a. Finland - Sodankylä, (Northern Finland)
Detection of Soil Freezing from L-Band Passive Microwave
Observations.........................................................................................................................28
5. OCEANIA
a. Australia – Clare Valley, South Australia
Vineyard Remote Sensing – Practical Applications. Early and pre harvest remote
sensing are being applied at Taylors as tools for improving vineyard
productivity..........................................................................................................................30
b. New Zealand – Auckland
Wireless Microclimate Sensor.............................................................................................32
FINAL COMMENT..............................................................................................................................34
REFERENCES......................................................................................................................................35
4
ABSTRACT
This paper shows how a different part of the world is predicted and / or monitors frost. It is
disclosed at least one case per continent, whose antiquity the study has not been published after
the year 2006, since the aim of this paper focuses on raising awareness of the current technology
with which it has worked.
The concern of this problem arises mainly due to the large amount of economic losses that occur
in agriculture, especially when the type of crop is the main income of the country concerned that
is why when this weather event happens, both farmers and government authorities should take
action.
As technology advances this problem is no longer covered when the loss is generated and if this is
not done proactively (or at least being monitored) in order to get to predict this event. It is
important to study and understand the previous cases, in which every year frosts occur, thus,
knowing the characteristics of the territory and how this event affects the different cultures
present in this territory.
Each country, depending on the economic resources that have been able to invest to cover this
issue in the best way possible, by studying various alternatives for early detection of frost is
achieved, and thus to avoid the loss of crops. There are different types of technology and are used,
which give satisfactory results have been avoiding crop losses because with early detection frost
can take action.
5
INTRODUCTION
Before starting to cases by country, it is necessary to consider several concepts, such as those
presented below:
1. FROST
It is designate frost on fruit growing at the climate event in which the ambient temperature
surround the plant or over the plant organ is under the ranges that allow the normal activity of the
plant.
Commonly frost is associated with the concept of below 0°C (subzero) ambient temperature, since
it is at this temperature at which the metabolism of a plant starts to be slowed down and
moreover is the temperature at which begins the liquid water to switch to its solid state.
Moreover, freeze concept is intimately related to freezing, because at temperatures below 0ºC
any tissue or plant organ begins to freeze.
FROST IN FRUIT
SERGIO ANTONIO TOLEDO VIVAS
AGRONOMIST - CATHOLIC UNIVERSITY VALPARAISO, CHILE
WWW.ECOPLANT.CL
According to its origin frost are classified into the following types:
For advection: It usually occurs in the fall or winter, and their appearance can be white or
black depending on the abundance or scarcity of moisture. Set when a mass of cold air
below 0 ° C, extends over an area, which may well be glen, valley or plain, with winds
exceeding 15 km / h, with negative temperature gradient without investment thermal.
Extensive grounds are affected, up to 100 km2, without the intervening clouds. Obviously
plants are cooled by contact.
For Radiation: Occurs in spring, autumn and winter. This is generated when the lower
layers of the atmosphere cools due to heat loss from the earth by radiation at night,
especially in the winter to be the longest nights. This is established in the presence of
temperature inversion, with calm winds and clear skies. Occurs when the relative humidity
of soil adjacent to the air is very low, which decreases more when a dry wind that takes
presented. This type of frost mainly affects flowers and vegetables that are grown in the
valleys and basins and canyons near the mountains.
Evaporative: It occurs in autumn and winter and is due to the evaporation of water from
the surface vegetation, so the air is dry and the relative humidity low, the change of state
of water from liquid to gas causes heat appears from plants and then this is cooled and
frozen.
Mixed: This is generated when the advection and radiation phenomenon are present
together; i.e. not soil inversion and loses heat by radiation.
6
The Frosts are also classified by their visual appearance, and such can be:
White: It occurs when there is high humidity and the temperature is near or below 0 ° C.
The Ice is present on the plants.
Black: This ice is not accompanied by the formation of ice, but the plant organs of the
plants turn a blackish color due to the destruction caused by the cold.
FORMULA FOR PREDICTING FROST
TOMÁS FERNÁNDEZ ÁBRICA
POSTGRADUATE IN GEOGRAPHY, NATIONAL UNIVERSITY OF MEXICO (UNAM)
2. FORECAST OF FROST (DEFINITIONS AND METHODS)
Assessing the value of frost forecasting involves complex decision analysis, economics and using
conditional probabilities. Accurate prediction can potentially reduce frost damage by frost, as it
provides an opportunity for the farmers to prepare against them.
While decision analysis is used in many disciplines, its application to the prediction of frost is
scarce. Banquet Articles, Halter and Conklin (1976) and Katz, Murphy and Winkler (1982) discuss
the use of decision analysis to evaluate the cost effectiveness of frost prediction relationship. Katz,
Murphy and Winkler (1982) investigated in detail the value of the prediction of frost in the Yakima
Valley in Washington State, USA. The authors have used the processes of decision-making on a
Markov model to structure the problem in identifying possible actions, events and consequences.
These authors evaluated the usefulness of three methods for predicting frost:
The first, by calculating the conditional standard deviation for the prediction using only
climate data.
The second corresponds to official forecasts conducted by the National Weather Service in
the USA.
The third corresponds to a perfect prediction where the minimum temperature prediction
is always correct. The standard deviation is "conditional" because it is based on an
assumed level of prediction accuracy.
FROST PROTECTION: PRINCIPLES, PRACTICE AND ECONOMICS
SERIES ON ENVIRONMENT AND NATURAL RESOURCES MANAGEMENT
ENVIRONMENT, CLIMATE CHANGE, BIOENERGÍA (MONITORING AND EVALUATION)
UNITED NATIONS FOOD AND AGRICULTURE (FAO)
ROME, 2010
The forecast of frost at time for the farmer and the cowman (mainly) and to alert other services
(NHTSA), to prepare the means available, it is not easy because of the large number of factors
involved, such as location, extent, topography, type and density of vegetation, trees, crops,
whether it is a valley or if instead the place is at a certain altitude.
It consists of the predetermination of the negative temperature that can be achieved overnight, its
minimum value and duration.
7
A good help for this forecast assumes the availability of climatological data means frost-free
period and in particular the first frost of fall and last spring, for special vigilance in that period.
For more than a century are being used empirical formulas for each particular place and all
different. It seems that the best are those that establish hygrometric equations relating the
minimum temperature to the dew point, relative humidity and certain statistically fitting
parameters obtained for each location.
PREDICTING RISK OF FROST
TIEMPO.COM FORUM - GENERAL WEATHER FORUM - GENERAL WEATHER - PREDICTING RISK OF FROST.
MARCH 2007
Today frosts are forecast:
By means of maps shown by isolines frost the record thereof.
Using the Weather Service, when they make forecasts of minimum temperatures.
By means of the hygrometer, as this gives the degree of humidity; if it is high, it is unlikely
that a frost is generated.
Through Evaporimeter because if evaporation is slow, there is higher humidity and the risk
is less than the frost occurs.
Through the Contact Thermometer, when this reaches the temperature approaches
freezing establishes an electrical contact which causes an alarm bell work.
Dew point temperature. If at night the temperature is high, then dew occurs and not frost.
Sometimes, maybe can generate fog.
Using models, e.g. Neural Networks.
FORMULA FOR PREDICTING FROST
TOMAS FERNANDEZ ABRICA
POSTGRADUATE IN GEOGRAPHY. NATIONAL UNIVERSITY OF MEXICO (UNAM)
8
FORECASTING TECHNIQUES TO FREEZE ON THE WORLD
BY CONTINENT
1. AFRICA
a. Kenya - Nyandarua, Kiambu, Murang’a and Nyeri
Application of GIS Techniques in Frost Risk Mapping for Mitigating Agricultural Losses
Susan Malaso Kotikot & Prof. Simon Onywere
Farmers in Nyeri, Nyandarua and Lari have in the past suffered huge losses amounting to millions
of shilling as a result of an overnight frost. As a result, USAID, Kenya Horticultural Competiveness
project has taken management measures to assist the farmers at the foot of Mt. Kenya by
encouraging them to plant trees, diversify crops, adopt greenhouse technologies and practice drip
irrigation (USAID-KHP Monthly Bulletin January 2012 paper).
Most frost events occur during clear and calm nights (Kalma, 1983). They are influenced by terrain
aspects combined with meteorological factors such as wind, clouds and humidity. An experiment
by Laughlin, et al., (1985) verified that cloud cover, wind and elevation, could be combined in an
equation to give a good prediction of the expected minimum ground temperature at any particular
site at night. Because cold air flow downs lope, much like water, the valley floors and lower
portions of the slopes are colder (Richards, 2003). This makes frost risk assessments relevant for
agricultural climatic hazard mapping.
The current paper shows the role satellite imagery technologies can play in frost risk mapping.
The study area is located within the Kenyan highlands, one of the major water towers in Kenya
and consisting the Aberdare forest and national park. Spatially, the area covers four counties,
Nyandarua, Kiambu, Murang’a and Nyeri counties. A 15 km buffer was however introduced in
order to include variations in altitude and topography beyond the high altitude Aberdare ranges.
The topography is characterized by mountain ranges, strong local variations and feature elevations
resulting in an undulating topography including valley systems. This makes the region suitable for
frost risk assessment since altitude is an important factor as far as choice of crops, temperature
variations; rainfall and wind patterns are concerned (Oludhe, 2008).
Main crops include maize, beans, peas, potatoes, tea and coffee.
COUNTRY CROP MATERIALS AND
METHODS
PRODUCT DEVELOPMENT PLACE
Kenya
Nyandarua,
Kiambu,
Murang’a
and Nyeri
counties
2013
Maize,
Beans,
Peas,
Potatoes,
Tea ,
Coffee
MODIS LST
(Terra and Aqua)
LandSat and
Aster 30m DEM
Frost risk
mapping
Kenyatta University,
Department of
Environmental Planning
and
Management
9
Satellite data MODIS (Moderate Resolution Imaging Spectroradiometer) Land Surface
Temperature was used in the study. The MODIS sensor is onboard the Terra and Aqua satellites
and provides twice daily observations; daytime and nighttime (Zhengming, 1999). Nighttime data
sets for year 2000 to 2013 were used. In addition to the MODIS LST, temperature data records (to
calibrate LST) and rainfall data was collected from weather stations within and 4 surrounding the
study area to investigate the influence of available moisture on frost occurrence. Other data sets
collected include Landsat imagery from the USGS, Aster 30m DEM, and frost occurrence data. The
quality of the data was controlled to exclude all the obvious suspicious data.
Using ENVI 5 software, minimum temperatures per pixel was extracted from the daily nighttime
datasets to create continuous coverage of monthly datasets in order to fill up potential data gaps
on the daily acquisitions. The frequency of occurrence for minimum temperatures below 273K was
computed through time frequency analysis to obtain a frequency map. Further classification of the
datasets was done using ENVI 5 decision tree classifiers to discriminate various levels of frost risk;
very severe frost (<250 K), severe frost (250 to 260), moderate frost (260 to 270 K), minor frost
pockets (270 to 280 K) and regions of no frost (>280 K). These are the levels used by the Kenyan
meteorological department. The determination of frost category threshold was based on the
frequency and level of minimum temperature occurrence evaluated since the year 2000 to 2013.
The frost hotspot map above represents occurrence and distribution of minimum temperatures
while the frequency map represents the number of months herein translated as the number of
times that temperatures below 00C have occurred. The relationship is such that the extent of the
region most frequented by low temperatures, correspond to the frost hotspots. There is a region
above a particular elevation where temperatures are constantly low enough to form frost. These
are at the peaks of Mount Kenya and the Aberdares which are often covered by snow.
10
Conceptual Framework
11
2. AMERICA
COUNTRY CROP MATERIALS AND/
OR METHODS
PRODUCT DEVELOPMENT PLACE
Argentina
Buenos
Aires,
Córdoba,
Santa Fe
2006
Corn,
Soybean,
Sunflower,
Wheat
NOAA-AVHRR
(Advanced Very
High Resolution
Radiometer)
Frost Images
percentage (by
region and period)
National Institute of
Agricultural Technology
(INTA), Argentina
Argentina
Mendoza
2014
Vine Wireless Sensor
Network (WSN)
Taking and storing
periodic
temperatures.
Remote access to
sensor status
reporting tools
GridTICs Laboratory,
Department of Electronics,
National Technological
University, Mendoza ,
Argentina
Regional Faculty.
Faculty of Agricultural
Sciences, National
University of Cuyo,
Argentina
Argentina
San Juan
2012
General Weather station
connected to the
cellular network
Send Short Message
Service (SMS) of
alarms generated
Institute of Automation,
National University of San
Juan, Argentina
Bolivia
Southwest
of the
Bolivian
highlands
2010
Potato,
Quinoa
Meteorological
ground data
available in
three
meteorological
stations (Salinas of
Garci Mendoza,
Irpani and Jirira),
MODIS Images
(Terra and Aqua)
and DEM SRTM
Maps of MODIS frost
occurrence in
southern Bolivia
Research Institute for
Development (IRD), France;
Centre D´Ecologie
Fonctionnelle &Evolutive,
France
Chile
Region of
O’Higgins
2009
Vine Data of weather
stations CRIA
(Regional
Information Center
Computational
Multilayer
Perceptron (MLP)
model for Frost
Geoinformatics Research
Centre, School of
Computing and
Mathematical Sciences,
12
Agrometerological)
of the Ministry of
Agriculture of
Chile,
Computational
Neural Network
(CNN) methods,
Kohonen’s Self-
Organizing Map
(SOM)
Prediction Auckland University of
Technology, New Zealand,
School of Computing and
Informatics, Faculty of
Engineering Catholic
University of Maule, Chile
Perú
Mantaro
Valley
Huayao,
Jauja,
Santa Ana
2006
Corn,
Potato
Mechanistic
Models (Cellier and
Lhomme),
Empirical Model
(multiple linear
regression
equations)
Freeze Forecast
Models
National Service of
Meteorology and
Hydrology (SENAMHI)
Uruguay
2014
General NOAA-AVHRR Probability of
Occurrence of Frost
National Institute for
Agricultural Research
(INIA),
Agroclimate and
Information Systems Group
(GRAS)
13
a. Argentina - Buenos Aires, Córdoba, Santa Fe
Spatial Characterization of Frost and Water Stress in the Argentine Pampa Region using Satellite
and Complementary Information
Straschnoy, J. V.; Di Bella, C. M.; Jaimes, F. R.; Oricchio, P. A.; Rebella, C. M.
The present work presents a methodology that allows the spatial characterization of the
agricultural production losses by frost or water stress in different areas of the Pampa’s Region
using satellite and complementary information.
For each period images of Absolute Minimum Surface Temperature and Frost Percentage were
generated. The analysis of the images allowed to observe the fact that the frost events were more
intense and frequent during the Late Frost periods.
This methodology can be applied to different areas and it could be integrated with other
information sources (e.g. topography, climate forecasts). In this way, is expect to improve the
analysis and evaluation of the risk related events affecting the regional agriculture production.
Remote sensing is a very advantageous tool in terms of cost benefit as it allows large areas of homogeneous data and performs multi-temporal analysis. From the spectral information captured in the thermal infrared is possible to estimate the surface temperature (Ts), which allows for example to study the occurrence of frost (e.g. Di Bella et al., 1997). In development of the climate risk maps information from various disciplines such as meteorology, climatology, remote sensing, soil science, crop management, plant physiology and production technologies integrated (http://www.ora.gov.ar/). This paper develops the "Risk Maps - Satellite Data" component of the project, the specific objective was to develop a methodology based on satellite information to establish indicators to measure the risk of frost at the regional level to be integrated with information from other project components. The study area included five regions representing a significant portion of the land used for domestic agricultural production: three regions corresponded to the province of Buenos Aires (north, central and west), one south of Santa Fe and other southeast Cordoba. The growing season for a crop available is defined by the time of occurrence of frost and the different sensitivity to them which presents each of the species. With the aim of establishing a planting date to reduce the risk of frosts in stages defined as critical for each crop, is very important to establish the mean dates of first and last frost, as well as anomalies that can happen (frost early and late) (Otegui and Pereyra, 2003). For example, frost present during the months of January to May cause damage to summer crops sown late, complicating the generation of yield components. As for winter crops, the growth stage during the early frost is not affected compared to a climatic anomaly. For the purposes of this study, two associated risk indicators of ice were generated. The first consisted of the minimum temperature recorded at each pixel, and the second was calculated as the percentage of available dates between, for which lower surface temperatures were recorded at 0 ° C. The images used to determine the risk indicators of ice were obtained from NOAA-AVHRR (Advanced Very High Resolution Radiometer) satellite, whose nominal spatial resolution is 1.1 x 1.1 km2 during the period 1996-2002. In accordance with the general objectives of the project as to estimate the risks of extreme, weather events likely to affect agricultural production, two periods of interest were defined: a) Early Frost (January to May) and b) Delayed Frost (September to December ). To select the images to be used the minimum temperatures recorded at weather stations located within the study area (Pehuajó, 9 de Julio parchment Laboulaye and Rosario) during the periods considered were
14
analyzed, and those dates were identified in which there was lower minimum temperatures at 3 ° C. Finally, the images were considered corresponding to the dates they met the following requirements:
Zero or low cloudiness;
Representation of the study area (> 75%);
No entry errors. Which were treated by the following procedure:
Importing the bands 1 (0.58 to 0.68 microns - red), 2 (0.72 to 1.10 microns - near infrared), 4 (10.30 to 11.30 microns - thermal infrared), and 5 (11.50 to 12.50 microns - thermal infrared) satellite NOAA AVHRR binary format Erdas ©;
Atmospheric correction of images using a method called split window (Sobrino et al, 1991.)
Defining criteria for filtering erroneous or missing data due to errors in capturing or processing and geometric correction of the images;
Screening of all images to the coordinate system Lat / Lon, because they do not originally have any projection.
Based on the surface temperature (Ts) defined three categories:
Cloud: pixels with less than five degrees Celsius (<-5 ° C) values;
No Frost: pixels with above zero degrees Celsius (> 0 ° C) values,
Freezing: pixels with values between -5 ° C and 0 ° C. To evaluate the risk of frost regions of study, two indicators were developed. In the first step, images Absolute Minimum Surface Temperature (Ts) was calculated for each pixel the minimum value of Ts among all dates available for each region. The second indicator, called Percent Glass, was generated by calculating the quotient between the number of dates in which the temperature was above -5 ° C and below 0 ° C (Category 3 Frost) and number of dates on which the temperature was above -5 ° C (categories 2 and 3, No Frost and Frost), i.e., from the ratio Frost / (+ Frost No Frost). Thus, values between 0 and 1 is obtained, which multiplied by 100 indicate the percentage of dates (from available) in which the surface temperature (Ts) was lower than 0 ° C (category Frost) in each pixel, regardless of the presence of clouds.
15
Frost Images percentage by region and period
16
b. Argentina - Mendoza
Low-Power Wireless Sensor Network for Frost Monitoring in Agriculture Research
Ana Laura Diedrichs, Germán Tabacchi, Guillermo Grünwaldt, Matías Pecchia, Gustavo Mercado
and Francisco González Antivilo
This work presents the development of a wireless sensor network (WSN), based on IEEE-802.15.4,
in order to be used for frost characterization in precision agriculture by measuring temperature.
The key objective is to reduce the power consumption of the network to the minimum, allowing
several measurement points per node and the remote monitoring of the sensors behaviour. For
the communication interface between a WSN node and the sensors, we have developed a serial
protocol inspired in SDI-12. Preliminary results show a low-cost and low-power WSN. The user can
access and use the data for agronomic research.
Precision agriculture (PA) uses decision support systems to manage the crops. The ambiental
parameters needed (e.g., temperature, humidity) can be measured using sensors.
Wireless sensor networks (WSN) consist of random or planned spatially distributed nodes, also
called motes, each of which is equipped with sensors, memory for storage, microprocessor for
computing their decisions, batteries for energy supply and are able to communicate wirelessly
with other nodes in a short-range. Thanks to the different technology advances, today it is possible
to build multifunctional sensor devices which are smaller, cheaper, able to communicate to each
other and can operate with low-power, in contrast with the traditional data loggers.
The network design focuses on achieving a local requirement of environmental parameter
monitoring in times of frost. The damage caused by the frost takes place when the temperatures
are below than a tolerable limit for the plants. Each phenological state, e.g. flowering, has a
variable cold hardiness, so the lethal temperature is also variable. Freezing climatic events are the
most dangerous, because they affect a large land surface. Mendoza is not an exception. According
to the Instituto Nacional de Vitivinicultura (INV), in 2013 the loss of the vine crop reached up to
27%. Big part of that lost of yield was during the early spring. In order to study the micro-climate
phenomenon of frost in Mendoza, a WSN was placed between the vineyards, where the sensors
are distributed vertically as well horizontally. Increasing the spatial sensor resolution is so
important because the air temperatures change vertically as horizontally, and the plant has also
different cold hardiness in the organs like trunk, flowers, and shoots. For a better understanding of
the phenomenon a precise measurement is necessary.
The protocol used for communication in the WSN is IEEE 802.15.4, which covers physical layer and
MAC (medium access control) sublayer and allows only peer to peer or star as possible topologies
in a wireless personal area network. Motes can be reduced function devices (RFD) or full function
devices (FFD). The RFD is the sensor node and has a low power duty-cycle and the FFD is the
coordinator of the network. Carrier sense multiple access with collision avoidance (CSMA-CA) is
used to control medium access, some motes are placed in the greenhouse and the others in the
vineyard.
17
Topology of the wireless sensor network
c. Argentina - San Juan
Weather Station with GSM Communication for Frost Prediction and Detection of wind Zonda.
C. Godoy, C. Carletto, J. C. Correa and A. Lage
The development of a weather station connected to the cellular network, which automatically
generates alarm signals to the user when it detects Zonda wind or no chance of frost is presented.
The frost prediction is performed with a least squares approximation sequence based on
distributed temperature measurements in the field and placed in the sights wireless sensors. The
team also has other sensors for different meteorological variables that the user can access
remotely at any time via a simple text message protocol.
For fruit growers Tulum Valley (San Juan) have a significant importance and late frosts Zonda
winds. Both phenomena occur if at the time of flowering and fruit set produced considerable
damage in production. "Late frosts" occurring at the beginning of spring, it is called when the crops
begin their growth cycle. The Zonda wind is a very hot dry wind coming from the Cordillera,
sometimes it manifests itself suddenly with a violent gusts reaching up to 120 Km / h.
Frosts typically occur in the early morning (before sunrise) and is important for the producer to be
forewarned and prepared to act (Snyder et al., 2010). It is for this reason that you must have an
alarm a few hours before the freeze occurs.
The aim of this work is to have a measurement system that allows for variable weather short-term
prediction of frost early detection of the arrival of Zonda winds and detecting temperature
threshold values, to generate corresponding alarms. In addition, all functions and information
collected by the system are available where there is GSM network coverage.
The team developed consists of a master station and one or more wireless stations. The control
and communication of all stations are in charge of microcontrollers (Microchip, 2003).
18
The main station has sensors for temperature, humidity, pressure on, rain gauge, wind speed and
direction. This station is responsible for making the connection to the wireless stations. It is
powered with 220V network and is responsible for attaching the body of a text message all the
information gathered, when required and send via SMS alarms generated.
A main station places a SIM card, which can belong to any cellular service provider company. The
main station receives all the information from its own sensors and wireless stations and performs
data processing and this is where it is analyzed whether the conditions exist for the alarms are
activated and is also responsible for responding to user request or modification of the system
configuration parameters. This program also allows discriminating the areas in which these
phenomena are occurring (Ibrahim, 2008).
With alarms provided by the system, the producer can be activated with a heating time, such as
fuel heaters or fires can do around crops. In the case of wind Zonda, can trigger water pumps
water on crops pulverized lowering its temperature by preventing them from burning. Remote
access for monitoring the evolution of meteorological variables is a useful feature of this project,
since the farmer can make monitoring the weather conditions in your field from anywhere with
access to the GSM network.
General diagram of the developed system
19
d. Bolivia – Andean Highlands
Downscaling MODIS-derived maps using GIS and boosted regression trees (BRT): The case of
frost occurrence over the arid Andean highlands of Bolivia
Robin Pouteau, Serge Rambal, Jean-Pierre Ratte, Fabien Gogé, Richard Joffre, Thierry Winkel
Frost risk assessment is of critical importance in tropical highlands like the Andes where human
activities thrives at altitudes up to 4200 m, and night frost may occur all the year round. In these
semi-arid and cold regions with sparse meteorological networks, remote sensing and topographic
modeling are of potential interest for understanding how physiography influences the local
climate regime. After integrating night land surface temperature from the MODIS satellite, and
physiographic descriptors derived from a digital elevation model, explored how regional and
landscape-scale features influence frost occurrence in the southern altiplano of Bolivia. Based on
the high correlation between night land surface temperature and minimum air temperature, frost
occurrence in early-, middle- and late-summer periods were calculated from satellite observations
and mapped at a 1-km resolution over a 45,000 km² area. Physiographic modeling of frost
occurrence was then conducted comparing multiple regression (MR) and boosted regression trees
(BRT). Physiographic predictors were latitude, elevation, distance from salt lakes, slope steepness,
potential insolation, and topographic convergence. Insolation influence on night frost was tested
assuming that ground surface warming in the daytime reduces frost occurrence in the next night.
Depending on the time period and the calibration domain, BRT models explained 74% to 90% of
frost occurrence variation, outperforming the MR method.
Low air temperature is one of the most important factors controlling vegetation zonation and key
processes such as evapotranspiration, carbon fixation and decomposition, plant productivity and
mortality in natural and cultivated mountain ecosystems (Chen et al., 1999; Nagy et al., 2003).
Depending on vegetation structure, landscape position or soil properties, frost can damage plant
tissues thus affecting forest, pasture and crop productivity (Blennow & Lindkvist, 2000).
In the southern Andes, sparsely vegetated areas juxtaposing extended flat plains around salt lakes
and steep slopes on the cordilleras and volcanos, display semi-arid and desert landscapes largely
dominated by terrain structure. Subjected to the night/day and sunlit/shaded slope contrasts
characteristic of the mountain climate, this environment is well suited for examining the influence
of regional and landscape-scale physiography on the local climate regime, and particularly frost
occurrence.
The present study resorts to an advanced form of regression, the boosted regression trees (BRT).
BRT use the boosting technique to combine large numbers of relatively simple tree models to
optimize predictive performance. BRT have been used successfully in human biology (Friedman &
Meulman, 2003), land cover mapping (Lawrence et al., 2004), biogeography (Parisien & Moritz,
2009), species distribution (Elith et al., 2008), and soil science (Martin et al., 2009). They offer
substantial advantages over classical regression models since they handle both qualitative and
quantitative variables, can accommodate missing data and correlated predictive variables, are
relatively insensitive to outliers and to the inclusion of irrelevant predictor variables, and are able
to model complex interactions between predictors (Elith et al., 2008; Martin et al., 2009). Though
direct graphic representation of the complete tree model is impossible with BRT, the model
interpretation is made easy by identifying the variables most relevant for prediction, and then
20
visualizing the partial effect of each predictor variable after accounting for the average effect of
the other variables (Friedman & Meulman, 2003).
The aims of the present work were:
To explore how regional and landscape-scale physiography influence frost occurrence in
Andean highlands through integration of field and remote sensing data, digital terrain
analysis, and GIS;
To downscale regional frost occurrence maps at a level relevant for farming and land
management decisions using BRT models.
This study was focused on the austral summer period, from November to April, when frost holds
the greatest potential impact for local farming activities.
The study area was located at the southwest of the Bolivian highlands, near the borders of
Argentina and Chile. This region, boarded by the western Andes cordillera, is characterized by the
presence in its centre of a ca.100×100 km dry salt expanse, the Salar of Uyuni, while another salt
lake, the Salar of Coipasa, lies at the north of the study area. The landscapes show a mosaic of
three types of land units: more or less extended flat shores surrounding the salt lakes (elevation
ca. 3650 m) and an alternation of valleys and volcanic relieves (culminating at 6051 m) in the
hinterland. The native vegetation of this tropical Andean ecosystem, also known as puna, consists
of a mountain steppe of herbaceous and shrub species (e.g. Baccharis incarum, Parastrephia
lepidophylla, and Stipa spp.) (Navarro & Ferreira, 2007) traditionally used as pastures but
progressively encroached by the recent and rapid expansion of quinoa crop (Chenopodium quinoa
Willd.) (Vassas et al., 2008).
These particular thermal conditions lead to high frost risks throughout the year. Advections of air
masses from the South Pole represent only 20% of the observed frosty nights (Frère et al., 1978)
and are four times less frequent in the summer than during the winter, when the intertropical
convergence zone goes northward (Ronchail, 1989).
Therefore, the main climatic threat lies in radiative frost, occurring during clear and calm nights. As
reported by local peasants, frost occurrence shows a strong topographical and orographical
dependence, as well as a marked seasonality. This seasonality lead us to split the active vegetation
period into three time periods characterizing the mean regional climate dynamics in the summer
rainy season:
November–December when precipitation and minimum temperature rise progressively,
January–February when precipitation and temperature are at their maximum, and
March–April when both begin to decrease
The regional BRT clearly outperformed the MR models, the latter being affected by strong non
linearities in the high frost occurrence range, showing its poor predictive capacities in the early
and late summer periods. On the other hand, the regional BRT negligibly overestimated the
satellite observations with practically no bias whatever the time period. With an error generally
less than 8% on predicted frost occurrence values, the regional BRT thus appear suitable for
predicting frost occurrence from physiographic variables alone.
21
One-kilometer resolution maps of MODIS-derived frost occurrence in southern Bolivia in three
successive time periods (the color scale at the right shows the frost probability)
e. Chile - Region of O´Higgins
Frost Prediction Characteristics and Classification using Computational Neural Networks
Philip Sallis, Mary Jarur, and Marcelo Trujillo
The effect of frost on the successful growth and quality of crops is well understood by growers as
leading potentially to total harvest failure.
Studying the frost phenomenon, especially in order to predict its occurrence has been the focus of
numerous research projects and investigations. Frost prone areas are of particular concern. Grape
growing for wine production is a specific area of viticulture and agricultural research. This paper
describes the problem, outlines a wider project that is gathering climate and atmospheric data,
together with soil, and plant data in order to determine the inter-dependencies of variable values
that both inform enhanced crop management practices and where possible, predict optimal
growing conditions. The application of some novel data mining techniques together with the use
of computational neural networks as a means to modeling and then predicting frost is the focus of
the investigation described here as part of the wider project.
22
The use of contemporary methods for mining data and knowledge discovery to classify and
quantify the variables that are identified as being critical for the production of robust
computational models is seen as both appropriate and necessary for this problem domain. This
paper describes early results from work in frost prediction being undertaken by academic
researchers at Auckland University of Technology in New Zealand, Universidad Católica del Maule
in Chile and Universidad de Talca, Chile.
In the research described in this paper, some derived values, especially the calculation of dew
point (the point in time just before frost falls) are used and the tendency (such as a falling
barometer and cloud cover, which determines solar intensity) are included to enrich the
combinatorial analysis of the data.
Computational Neural Network (CNN) is used method to assist with the analysis and modeling of
this problem. CNN technologies enable to develop highly non-linear mathematical models with a
relatively low level of computational complexity.
With this model and data processing can be analyzed large data sets still retain the integrity
between the dependent variables while construct scenarios is done for different result outcomes.
The complexity of data dependencies is represented by a network of relationships, trainable to
adapt to new data as it is ingested into the model without losing the intrinsic value of the variable
dependencies. By reducing the relationship complexities it can reach a network size that is deeper
in ‘meaning’ than the superficial data set might reflect and thus, be discovered the extent to which
the dependencies (factors relating to frost prediction in this case) really exist. The identification (or
classification) of the variables present in such a model, using the Perceptron Multi-layer (MLP)
method, is not in itself a sufficient analytical framework for complex data sets such as for the frost
prediction problem. Using Kohonen’s Self-Organizing Map (SOM) techniques in conjunction with
MLP does establish an appropriate level of participation by the variables in the phenomenon being
studied. In this respect, it has been used Neucom [www.kedri.aut.ac.nz] to implement a
combinatorial approach to building the final model from the findings that is obtained through
other data mining techniques. It propose this as a robust analytical platform for application to the
frost prediction problem.
It is not sufficient to only consider the variables of humidity and temperature for the prediction
process and therefore, to consider the dependency of other variables, such as precipitation,
atmospheric pressure and wind velocity.
Concept Proposal for the analysis of Weather Data using Self-Organizing Maps (SOM) in order to
develop a Computational Multilayer Perceptron (MLP) model for Frost Prediction
23
f. Peru – Mantaro Valley - Huayao, Jauja, Santa Ana
Calibration and Validation of Models Forecast frost in the Mantaro Valley
Janeet Sanabria, Jerónimo García, Jean-Paul Lhomme
Today, with available technology, agricultural activity has various models forecast frost that
mitigate or control the effects of this phenomenon agroclimatic; however, for hill farmers benefit
from this technology available, it is necessary to identify, calibrate and validate models to
determine which one will be used with a good level of reliability because the available models
were developed to climatic and topographic conditions different.
The study aims to calibrate and validate mechanistic models Cellier and Lhomme, and identify and
validate an empirical model in order to predict crop frost level in the Mantaro Valley.
For this, first the models were calibrated and validated and second forecast models appropriate
frost defined.
Cellier Model: The model is based on the transfer processes of energy conservation
between the ground and the lower atmospheric layers. This model predicts minimum air
temperature at different levels from 0 to 300 m, in clear sky conditions.
Lhomme Model: The model describes the transfer of mass and energy in static equilibrium
of the system soil - plant - atmosphere, at the end of the night when the minimum
temperatures usually occur. Energy transfers within the air are based on the theory K (air
transfers at a particular time). This model predicts the temperature of the areal elements
of a crop and topsoil.
Empirical Model: was determined using the developed by García (1993) PROTMI software.
He poses in nocturnal cooling process two controllable factors: radiation balance factor
and factor atmosphere greenhouse. Also considered most influential variable in the night
cooling is the minimum temperature of the previous day to be forecast (T min-1), which
will indicate the base or initial energy available in the environment. This model will predict
the minimum air temperature at level instrument meteorological shelter (1.60 m).
The first phase consists of the calibration or analysis and interpretation of physical models;
identify which parameters are having greater influence; then estimate all parameters and adjust to
those deemed necessary by change in values. In the empirical model this phase consisted of
determination of empirical equations for predicting minimum temperatures for both the
beginning (September to December) and the end of the campaign agricultural (February to May),
the multiple correlation analysis was used with variables predictors (minimum temperature of a
previous day, air temperature bulb dry and wet bulb and wind speed).
The second phase or validation was to compare the data of the outputs of the models with field
measured data. For which they were used the statistical indicators such as the residual means
square error (RMSE, measured global data scatter around the regression line), the bias (B,
measured the mean difference of the outputs to measurement)
24
g. Uruguay
Remote Sensing and Geographic Information Systems, some applications in Uruguay
Publication National Institute for Agricultural Research (INIA)
Remote sensing or remote sensing can be defined as "the use of sensors for acquiring information
about objects or phenomena without contact between them.”These sensors may be capable
photographic or optoelectronic recording systems images as the energy emitted by distant objects
and can be mounted in different platforms such as aircraft, satellites, etc.
Remote sensing allows the study of natural resources and vegetation systems agricultural, being
able to obtain accurate information in near real-time status sector. Furthermore, Geographic
Information Systems (GIS) allow the integrated analysis remote information, and cartographic data
bases georeferenced sensors.
The National Institute for Agricultural Research (INIA) through the Irrigation Group Agro-Climate,
Environment and Agriculture Satellite (GRAS) is developing applications using remote sensing and
GIS applied to the sustainable use of natural resources in the agricultural sector.
Applications of Remote Sensing: According to the resolution of the sensors used most important
applications that GRAS is developing, can be grouped into two areas of study:
NOAA-AVHRR Images AVHRR images from NOAA satellites are of low spatial resolution,
1km x 1km pixel size medium, and high temporal resolution of daily visit frequently, and
allow studies at the country level, departments or regions. Processed the following
variables:
Vegetation index (NDVI)
Frosted Surface
Evapotranspiration
LANDSAT 5 and 7: The images captured by the Landsat satellites are high spatial
resolution, 30 x 30 m pixel size medium and low temporal resolution with a frequency of
visit of each 7 days. Each image covers an area approximately 180 by 180 km.
The higher spatial resolution of this satellite, allows studies at the departmental level,
police sections, farms and small production units in particular. In the Today, processes
images received GRAS for:
Land Use resource
Determination of the use of natural resources
Variability studies
Daily monitoring of the surface temperature. Thus determines the impact of frost, with emphasis
on the early and late occurring, the effects of these on crops and pastures, identifying the affected
areas.
25
Early frost
Frost Forecast
Powered by CPTEC Brazil and available on the website of the GRAS Unit www.inia.org.uy/gras
26
3. ASIA
a. China – North of China
Hyperspectral characteristics of winter wheat under freezing injury stress and LWC inversion
model
Huifang Wang, Jihua Wang, Qian Wang, Naizhe Miao, Wenjiang Huang, Haikuang Feng, Yingying
Dong
Winter Wheat Freezing Injury (WWFI) is one of the meteorological disasters causing severe yield reduction of winter wheat in the north of China. It includes freezing injuries through winter and abnormal cold spells in early spring before jointing. Freezing injury of various extents might cause similar large-scale death or growth delay of crops, causing significant reduction in crop yield. The occurrence of freezing injury of winter wheat can cause irreversible loss in production in the winter, so monitoring of freezing injury stress is of great significance. Leaf Water Content (LWC) is one of the important indexes in the diagnosis of freezing injury of wheat. When the winter wheat is under freezing injury stress, the most obvious feature is the loss of water and green color in the leaves, and the tip and blade of the leaf become dry, or even shed. With the development of remote sensing technology, extraction of the physiological and biochemical changes of crops through hyperspectral data has distinct advantages, especially through the estimation model of physiological and biochemical parameters built with the hyperspectral information. The impact and degree of crops under environmental stress can be easily and quickly determined by the model. The study of leaf spectra and qualitative and quantitative research on freezing injury hyperspectral characters will lead to understanding winter wheat hyperspectral characters and freezing injury mechanisms. The relationship was studied between hyperspectral characteristics, of the freezing injury stress of winter wheat in the late jointing stage period with the simulated artificial frost chamber, and chlorophyll content, and the result shows a significant correlation between them. The correlation between the vegetation indexes and four physiological and biochemical parameters of 5 crops were built studied based on the hyperspectral information, and the narrow waveband information of the physiological and biochemical sensitive parameters was extracted, which demonstrates that hyperspectral information can reflect the physiological and biochemical information. The frost injury characteristics of winter wheat in the booting stage through the hyperspectral infrared band characteristics were studied. Such studies suggest that the monitoring of the environmental stress of crops through hyperspectral remote sensing is practical and feasible. In this paper,
COUNTRY CROP MATERIALS AND/
OR METHODS
PRODUCT DEVELOPMENT PLACE
China
North of
China
2012
Wheat Hyperspectral Data Leaf Water Content
(LWC) Inversion
Model
Beijing Research Center for
Information Technology in
Agriculture Beijing, China;
Institute of Agricultural
Remote Sensing &
Information System
Application Zhejiang
University Hangzhou, China
27
hyperspectral characteristics of winter wheat from the seedling stage to the regreening stage of various freezing resistance varieties of winter wheat were measured in the same planting conditions, the hyperspectral characteristics and the differences of winter wheat of various kinds under the temperature stress during the wintering stage were analyzed in detail, and the physiological response under the low temperature stress or in the adverse environment and the freezing injury hyperspectral response mechanism were preliminarily analyzed. Through the hyperspectral sensitivity analysis, the sensitive wavebands of freezing injury could be found, and the LWC Inversion Model could be built through the correlation among the hyperspectral response curve, the hyperspectral characteristic parameters and the LWC under the low-temperature stress. It can make better use of hyperspectral remote sensing technology for agricultural disaster prevention and precision agriculture and provide necessary information support for further study. The spectral sensitivity (SS) is defined as: the specific radio between the difference value of the spectra of the stressed plant and the spectra of the normal plant. The vegetation indexes of various combinations of blue, green, near-infrared and middle-infrared wavebands can be adopted to conduct the large-area freezing injury monitoring, and the severity of the injured degree can be determined through the size of the vegetation index difference. The correlation analysis was conducted through the integration of the LWC and spectral reflectance ratio and the first-order differential data of the winter wheat from the whole seedling stage to the regreening stage, to seek the LWC-sensitive wavebands characterizing the severity of freezing injury and establish the leaf water content monitoring model for the winter. The occurrence of freezing injury is a long and complex process, which is closely related to the climatic conditions, cultivation management conditions, wheat varieties and growing environment conditions. The experiment was carried out under the field conditions; natural conditions cannot be controlled. The cumulative precipitation from Oct. 1, 2010 to Mar. 30, 2011 was only 7 mm in Beijing, and although the irrigation had been carried out before winter, less precipitation during the whole wintering stage increases the freezing injury stress to the winter wheat. In the whole growth period of winter wheat, the wintering stage water requirement is minimal. If in this period there is a drought stress, usually the freezing injury will get worse, because the soil can't frozen, the wheat tillering node exposed in the low temperature will die, so the drought adds to the severity of the freezing injury. This article studied the inversion of winter wheat freezing severity at a single leaf scale in the wintering stage. It should be practical, but when winter wheat under the low temperature stress, chlorophyll, cell structure, etc., could be also influenced, so further study need to conduct on the simulation model extensionality. Furthermore, multiple affecting factors should be comprehensively considered and utilized in monitoring freezing injury in the future.
Comparison between the predicted value and measured value of LWC of winter wheat in four variables models.
28
4. EUROPE
a. Finland - Sodankylä, (Northern Finland)
Detection of Soil Freezing from L-Band Passive Microwave Observations
Kimmo Rautiainen , Juha Lemmetyinen, Mike Schwank, Anna Kontu, Cécile B. Ménard, Christian
Mätzler , Matthias Drusch, Andreas Wiesmann, Jaakko Ikonen, Jouni Pulliainen
Dual-polarized observations of L-band brightness temperature at a range of observation angles
were collected from a tower-based instrument, and evaluated against ancillary information on soil
and snow properties over four winter seasons. During the first three winter periods the
measurement site was located over mineral soil on a forest clearing, for the fourth winter the
instrument was moved to a wetland site. Both sites are located in Sodankylä, Northern Finland.
The test sites represent two environments typical for the northern boreal forest zone. The data
were applied to derive an empirical relation between the onset and progress of soil freezing and
the observed passive L-band signature.
A retrieval algorithm was developed using the observations at the forest opening site. The
algorithm exploits the perceived change in brightness temperature and the change in the relative
difference between the signatures at horizontal and vertical polarization. With the collected
experimental dataset, these features were linked optimally to the progress of soil freezing by
choice of observation angle, polarization and temporal averaging. The wetland site observations
provided the first opportunity for demonstrating the developed algorithm over a different soil
type, giving a first estimate of the algorithm performance over larger heterogeneous targets.
Areas affected by soil freezing cover more than half of the total landmass of the Earth. This
includes both areas with perennial and seasonal soil freezing; seasonal freezing occurs on
approximately 51% of the landmass, while permafrost areas cover 24% (Smith & Brown, 2009;
Zhang, Barry, Knowles, Ling, & Armstrong, 2003). Soil freezing affects both the latent heat
exchange and the surface radiation balance between the soil surface and the overlying medium.
Also, the evapotranspiration of different atmospheric trace gases such as CO2 and methane is
affected by soil freezing (Skogland, Lomeland, & Goksoyr, 1988; Zhang et al., 2003). This impacts
carbon balance estimates in environments affected by seasonal soil freezing (e.g. Frolking et al.,
1996).
COUNTRY CROP MATERIALS AND/
OR METHODS
PRODUCT DEVELOPMENT PLACE
Finland
Sodankylä,
(Northern
Finland)
2012
Boreal
Forest
Tower-based SMOS
reference
radiometer,
ELBARA-II, on loan
from ESA
Algorithm for
detecting seasonal
Soil Freezing
processes using L-
Band Microwave
Radiometry
Finnish Meteorological
Institute, Finnish
Environment Institute,
Finland
29
Furthermore, the thermal state of the soil controls the mass and rate of water infiltration and, by
extension, the partitioning between surface and sub-surface runoff. It is therefore essential to
have accurate information on the thermal state of the soil for hydrological and climatological
forecasts (e.g.Willis, Carlson, Alessi, & Hass, 1961). Microwave observations at L-band (1–2 GHz)
provide a useful tool for monitoring soil properties because of:
The relatively high emission depth at these wavelength (Ulaby, Moore, & Fung, 1982) and
Low scattering by surface vegetation (Jackson & Schmugge, 1989).
Experimental data on L-band microwave radiometry for the detection of soil freezing was
presented by Schwank et al. (2004) and Rautiainen et al. (2012). The objective of the present study
was to extend the findings reported by Rautiainen et al. (2012), covering several winter seasons
and a second observation site over a different soil type, allowing to establish a robust empirical
algorithm for soil freeze/thaw state detection from L-band microwave radiometry.
The detection of soil freezing from passive microwave observations is based on the difference in
emissivity between unfrozen and frozen soil. However, this difference is not discrete as ice and
liquid water and also bound water may coexist in soil; the suppression of the melting point of ice
increases with the increase of the specific surface area due to the Gibbs–Thomson effect
(Watanabe & Mizoguchi, 2002). The presence of free liquid water in unfrozen soils increases the
effective dielectric constant compared to situation when frozen, thus decreasing emissivity and
brightness temperature. On the other hand, freezing of the liquid soil water-phase affects the
detected emission in the same way as drying does, i.e. emissivity is increased.
Algorithm characteristic:
Was developed for passive L-band observations (ELBARA-II and SMOS)
Is based on increase in observed brightness temperature and decrease in polarization
difference due to soil freezing
Using the SMOS data from the first two winter periods (2010-2011 and 2011-2012) and
the data from Finnish Environment Institute’s Soil Frost Observation network
Soil state is defined from relative frost factor value
Wetland soil frost depth estimates based on frost detection algorithm.
30
5. OCEANIA
a. Australia – Clare Valley, South Australia
Vineyard Remote Sensing – Practical Applications. Early and pre harvest remote sensing are
being applied at Taylors as tools for improving vineyard productivity.
Colin Hinze, Taylors Wines
Taylors Wines was established in the Clare Valley, South Australia, in 1969, when the first
Cabernet Sauvignon vines were planted on the original 123 hectare property, near Auburn. In
2007, the vineyard operation has expanded to an Estate of over 500 hectares of established
plantings. Many of the original vineyards are in the process of being removed and replanted.
Since 2005, has been an introduced precision viticulture (PV) application into work practices, to
help improve factual knowledge of the vineyards. One of the tools found particularly helpful is
aerial remote sensing, in the form or digital multi-spectral imagery (DMSI).
DMSI is essentially a digital photograph of the blue, green, red and infra-red light reflected from
the vine canopy. This data can be expressed as either NDVI (normalized difference vegetation
index) or PCD (plant cell density). PCD data is delivered to us via the internet, and loaded into a
free software package called Viewpoint. This software allows us to view the imagery and analyze
one or more data layers (www.deltadatasystems.com).
The data from the PCD maps has helped to improve vineyard outputs.
Based on the maps which have begun to divide blocks into differential management zones, in an
attempt to reduce the inherent variability, or to exploit differences to maximize returns in quantity
and/or quality.
In recent seasons, some growers and service providers have been investigating the value of early
season remote sensing, targeted at flowering time (October to December, depending on the
region). As there is much greater time between capture and harvest, there are greater
opportunities to influence the outcome of that season, typically through water and nutrient
COUNTRY CROP MATERIALS AND/
OR METHODS
PRODUCT DEVELOPMENT PLACE
Australia
Clare
Valley,
South
Australia
2007
Vine Digital
Multispectral
Imagery (DMSI)
Images of Prevention
and Protection In the
future in Vineyards
Taylors Wines
New
Zealand,
Auckland
2007
Large
Horticultural
Fields
Wireless Sensor
Network (WSN)
Frost Alarm System Electrical & Electronics
Engineering, Auckland
University of Technology,
Auckland, New Zealand
31
management. While have not yet undertaken this at Taylors Wines, was obtained a spring 2006
PCD image for a different reason . . . frost.
Like many parts of Southern Australia, frost events during October 2006 had a significant impact
on vineyards. An opportunity arose to capture digital multi-spectral imagery of the property soon
after our worst frost, giving a very accurate record of the impact. will be used this information
during Vintage 2007 to carry out targeted harvesting of partially affected blocks, and continue to
refer to it as look at frost prevention and protection options in the future.
It is worth noting a word of caution at this point – that all remote sensing maps will look variable.
It is critical that the user of the data determines if the variability shown by the imagery is
significant enough to warrant further action.
Equally as important as ground truthing, is measuring the impact of changes made in
management. It is important that assess any change in yield and economic impact, for know if
decision was correct. To help achieve this, have installed grape yield monitors on both of machine
harvesters to enable to map grape production. These devices have been quickly adopted by
harvester operators, and are now integrated into the process of vintage.
Image showing the extent of frost damage in one area of the Taylors Wines Estate
November 2006
32
b. New Zealand - Auckland
Wireless Microclimate Sensor
David Grant, Adnan Al-Anbuky
The paper discusses the development of wireless microclimate sensor. The sensor forms the
generic component of a microclimate Wireless Sensor Network (WSN). The network is based on
IEEE 802.15.4 low power wireless protocol supported by ZigBee stack. The key objective of the
network is to support remote monitoring and management of large orchards. The designed sensor
together with a locally developed cell modem facilitates the main components for defining a
microclimate wireless sensor network that could be accessed remotely.
The developed sensor takes care of the critical environmental sensing for the horticultural
application needs.
It also takes care of energy management and self replenishment.
Preliminary results of the design sensor reflect good performance in coverage of large horticultural
fields at a reasonably low cost. Five Sunlight hours a day would keep the sensor functional through
its storage mechanism. The architecture allows for both power and location awareness.
Identification of the field areas that require attention could be made through either light emission
or embedded location awareness algorithm. Range of view for the frost alarm system would be in
the region of 500 meters. Initial testing reflects promising results.
In a horticultural setting it is often found that critical environmental characteristics may need
measuring. This is necessary for providing the necessary alarm when extreme environmental
conditions take place. These characteristics may include ambient light, Photo synthetically Active
Radiation (PAR) light, humidity, temperature and pressure of the surroundings. These
characteristics may represent a large physical area with equally large dynamic variations over the
area.
Wireless Sensor Networks give this opportunity by measuring the raw data at a node level and
sending it processed or unprocessed to an external link. The information gathered from sensors
covering a large field could then be integrated. This would result in identifying areas within the
larger field that require attention. Example to that is the possible frost that may damage the grape
within a vineyard.
ZigBee technology is a low data rate, low power consumption, low cost; wireless networking
protocol targeted towards automation and remote control applications. This makes it the
industrial standard for embedded monitoring or sensing applications.
ZigBee is a protocol standard that incorporates sensors, Micro Electro Mechanical Systems
(MEMS) with wireless communications. It is classed as a network that is self organizing, self
healing and is robust and reliable. This is suited to a scalable network of sensors that is to be setup
quickly and provide wide area coverage through mesh communications.
Comparison of the range of wireless protocols leads to a choice in ZigBee for its low power
requirements, network scalability reliability, and future protocol development.
The sensor node is designed such that a typical WSN Node has a microcontroller, sensing
elements, alarm indicator, through Light Emitting Diode (LED), radio communication through the
antenna and backup communication protocol & transceiver elements, and a power regulation and
supply system.
33
Using a dedicated Temperature and Humidity Sensor and a Photo synthetically Active Radiation
PAR corrected Light Sensor the node will be able to measure the needed environmental
characteristics. With energy being a major consideration, one of the key features of the system is
the effective energy harvesting from the solar panel. This takes place even if the battery does not
have the energy left to power the node. This circuit will allow for the charging of the battery
without needing the microcontroller. This ensures even in very low power situations the node will
be able to replenish the battery. Included in the key features would be the monitoring of energy
from the solar panel and the battery giving an indication of the nodes energy resource and the
likelihood of node failure.
For Frost Alarm the RGB LED is the flashing frost warning indicator. It will have differing light
output colour for differing temperature and humidity conditions. It will illuminate green for a frost
safe condition, blue for approaching frost danger condition, and red for frost danger condition.
This will ensure anyone watching the nodes or vineyard will have a clear understanding of how the
temperature is changing. Most likely use would be for the helicopter pilots flying above the
vineyard/orchard. The required intensity for night viewing gained through testing of the LED
would be ≥2000mCd (mill-candelas) resulting from tests. Distance of observation would be from
15 – 150m vertical height above ground level (the height range which frosts occur at). Also from
the tests the optimum flashing duty cycle was found, this is to conserve energy and obtain the
most contrasting attention when indicating frost conditions. The angle of observation would likely
be ≥45° above horizontal, thus a diffusing element would be used to spread out the light to
achieve this angle.
WSN Node Architecture
34
FINAL COMMENT
Technological advances seem to be moving very fast every year, connectivity and scientific study
has provided the solution, in many areas, several problems that are globally present. Each country
faces problems related to health, food, education, etc. different focuses its resources in trying to
respond and provide a useful alternative to face in the best way possible solution this situation.
Agriculture is an activity that affects very significantly in a country, therefore the losses that may
occur in this area generates a significant economic disruption, is why it is a topic that often
different agencies are working, and thus is no stranger to technological advances that exist today.
Frost over the years has led to significant economic losses in different crops, therefore it has
become necessary to treat and evaluate them from a preventive point of view, i.e., achieve to
obtain useful data generated early information that allow to take measures necessary to avoid
losing the plantations at risk.
This paper is focused on showing how in recent years the subject of worldwide frost covers,
revealing the various publications that have been written about it, pointing out the scientific
development of systems covering such events that affect the agriculture.
35
REFERENCES
CONCEPTS:
[1] «Las Heladas en Fruticultura». [Sergio Antonio Toledo Vivas].
[2] «Fórmula para Pronosticar Heladas». [Tomás Fernández Ábrica].
[3] «Protección contra las Heladas: fundamentos, práctica y economía - Volumen 1».
[Organización de las Naciones Unidas para la Agricultura y la Alimentación (FAO)].
[4] «Predicción del Riesgo de Heladas». [Foro del Tiempo].
EXAMPLES OF DIFFERENT METHODS USED WORLDWIDE:
[5] « Application of GIS Techniques in Frost Risk Mapping for Mitigating Agricultural Losses».
[Susan Malaso Kotikot & Prof. Simon Onywere.].
[6] «Caracterización espacial del Estrés Hídrico y de las heladas en la región pampeana a Partir
de Información Satelital y Complementaria». [Straschnoy, J. V.; Di Bella, C. M.; Jaimes, F.
R.; Oricchio, P. A.; Rebella, C. M.].
[7] «Low-Power Wireless Sensor Network for Frost Monitoring in Agriculture Research». [Ana
Laura Diedrichs, Germán Tabacchi, Guillermo Grünwaldt, Matías Pecchia, Gustavo
Mercado and Francisco González Antivilo].
[8] « Estación Meteorológica con Comunicación GSM para Predicción de Heladas y Detección
de Viento Zonda». [C. Godoy, C. Carletto, J.C. Correa and A. Lage].
[9] «Downscaling MODIS-derived maps using GIS and boosted regression trees (BRT): The case
of frost occurrence over the arid Andean highlands of Bolivia». [Robin Pouteau, Serge
Rambal, Jean-Pierre Ratte, Fabien Gogé, Richard Joffre, Thierry Winkel].
[10] «Frost Prediction Characteristics and Classification using Computational Neural Networks».
[Philip Sallis, Mary Jarur, and Marcelo Trujillo].
[11] « Calibración y Validación de Modelos de Pronóstico de Heladas En El Valle Del Mantaro».
[Janeet Sanabria, Jerónimo García, Jean-Paul Lhomme].
[12] «La Teledetección y los Sistemas de Información Gergráfica, algunas aplicaciones en el
Uruguay». [Instituto de Investigaciones Agropecuarias (INIA)].
[13] « Hyperspectral characteristics of winter wheat under freezing injury stress and LWC
inversion model ». [Huifang Wang, Jihua Wang, Qian Wang, Naizhe Miao, Wenjiang Huang,
Haikuang Feng, Yingying Dong].
[14] « Detection of Soil Freezing from L-Band Passive Microwave Observations ». [Kimmo
Rautiainen , Juha Lemmetyinen, Mike Schwank, Anna Kontu, Cécile B. Ménard, Christian
Mätzler , Matthias Drusch, Andreas Wiesmann, Jaakko Ikonen, Jouni Pulliainen].
[15] « Vineyard Remote Sensing – Practical Applications. Early and pre harvest remote sensing
are being applied at Taylors as tools for improving vineyard productivity». [Colin Hinze,
Taylors Wines].
[16] «Wireless Microclimate Sensor». [David Grant, Adnan Al-Anbuky].