Research Advances in Monitoring Agro-meteorological Disasters Using Remote Sensing

Xueyan SUI，Rujuan WANG，Huim in YAO，Meng WANG，Shaokun LI，Xiaodong ZHANG*

1.Shandong Academy of Agricultural Sciences，Jinan 250100，China；2.Center of Agricultural Technique Extension of Longkou City，Longkou 265700，China；3.Institute of Crop Sciences，Chinese Academy of Agricultural Sciences／The National Key Facilities for Crop Genetic Resources and Improvement，Beijing 100081，China

1 Introduction

Agricultural production is closely related to meteorological condition.Excellent meteorological condition and favorable weather can promote crop growth and development，while frost，aridity，hail，rainstorm，and dry-hot wind will seriously influence production.Meteorological disasters have brought huge harm to China's agriculture，rural areas and farmers.In 2010，the area covered by floods and drought was up to 37.43million hectare，and the area affected by disasters reached 18.54million hectare，accounting for 11.54%of the total sown area［1］.Based on uncertainty of grain production due to unfavorable factors such as agricultural disasters，number1 document of central government in 2010-2012 energetically supports agriculture science and technology，to guarantee stable growth of China's grain yield in disasters.Monitoring and evaluation of meteorological disasters are important sections in disaster relief and reduction.Disaster evaluation results can provide objective reference for decision-making department formulating pertinent disaster reduction and relief schemes，and provide support for stable increase of grain yield，which is favorable for sustainable development of China's grain production.

In the past，disaster monitoring and evaluation methods are mainly field survey and sampling.These methods not only waste time，butalso waste labor.They are drawbacks of poor representativeness，strong subjectivity and poor timeliness，thus it is difficult to satisfy real-time monitoring demands of large area of disasters.What's worse，farmland and transportation would be seriously damaged after disasters，which will greatly influence formulation of disaster relief schemes.With the development of science and technology，the remote sensing provides a scientific and effective method for monitoring and evaluation of agricultural disasters，provides important means for large area and rapid obtaining crop and environment information，and is an essential means for future agricultural disaster monitoring and evaluation.

2 Remote sensing principle of agricultural disaster monitoring

The remote sensing technology is based on electromagnetic wave theory.Once disaster occurs，biological and chemical components and canopy structure will have change，electromagnetic wave received by sensor will change accordingly.Through comparing and analysis of electromagnetic wave information before and after disasters，it is able to obtain crop disaster situations.At present，remote sensing monitoring has been launched for floods，droughts，snow storms，sandstorms，plant diseases and insect pests，and typhoons.Different disasters exert different influence on physiological and biochemical component of crops.In the event of floods，too much water will lead to liquid phase replacing gas phase，consequently plants grow in the anaerobic environment.In this situation，plants will become small and short，leaves become yellow，leafstalk grow slightly at the top，root system becomes shallow and thin，and root hair will significantly decrease.The accumulated water will stop growth of root system，which will gradually blacken and rot and the entire plant will wither.In the event of droughts，soil will lack water，crops will wilt.In the event of frost，plant tissue will freeze，leading to injury or even death of plants［2］.Suffered from different disasters，crops will take on different features，which will show respective characteristics in spectrum.Thus，the monitoring method of remote sensing is varied.

2.1 Monitoring floods by remote sensingAfter rainstorm，water level of river systems will rise and water body coverage will increase.Main points of present flood monitoring remain in increase of water body coverage and existence time of water body.In 1998，devastating floods hit the whole country.The Chinese Academy of Science sused the remote sensing data of time series to rapidly recognize flood and the dynamic information.It completed more than 70 remote sensing monitoring images and pictures，and more than 50 disaster analysis reports and bulletins，and rapidly transmitted to the State Council and related ministries and commissions，which greatly supports the struggle in fighting flood and relieving victims［3］.The submerged area is generally classified for images by multi band satellite data.Apart from common artificial recognition［4］，computer image classification method（such as various supervised classification and non-supervised classification methods）；now there are some new easy and flexible methods，such as relationship between remote sensing wave band and spectrum［5］and water body discriminant function method［6］.However，there are no pertinent researches about remote sensing monitoring of crop floods，such as the relationship between different crops，levels，spectrum after floods and corresponding cultivation indicators.

2.2 Monitoring droughts by remote sensingThere are more than 150 kinds of definitions of drought［7］，mainly including meteorological drought，hydrological drought，agricultural drought，and economic drought.Agricultural drought occurs when there isn't enough moisture to support average crop production on farms or average grass production on range land.Although agricultural drought often occurs during dry，hot periods of low precipitation，it can also occur during periods of average precipitation when soil conditions or agricultural techniques require extra water.Major factors reflecting drought degrees include precipitation，water content in soil，soil texture，temperature，crop variety and yield，as well as season when the drought occurs.The two key factors are soil water content and crop water demand.Therefore，the direct objective of using remote sensing to monitor drought is to monitor the soil water content.The soil water content is highly correlated with the soil thermal inertia，a quantity that characterizes a material's resistance to changes in temperature.It is able to calculate soil water content through calculating changes of soil temperature difference between day and night.For exposed soil，this method has high accuracy.Drought threat not only remains in seed germination period，but also runs through the whole breeding period.Thus，how to obtain water content in the soil covered by crops is the key for successful monitoring.

When winter wheat suffers winter or spring droughts，introducing Enhanced Vegetation Index（EVI）into the thermal inertia model can increase the inversion accuracy of soil water content［8-10］.Leaf is one of the organs vulnerable to damage in adverse environment and is easily observed.Through vegetation index and multi-year anomaly data obtained through remote sensing or changes of vegetation index series in the event of droughts，it is able to directly monitor crop drought situation［11-13］.Although remote sensing monitoring of droughts has been studied for a long time，nearly all crops are faced with influence of drought disasters.There are a lot of works to do to obtain accurate soil water content and establish drought monitoring and evaluation methods through remote sensing and other data support.

2.3 Monitoring snowstorm s by remote sensingSnowstorms have been monitored since 1966.Many kinds of satellite sensors are used to obtain information of dynamic changes of snow cover，such as NOAA／AVHRR，NASA／MODIS，Landsat／MSS，TMand environment monitoring satellite sensors［14］.At present，binary snow cover mapping methods widely used include brightness threshold method，normalized difference snow index（NDSI）based on high reflectance of visible light wave band and low reflectance of near-infrared band of snow disaster，and image supervised classification method based on multi spectral image operation［15-17］.With the aid of these methods，many international organizations and institutions have developed a lot of snow cover monitoring system.

Combining snowstorms in pastoral areas of China，some domestic scholar shave proposed many snow disaster evaluation methods.They took dominant factor of snow disaster＂constant snow cover days＂as criteria for evaluation snow disaster scale，and divided snow disasters into light snow disaster，mediate snow disaster，heavy snow disaster，and extremely heavy disaster with 5，7，10 and more than 30 days of snow cover days［18］.Snow disasters in China mainly occur in November，March and winter（December to next February）.According to average precipitation in winter，or November and March in pastoral areas subject to snow disasters（namely，perennial ground snow cover）and the precipitation anomaly percentage（namely，current year actual ground snow cover depth），they divided the snow disasters into general and serious disasters［19］.Snow disasters generally occur in high altitude areas.Crops subject to snow disasters mainly include forage grass and protected vegetables.Monitoring indicators are few and it is easy to form unified remote sensing monitoring technologies and standards and popularize and apply these technologies and standards.

2.4 Monitoring sandstorms by remote sensingSince there is certain difference in albedo and temperature of top，ground surface and cloud layer of areas subject to sandstorms，and the polar orbiting meteorological satellite and EOS series satellites have channels for detecting albedo and temperature，it is feasible to use satellite remote sensing to monitor sand storms.Ma Yunjunet al.［20］divided sandstorm risks into potential risks and existing risks and made regional evaluation of potential risks of sandstorms in central and northern regions of Ningxia from influencing sandstorm risks and cultural factors with overall consideration of danger of disaster-inducing factors，stability of disaster-formative environment and the vulnerability of hazard.Li Jinrong［21］，based on the-ory of regional disasters，by RSand GIS technical means，through analysis on sandstorm features，established evaluation indicators of sandstorms and builts and storm evaluation model by hierarchical analysis method and fuzzy comprehensive evaluation method.In China，sandstorms often occur in spring.They can expose root system of crops and forage grass，blow seeds and seedlings.Thick sands covering plant leaves will influence normal photosynthesis and lead to yield reduction of farming and animal husbandry.In the existing sandstorm remote sensing monitoring，there are relatively few researches about the influence of sandstorm intensity on crops.

2.5 Monitoring plant diseases and insect pests by remote sensingCompared with meteorological disasters，researches on plant diseases and insect pests are more pertinent and have detailed to specific crop varieties and types of plant diseases and insect pests.Once there are plant diseases and insect pests，crops will have changes of absorption and reflection features in different wave bands in varying degrees，namely，spectrum response of plant diseases and insect pests.These features will become basic references for monitoring plant diseases and insect pests by remote sensing.Scholars both at home and abroad have made extensive experimental researches on remote sensing monitoring of plant diseases and pest insects of major crops including wheat，rice and cotton，analyzed physiological mechanism of spectrum response of plant diseases and pest insects［22］，position of spectrum response features of plant diseases and pest insects［23-24］，and established vegetation index for monitoring of plant diseases and pest insects and algorithms for remote sensing recognition and degree distinguishing of plant diseases and pest insects［25-26］.Once plant diseases and pest insects occur，the reflectance of visible light increases，the reflectance of near infrared light decreases，the red side moves towards short wave direction，but different crops have different plant diseases and pest insects，so the changes of pigment，water and protein in crops are different，and the influence on overall structure of crop plant is also different.To make discrimination，the key is to increase resolution of spectrum.

2.6 Monitoring typhoons by remote sensing Typhoons can destroy crops，blow off leaves，break tree branches and trunks，and even uproot trees，so they greatly influence plant growth.Traditional methods mainly include field survey［27］and aerial photos combined with field survey［28］.With the development of satellite remote sensing，the satellite image was gradually applied in the evaluation of influence of typhoons in the1990s.Combining SPOT images and surface DTMdata，Leeet al.［29］concluded that typhoons lead to different NDVI changes of different types of terrain and species.At windward slopes due to influence of stronger wind，NDVIof trees have the highest drop.As for NDVI drop of different species，natural broad-leaf forest is greater than artificial coniferous forest.Wanget al.［30］，with the aid of Moderate Resolution Imaging Spectroradiometer（MODIS），calculated changes of NDVI，EVI，NDII，LAI and Fraction of Absorbed photosynthetically Active Radiation（FPAR）after typhoons and analyzed influence on these changes on forest，and found that NDVI changes best reflect the damage.

Crop lodging is a general problem resulted from typhoons.Shandong Institute of Agricultural Sustainable Development launched remote sensing monitoring and evaluation researches on wheat lodging since 2007.Through satellite synchronized positioning test in 2007 and 2013 and artificial simulation experiment of wheat lodging in 2010，it analyzed change characteristics of spectral curve before and after wheat lodging and preliminarily established extraction method for wheat lodging disaster at grain-filling stage［31-32］.Liu Liangyunet al.［33］successfully monitored degree of wheat lodging using changes of NDVI values with the aid of 2 Landsat ETM satellite images of wheat lodging on April 7，2003 and May 9，2003.

Typhoon is a violent tropical cyclone.Wind directions change from time to time and are unexpected.Disasters resulted from typhoonshave poor homogenieity，thus it isgreatly difficult to set up samples for remote sensing researches.

3 Development demands and major problems to be solved

3.1 Rapid response after disaster is the demand for agricultural industrialization development Scientific and technological development promotes increase of productivity but also brings about surplus of agricultural production labor.To seek higher living quality，rural residents graduallymove to urban areas and fewer and fewer farmers are willing to do farming work.Land circulation and agricultural industrialization have become inevitable result in the special historical period.Fewer farmers doing farming work will have higher dependence on science and technology.To increase agricultural production efficiency，3S technology with remote sensing asmajor part becomes important tool for agricultural production management.Especially after disasters，since the time is urgent and resources are in short，rapid monitoring and evaluation have great significance.

3.2 Carrying researches on remote sensing monitoring of agricultural meteorological disasters is import ant basis for agricultural insurance claims China's agricultural production is weak and seriously influenced from disasters and has high operating risk，so it objectively needs agricultural insurance to transfer，diversify risks and share economic losses.The＂three rural issues＂concerning rural areas，agriculture and farmers are always hot spots of social reform and economic development，and agricultural insurance is receiving closer and closer attention of various circles of society.Shandong Province launched pilot projects of agricultural insurance in 1982，but the business scale is still not large.Reasons for this would be（i）small scale of farmers'household operation and low awareness of participating in agricultural insurance，and（ii）lack of reference of agricultural insurance claims and certain difficulty in disaster survey.In recent years，agricultural industrialization organizations are rich and varied and probability of farmers participating in industrialized operation becomes higher.Therefore，establishing agricultural meteorological disaster monitoring and evaluation system in short time will play certain role in promoting smooth implementation of agricultural insurance.

3.3 Problems faced by remote sensing monitoring of agricultural meteorological disastersResearch of agricultural remote sensing was started in the beginning of the 1980s and firstly applied in agricultural production management in Xinjiang and Heilongjiang，but the remote sensing monitoring of agricultural meteorological disasters still remains at the stage of scientific researches.Major reasons for this are as follows.

Firstly，agricultural production is situated in open environment and subject to influence of many factors such as seed quality，cultivation management measures，and environment，etc.Different crops present different characteristics in disasters；even the same crop shows different characteristics in disaster in different growth period.From the perspective of principle of remote sensing monitoring，the problem of same species having different spectrum and different species having the same spectrum is a general phenomenon.

Secondly，past researches of cultivation of various crops lack systematic researches about loss due to meteorological disasters.Therefore，when making disaster evaluation using remote sensing technology，it is necessary to establish a complete set of disaster monitoring and evaluation system from the perspective of cultivation physiology，then transit from spectrum data to cultivation indicators and from qualitative analysis to quantitative evaluation，so as to establish remote sensing monitoring and evaluation system for agricultural meteorological disasters.

Thirdly，agricultural remote sensing is generally based on spectrum analysis.Onceagricultural disaster occurs，it is impossible to obtain basic information in bad weather（for example，in cloudy condition）.In addition，there is a certain period for obtaining satellite data.It is difficult to make remote sensing monitoring.For instance，in Wenchuan earthquake，due to bad weather，it was impossible to use remote sensing technology in the beginning.Therefore，it is necessary to carry out researches and application of unmanned aerial vehicle remote sensing and radar remote sensing monitoring technologies.

Finally，the occurrence and development of agricultural meteorological disasters are mainly influenced from meteorological factors and present on sudden characteristics.Thus，positioned observation only bring saccidental data of certain condition.To obtain systematic data，it needs a long time.

［1］China Statistical Yearbook（2011）［M］.Beijing：China Statistics Press，2011：492.（in Chinese）.

［2］JIANG GM.Plant ecophysiology［M］.Beijing：Higher Education Press，2004：161-212.（in Chinese）.

［3］WEICJ，WANG SX，YAN SY，et al.The main accomplishemnt of flood hazard monitoring and evaluating for China by remote sensing in 1998——the application of flood hazard'squick-reporting system remote sensing based on network［J］.Journal of Natural Disasters，2000，9（2）：16-25.（in Chinese）.

［4］Cihlar J，Xiao QH，Chen J，et al.Classification by progressive generalization：A new automated methodology for remote sensing multichannel data［J］.Int.J.Remote Sensing，1998，19（14）：2685-2704.

［5］ZHOU CH，DU YY，LUO JC.A description model based on knowledge for automatically recognizing water from NOAA／AVHRR［M］.Beijing：Surveying and Mapping Press，1996：221-232.（in Chinese）.

［6］PEIZY，YANG BJ.Study on macroscopical flood hazard monitoring by remote sensing using NOAA image［J］.Transactions of the Chinese Society of Agricultural Engineering，1999，15（4）：203-206.（in Chinese）.

［7］Wilhite DA.Preparing for drought：a methodology［A］.Wilhite，D A.（Ed.），Drought：A Global Assessment，Hazards and Disaster Series，vol.II［M］.Routledge，New York，2002：89-104.

［8］YANG LP，SUIXY，YANG J，et al.Construction of remote sensing monitoring model for spring soil moisture in Shandong Province［J］.Shandong Agricultural Sciences，2009（5）：17-20.（in Chinese）.

［9］YANG YY，GUO HH，SUIXY，et al.Construction of drought monitoring system by remote sensing in major wheat areas of Shandong Province［J］.Shandong Agricultural Sciences，2009（1）：15-18.（in Chinese）.

［10］YANG YY.Research on business model running of wheat drought remote monitoring beforemound closure in Shandong Province［D］.Ji'nan：Shandong Normal University，2009：38-69.（in Chinese）.

［11］FENG Q，TIAN GL，LIU QH.Research on the operational system of drought monitoring by remote sensing in China［J］.Journal of Remote Sensing，2003，7（1）：14-18，T001.（in Chinese）.

［12］FENG Q，TIAN HL，WANG AS，et al.Remote sensing monitoring of soil humidity using vegetation condition index［J］.Journal of Natural Disasters，2004，13（3）：81-88.（in Chinese）.

［13］JUWM，SUN H，TANG ZC.The application of meteorological satellite remote sensing in drought monitoring［J］.Journal of Catastrophology，1996，11（4）：25-29.（in Chinese）.

［14］Rango A，Martinec J.Snow accumulation derived from modified depletion curves of snow coverage［J］.Symposium on Hydrological Aspects of Alpine and High Mountain Areas，IAHSPublication No.1982，138：83-90.

［15］Hall DK，Riggs GA，Salomonson VV.Development of methods for mapping global snow cover usingmoderate resolution imaging spectroradiometer data［J］.Remote Sensing of Environment，1995，54（2）：127-140.

［16］CAOMS，LI X，CHEN XZ，et al.Remote sensing of cryosphere［M］.Beijing：Science Press，2006：10-45.（in Chinese）.

［17］JIQ，SUN LX，WANG Y.Snow monitoring using MODIS data［J］.Remote Sensing Information，2006（3）：57-58.（in Chinese）.

［18］ZHOU LS，WANG QC.Study on real-time predictive assessment of snowstorm disaster in eastern pastoral area of Qinghai-Tibet Plateau［J］.Journal of Natural Disasters，2001，10（2）：58-65.（in Chinese）.

［19］GAOQH，MA ZJ，ZHANGYC.Natural disaster assessment［M］.Beijing：China Meteorological Press，2007：80-120.（in Chinese）.

［20］MA YJ，WANG JA，LIXY.Regional comparative assessment of potential risk ofwind sand disaster in north andmiddle partsof Ningxia［J］.Journal of Natural Disasters，2007，16（5）：1-8.（in Chinese）.

［21］LIJR.Study on risk assessment of sand storm disaster based on RS and GIS［D］.Beijing：Beijing Forestry University，2011：28-139.（in Chinese）.

［22］WANG JH，ZHAO CJ，HUANGWJ，et al.Quantitative remote sensing basis and application［M］.Beijing：Science Press，2008：373.（in Chinese）.

［23］YANGCM，CHENG CH，CHEN RK.Changes in spectral characteristics of rice canopy infested with brown plant hopper and leaf folder［J］.Crop Science，2007，47（1）：329-335.

［24］LIU ZY，WU HF，HUANG JF.Application of neural networks to discriminate fungal infection levels in rice panicles using hyper spectral reflectance and principal components analysis［J］.Computers and Electronics in Agriculture，2010，72（2）：99-106.

［25］Jones CD，Jones JB，LeeW S.Diagnosisofbacterial spotof tomato using spectral signatures［J］.Computers and Electronics in Agriculture，2010，74（2）：329-335.

［26］LIU ZY，WANG DC，LIB.Discrimination of lodged rice based on visible／near infrared spectroscopy［J］.Journal of Infrared and Millimeter Waves，2009，28（5）：342-345.（in Chinese）.

［27］Miura M，Manabe T，Nishimura N，et al.Forest canopy and community dynamics in a temperate old-growth evergreen broad-leaved forest，southwestern Japan：a7-year study of a4-ha plot［J］.Journalof Ecology，2001，89（5）：841-849.

［28］KuPfer JA，Myers AT，Mclane SE，et al.Patterns of forest damage in a southern Mississippi landscape caused by hurricane Katrina［J］.Ecosystems，2008，11（1）：45-60.

［29］Lee MF，Lin TC，Vadeboneoeur MA，et al.Remote sensing assessment of forest damage in relation to the1996 strong typhoon herb at Lienhuachiexperimental forest，Taiwan［J］.Forest Ecology and Management，2008，255（8）：3297-3306.

［30］WANGWT，QU JJ，HAOXJ，et al.Post-hurricane forest damageassessment using satellite remote sensing［J］.Agricultural and Forest Meteorology，2010，150（1）：122-132.

［31］ZHANG J.A study on the lodge of winter wheat by remote sensing monitor［D］.Ji'nan：Shandong Normal University，2011：8-48.（in Chinese）.

［32］HU ZJ，ZHANG J，WANG ZH.Spectral variation characteristics of wheat lodging in the filling period［J］.Journal of Anhui Agricultural Sciences，2011，39（6）：3190-3192.（in Chinese）.

［33］LIU LY，WANG JH，SONG XY，et al.The canopy spectral features and remote sensing of wheat lodging［J］.Journal of Remote Sensing，2005，9（3）：323-327.（in Chinese）.