[ieee 2014 third international conference on agro-geoinformatics - beijing, china...

6
Assessment of Agricultural Drought Indicators Impact on Soybean Crop Yield: A Case Study in Iowa, USA Youxin Huang, Xiuguo Liu, Yonglin Shen China University of Geosciences (Wuhan) Faculty of Information Engineering Wuhan, Hubei China [email protected] JiangFeng Jin Institute of Surveying Mapping and Geoinformation of Henan Zhengzhou Zhengzhou, Henan China [email protected] AbstractAgricultural drought is a condition of insufficient soil moisture caused by a deficit in precipitation over some time period. Soil moisture drops to a certain extent, adverse to the crop yield, and then reduces the production of crops. Soybean is one of the most important sources of oil and protein in the world. It is vulnerable to recurrent drought condition in the U.S. state of Iowa. This study was conducted to identify agricultural drought indicator that strongly correlated with soybean crop yield. Detail crop data (e.g., soybean crop yield, etc.) were collected from the USDA’s National Agricultural Statistics Service (NASS). A region of interests is defined based on the MODIS 16-days 250m resolution vegetation index synthetic products (MOD13Q1) and daily land surface Temperature/Emissivity 1km resolution products (MOD11A1) from 2000 to 2013 in Iowa, which were used to compare three kinds of remote sensing derived agricultural drought monitoring indicator of crop water demand status: (i) Crop morphological indices (e.g., NDVI/VCI); (ii) Crop physiological indices (canopy temperature, e.g., TCI); and (iii) Crop comprehensive indices (e.g., VSWI). Drought cumulative effects were considered according to the specific soybean crop growth stages including from planted to emerged, vegetative period (from emerged to blooming), reproductive period (from blooming to setting pods), and growing season (from emerged to dropping leaves). The impacts of drought duration on the soybean crop yield by both of indices were analyzed. These results imply that physiological indices and comprehensive index were more correlated to assess the effect of drought on soybean yield. Drought indices accumulated in reproductive period (from blooming to setting pods) are highly superior to other accumulated for impacting on soybean yield, while over the growth season (from emerged to dropping leaves) is also highly correlated with the total yield assessment. The results can aid on evaluating the effects of drought on soybean yield in different growth stage. This could be very useful to providing auxiliary decision-making information for drought relief, agricultural manager and grain merchant to plan solutions and prepare for potential drought in advance. Keywords—agricultural drought indicator; soybean crop yield; remote sensing; drought cumulative effects I. INTRODUCTION Drought is a recurring phenomenon with varying frequency, intensity, duration, and spatial extent affects most land areas of the world [1]. It is also one of the most costly related natural disasters in the world, causing the average annual losses attributable to drought at US$6-8 billion from the assessment of the United States Federal Emergency Management Agency (FEMA, 1995). Most regions of US have experienced drought; however, agricultural production and crop yield are more susceptible to drought. Agricultural drought is a condition of insufficient soil moisture caused by a deficit in precipitation over some time period. Soil moisture drops to a certain extent, adverse to the crop yield, and then reduces the production of crops. The agricultural production costs and losses associated with drought are increasing dramatically, although it is difficult to quantify this trend precisely in terms of spatial extent and intensity [7]. The success of drought preparedness and mitigation, to some degree, depends on timely information on drought onset, progress and extent through drought monitoring. The drought monitoring usually be performed by using the drought indicators, which can be used to estimate the drought severity and track the progress of the drought. With the development of remote sensing (RS) technology, the RS technology has been more widely adapted to monitoring agricultural drought due to the advantages of the macroscopic, cheap, wide coverage and high space-time resolution. A recent review of agricultural drought indices derived from remote sensing illustrates its research progress and classification, including soil moisture indices and crop water demand indices. RS microwave and thermal inertia is considered as potential method of monitoring soil moisture. Monitoring Crop water demand conditions derived from RS is generally reflected through crop morphological indices, crop physiological indices and comprehensive index. Soybean is one of the most important sources of oil and protein in the world. However, it is vulnerable to recurrent drought condition in the U.S. state of Iowa. The goal of the study was conducted to identify agricultural drought indicator that correlated closely with soybean crop yield. This study utilized detailed crop data and weather remote sensing data collected from 2000 to 2013 from a series of soybean trials analyzed across the U.S. state of Iowa to evaluate the effects of growing season drought on the soybean yield. This work was supported in part by the National Development and Reform Commission satellite and application of industrial development special project (Development and reform office high-tech [2012] 2083).

Upload: jiangfeng

Post on 01-Apr-2017

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: [IEEE 2014 Third International Conference on Agro-Geoinformatics - Beijing, China (2014.8.11-2014.8.14)] 2014 The Third International Conference on Agro-Geoinformatics - Assessment

Assessment of Agricultural Drought Indicators Impact on Soybean Crop Yield: A Case Study in

Iowa, USA

Youxin Huang, Xiuguo Liu, Yonglin Shen China University of Geosciences (Wuhan)

Faculty of Information Engineering Wuhan, Hubei China

[email protected]

JiangFeng Jin Institute of Surveying Mapping and Geoinformation of

Henan Zhengzhou Zhengzhou, Henan China

[email protected]

Abstract—Agricultural drought is a condition of insufficient

soil moisture caused by a deficit in precipitation over some time period. Soil moisture drops to a certain extent, adverse to the crop yield, and then reduces the production of crops. Soybean is one of the most important sources of oil and protein in the world. It is vulnerable to recurrent drought condition in the U.S. state of Iowa. This study was conducted to identify agricultural drought indicator that strongly correlated with soybean crop yield. Detail crop data (e.g., soybean crop yield, etc.) were collected from the USDA’s National Agricultural Statistics Service (NASS). A region of interests is defined based on the MODIS 16-days 250m resolution vegetation index synthetic products (MOD13Q1) and daily land surface Temperature/Emissivity 1km resolution products (MOD11A1) from 2000 to 2013 in Iowa, which were used to compare three kinds of remote sensing derived agricultural drought monitoring indicator of crop water demand status: (i) Crop morphological indices (e.g., NDVI/VCI); (ii) Crop physiological indices (canopy temperature, e.g., TCI); and (iii) Crop comprehensive indices (e.g., VSWI). Drought cumulative effects were considered according to the specific soybean crop growth stages including from planted to emerged, vegetative period (from emerged to blooming), reproductive period (from blooming to setting pods), and growing season (from emerged to dropping leaves). The impacts of drought duration on the soybean crop yield by both of indices were analyzed. These results imply that physiological indices and comprehensive index were more correlated to assess the effect of drought on soybean yield. Drought indices accumulated in reproductive period (from blooming to setting pods) are highly superior to other accumulated for impacting on soybean yield, while over the growth season (from emerged to dropping leaves) is also highly correlated with the total yield assessment. The results can aid on evaluating the effects of drought on soybean yield in different growth stage. This could be very useful to providing auxiliary decision-making information for drought relief, agricultural manager and grain merchant to plan solutions and prepare for potential drought in advance.

Keywords—agricultural drought indicator; soybean crop yield; remote sensing; drought cumulative effects

I. INTRODUCTION Drought is a recurring phenomenon with varying frequency,

intensity, duration, and spatial extent affects most land areas of

the world [1]. It is also one of the most costly related natural disasters in the world, causing the average annual losses attributable to drought at US$6-8 billion from the assessment of the United States Federal Emergency Management Agency (FEMA, 1995). Most regions of US have experienced drought; however, agricultural production and crop yield are more susceptible to drought. Agricultural drought is a condition of insufficient soil moisture caused by a deficit in precipitation over some time period. Soil moisture drops to a certain extent, adverse to the crop yield, and then reduces the production of crops. The agricultural production costs and losses associated with drought are increasing dramatically, although it is difficult to quantify this trend precisely in terms of spatial extent and intensity [7]. The success of drought preparedness and mitigation, to some degree, depends on timely information on drought onset, progress and extent through drought monitoring. The drought monitoring usually be performed by using the drought indicators, which can be used to estimate the drought severity and track the progress of the drought.

With the development of remote sensing (RS) technology, the RS technology has been more widely adapted to monitoring agricultural drought due to the advantages of the macroscopic, cheap, wide coverage and high space-time resolution. A recent review of agricultural drought indices derived from remote sensing illustrates its research progress and classification, including soil moisture indices and crop water demand indices. RS microwave and thermal inertia is considered as potential method of monitoring soil moisture. Monitoring Crop water demand conditions derived from RS is generally reflected through crop morphological indices, crop physiological indices and comprehensive index.

Soybean is one of the most important sources of oil and protein in the world. However, it is vulnerable to recurrent drought condition in the U.S. state of Iowa. The goal of the study was conducted to identify agricultural drought indicator that correlated closely with soybean crop yield. This study utilized detailed crop data and weather remote sensing data collected from 2000 to 2013 from a series of soybean trials analyzed across the U.S. state of Iowa to evaluate the effects of growing season drought on the soybean yield.

This work was supported in part by the National Development and Reform Commission satellite and application of industrial development special project (Development and reform office high-tech [2012] 2083).

Page 2: [IEEE 2014 Third International Conference on Agro-Geoinformatics - Beijing, China (2014.8.11-2014.8.14)] 2014 The Third International Conference on Agro-Geoinformatics - Assessment

II. STUDY AREA

A. Study Area and Dataset Soybean is one of the major crops grown on the United

States. A region of interests is picked in the states of Iowa (see Fig. 1) since it is in the top 10 list of the second big agricultural areas, and Iowa contributed to about 12% of soybean productions yearly in the United States, which the soybean yield is higher than other countries. Moreover, it almost have experienced drought every year.

Fig. 1. Iowa contributed to about 12% of soybean productions in U.S.(data source: crop production report was released by NASS’s USDA in 2013.)

Three main data sets were collected in recent decade from 2000 to 2013 of soybean growing seasons.

1) To obtain the agricultural drought data source of the calculation of crop morphological indices and physiological indices, two kinds of MODIS products are collected from 2000 to 2013 in Iowa, including MOD13Q1 and MOD11A1. Global MOD13Q1 data are provided every 16 days at 250-meter spatial resolution as a gridded level-3 product in the Sinusoidal projection, which has MODIS NDVI and Enhanced Vegetation Index (EVI) products that have been computed from atmospherically corrected bidirectional surface reflectance. MOD13Q1 may be used as input for drought indices. The daily land surface temperature (LST) data is collected through MOD11A1, which is used to calculate canopy temperature indicator. The band 31 and band32 is especially suitable for agricultural drought monitoring the surface temperature derived from remote sensing inversion, although there are eight MODIS thermal infrared band is designed for monitoring the surface thermal changes, so MOD11A1 products are chosen as the study of crop canopy temperature indices.

2) Land Use Data. The land use vector data and annual land acreage of the soybean in Iowa are collected from the Crop Scape – Cropland Data Layer (CDL) Home that is available online at http://nassgeodata.gmu.edu/CropScape/. The CDL categorized geospatial product is supported by Department of Agriculture’s NASS at 30m or 51m resolution from satellite observation data and ground truth data. The CDL data is used as a mask data to separate the soybean layer from other crop layer, on account of the same land in different years, and the cultivation of crops and the scope have their different patterns.

3)Detail Crop Statistics Data. Soybean is one of main farm crops in Iowa and the local farmers’ economy depends on

this production. Soybean phonological stages are described as follows: planted, emerged, blooming, setting pods, dropping leaves, and harvested. Palle Pedersen, who is a soybean extension agronomist, provides detailed information on the soybean growth and development of two key phases: vegetative stages (from emergence to blooming, including VE, V1, V2, V3,…, Vn) and reproductive stages (from start blooming to dropping leaves, including R1, R2, R3,…, R8). The date of soybeans progress stages, in general, is from 18 weeks to 42 weeks for the whole soybean growing season. The statistical information of the soybean growth process is captured in weeks from USDA's NASS Quick Stats 2.0 website [5]. Quick Stats 2.0 website is the most comprehensive tool for accessing agricultural data published by USDA's NASS and available online at http://www.nass.usda.gov/Quick_Stats/. The historical soybeans yield data is also collected in bushel/acre from this website.

III. DATA PROCESS Two kinds of MODIS data (MOD13Q1 and MOD11A1)

need to go through the following pretreatment of five steps: mosaic, projection, clipping, resampling, and mask. And soybean growth process data needs to be split in order to consider the drought cumulative effect.

A. Procedure • The Sinusoidal Projection (SIN) is used for most of the

gridded MODIS land products, but this way of projection and CDL images projection is different because of CDL’s projection is “USA Contiguous Albers Equal Area Conic USGS”, so we usually converted into accordant projective coordinates before application of MODIS product.

• The U.S. state of Iowa covers the three grids of h10v04, h11v04 and h12v04, and there is a long time series MODIS dataset from 2000 to 2013. Therefore, the batch program is designed using the ENVI IDL tools to mosaic the time series dataset.

• The two kinds of MODIS data are batch clipped respectively by the state of Iowa vector data to obtain the corresponding state image.

• when analyzing soybean crop yield, other crop layer should be excluded from our study interesting region. So the soybean CDL as a mask data is used to extract soybean pixels from the two kinds of MODIS data. Due to the MODIS image and their corresponding CDL have different spatial resolution. e.g., MOD13Q1 resolution is 250 m, the resolution of the MOD11A1 is 1 km, but the CDL resolution is 30 m, so resampling operation is needed before the mask processing.

• Finally, we obtain only areas with non-soybean land are masked out, which is a pure NDVI and LST pixel value, supporting the calculation of the crop morphological and physiological indices in this study.

B. Soybean Crop Growing Season Partitioning According to the USDA’s NASS, Soybean phenological

phase mainly includes the following several stage: planted,

Page 3: [IEEE 2014 Third International Conference on Agro-Geoinformatics - Beijing, China (2014.8.11-2014.8.14)] 2014 The Third International Conference on Agro-Geoinformatics - Assessment

emerged (as soon as the plants are visible), blooming (a plant should be considered as blooming as soon as one bloom appears), setting pods (pods are developing on the lower nodes with some blooming still occurring on the upper nodes), dropping leaves (leaves near the bottom of the plant are yellow and dropping, while leaves at the very top may still be green), and harvested. In this study, we select four growth stages, which is from planted to emerged, vegetative period (from emerged to blooming) and reproductive period (from blooming to setting pods) and growing season (from emerged to dropping leaves). In order to take into consideration the cumulative effects of drought in each growth stage, we need to use the piecewise cubic Hermite interpolating polynomial algorithm (PCHIP) [6] method to interpolate the weekly the given percent of soybean growth process data. When the weekly process more than 80%, we set the interpolation points as the growth phase of the point in time, and then we can obtain the corresponding specific date each phonological stage (Table І).

TABLE I. SOYBEAN GROWTH DATES OF PLANTED, EMERGED, BLOOMING, SETTING PODS AND DROPPING LEAVES FROM 2000 TO 2013 IN

IOWA, USA

Year Area Soybean Growth Stage Date a planted emerged blooming setting pods dropping leaves

2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000

Jun. 17 May 19 May 23 May 25 May 24 May 31 May 25 May 23 May 26 May 18 May 28 May 25

Jun. 9 May 13

Jun. 25 Jun. 1 Jun. 6 Jun. 5 Jun. 7

Jun. 17 Jun. 5 Jun. 2 Jun. 7

May 31 Jun. 10

Jun. 5 Jun. 18 May 25

Aug. 5 Jul. 19 Jul. 24 Jul. 24 Jul. 26 Aug. 1 Jul. 20 Jul. 20 Jul. 19 Jul. 22 Jul. 26 Jul. 17 Jul. 30 Jul. 14

Sept. 12 Aug. 29 Sept. 4

Aug. 29 Sept. 2

Sept. 16 Aug. 23 Aug. 24 Aug. 18 Aug. 22 Aug. 24 Aug. 21 Sept. 2

Aug. 17

Oct. 9 Sept. 24

Oct. 1 Sept. 28 Sept. 29

Oct. 3 Sept. 22 Sept. 24 Sept. 21 Sept. 25 Sept. 24 Sept. 23

Oct. 6 Sept. 17

a. the unit of area soybean growth stage is the day, cumulative calculations that need to be converted to the corresponding day, such as blooming on August 5, 2013, corresponding to the 217th day.

IV. METHOD The three major remote sensing derived agricultural

drought monitoring indicators of crop water demand status are: (A) crop morphological indices (NDVI/VCI); (B) Crop physiological indices (canopy temperature index of TCI); (C) Crop comprehensive indices (combination of morphological and physiological, e.g., VSWI), and a measure method of the accumulated magnitude of the agricultural drought can be adopted. Next, these methods will be talked about in details.

A. Crop Morphological Indices Crop morphological indicators reflect crop growing or

appearance. The growth condition of vegetation will have corresponding change when vegetation under water stress. The method of intuitive observation is often used in agronomy, to qualitatively describe the crops developing in a small area. However, quick and real-time monitoring of crop growth status and the damage to crops by drought based on remote sensing can support a wide range of the decision-making on precision crop management. Crop morphology indicators derived from remote sensing generally has normalized difference vegetation

index (NDVI), ratio vegetation index (RVI), Anomaly Vegetation Index (AVI), vegetation condition index (VCI), etc.

• NDVI

The normalized difference vegetation index (NDVI) is considered to be effective ecological indicators of regional or global monitoring vegetation and environmental change. NDVI primary usage is to compare the current state of vegetation with the previous time periods as to detect anomalous condition. Due to vegetation in near infrared band has the higher reflectance, but low reflectance in red band, so the vegetation NDVI value is larger; and rock and bare land in this two band has similar reflectance, NDVI value tend to zero; and cloud, water and snow of red light band reflectance less than near infrared band, NDVI is less than 0. Therefore, NDVI can be used to reflect the vegetation coverage and crop growth status. The higher the NDVI value, the bigger the vegetation coverage degree, and the better the crop condition. These features can also be used for analysis and monitoring drought. In this regard, many scientists discuss and explore how to apply the AVHRR and MODIS data, and achieved many breakthroughs [8].

• VCI

In order to eliminate the spatial variation of the NDVI, reduce the geographical and ecological system variables (mainly include the weather, soil, Vegetation and landform, etc.), and make the NDVI is comparable between the different areas and different time, and Kogan [9] put forward the Vegetation Condition Index (VCI). Formula is as follows:

VCI = NDVIi – NDVImin NDVImax – NDVImin

× 100 (1)

As in (1), NDVIi is a particular year the NDVI value of the ith period; NDVImax and NDVImin represent NDVI value of the maximum and the minimum within the ith period respectively. Kogan and others use VCI index further separate the NDVI derived from geographic attribute changes to estimate the influence of regional drought, and the results prove that VCI is to analyze the regional drought spatial-temporal evolution and quantitatively evaluate crop productivity as a great tool.

B. Crop Physiological Indices Crop canopy temperature is related to the process of energy

absorption and release based on the energy balance, and crop transpiration process will consume heat causes the low canopy temperature. The canopy temperature, when plant has not enough moisture, is higher than applying adequate water, so the canopy temperature can be used as a diagnosis indicator of crop drought.

At present remote sensing monitoring of the canopy temperature indicators commonly have temperature condition index (TCI). In 1995, Kogan [10] believed that only with VCI is not enough, in some cases, to monitor the accuracy of the drought. With the development of the hot channel can receive additional drought information, TCI was developed, and it emphasizes that the relationship between the temperature and plant growth.

TCI = BTmax – BTiBTmax – BTmin

× 100 (2)

Page 4: [IEEE 2014 Third International Conference on Agro-Geoinformatics - Beijing, China (2014.8.11-2014.8.14)] 2014 The Third International Conference on Agro-Geoinformatics - Assessment

As in (2), BTi is a particular year the brightness temperature values of the AVHRR band 4 or the MODIS band 31 and 32 of the ith period, BTmax and BTmin express a brightness temperature during the ith period of maximum and minimum values, respectively.

C. Crop Comprehensive Indices Defining droughts indices based on a single variable or

factor derived from remote sensing method, such as crop morphological indices or crop physiological indices, to some extent, there is a certain lack of their own to reflect the crop drought more accurately and effectively. The comprehensive index may be a practical method of drought monitoring. It can synthetically reflect the effects of drought on crop, many experts and scholars try to synthesize the characteristics of drought indices, in this study, a practical and operable comprehensive index (Vegetation Supply Water Index, VSWI) is chosen [11], which is taking into account the crop response to drought in red light, near infrared and thermal infrared wavelengths. VSWI is usually defined as follows:

VSWI = NDVI / TS (3)

VSWI is a combination of two important parameters of vegetation index (NDVI) and canopy temperature (TS). When drought and high temperature occurs, and the canopy temperature will increase because of crop stomata was forced to shut down, and then the value of VSWI tend to increase, thus, the greater the VSWI value, the more serious the drought. This index is suitable for vegetation coverage; especially the cover crop under the condition of good is very effective. Many researches also show that it is appropriate for vegetation transpiration strong season.

D. Calculating Drought Cumulative Effects The definition of drought has included a beginning date,

ending date, duration, and intensity. Duration of agricultural drought can be either a current duration in each crop progress stage or a whole growth period of drought events from beginning to ending. The above three types of drought index derived from remote sensing were considered the cumulative effect of drought. Drought cumulative Magnitude (DM) is defined as follows:

DM = (∑ ) ⁄ (n-m) (4)

Where m start with the previous stage date and continues to increase until the beginning of the next stage (n) to end point in time, Valuei is the corresponding accumulated value of the growth stage using different remote sensing drought indices. In fact, the accumulation of different drought index has DM is very similar because they reflect the accordant characteristics of drought during corresponding growth stage.

V. RESULTS AND DISCUSSION Based on the above data and three kinds of indicator, which

is agricultural drought monitoring indicator derived from remote sensing of crop water demand, was used to compare which agricultural drought indices that correlated more closely with the soybean crop yields. NDVI and VCI is calculated by using of NDVI values from MOD13Q 16-days vegetation index synthetic products; TCI is calculated by using of Land

Surface Temperature (LST) value from MOD11A1; VSWI is acquired by using of NDVI and LST from MOD13Q1 and MOD11A1 products.

A. The Different Growth Periods of Accumulated Drought To accumulate in different growth period of soybean

drought duration based on the characteristic value of three kinds of drought indices, and four growth stages of soybean special date need to be calculated using interpolation method of PCHIP based on the date of a given progress percentages. Fig. 2 shows the time point of this four growth stages annually (planted, emerged, blooming, and dropping leaves), and their curve keeps accordant variation direction. Every corresponding color vertical line segments represents the average value in this growth stage. We can distinctly see that soybean growing season time are delayed in 2013, 2008 and 2001 year. The duration of the interval between two phases is from 8 to 17 day, from 41 to 52 day, and from 29 to 42, respectively corresponding growing period from planted to emerged, and from emerged to blooming, and from blooming to setting pods. We can lay the foundation for analyzing the effect of drought on soybean crop yield by finding out accumulation effect of drought in different growth stages.

Fig. 2. Soybean annual growing stage date of planted, emerged, blooming,

setting pods and dropping leaves from 2000 to 2013 in Iowa

B. Soybean Yield Parameters and Crop Condition Analysis Iowa is divided into agricultural nine statistics districts for

convenience in compiling and presenting statistical information on crops and livestock. There is a fluctuation of soybean yield statistics output curve graph from 2000 to 2013 in Iowa and its nine districts on the map. The red line represents the annual soybean yield in Iowa, and the blue line represents the nine districts of soybean yield. The graph clearly indicated that the soybean yield data point toward the same trend between the state of Iowa and its nine districts.

Fig. 3. Soybean annual yield value in Iowa (red line) and its nine districts

(blue line) from 2000 to 2013, the unit is BU/ARCE. (data source: USDA NASS Quick Stats)

Page 5: [IEEE 2014 Third International Conference on Agro-Geoinformatics - Beijing, China (2014.8.11-2014.8.14)] 2014 The Third International Conference on Agro-Geoinformatics - Assessment

The wave trough of the soybean yield in Fig. 3 has obvious inflection points from 2000 to 2013, and the kurtosis of the yield curve increases in 2012, 2008 and 2003 led to corresponding annual soybean yield decreased. According to the U.S. National Drought Mitigation Center reported about agricultural weather assessments on July, 2012 (Fig. 4), and we can learn about approximately 87% of soybeans grown in the U.S. is within an area experiencing drought, based on historical NASS crop production data. Moreover, coverage of Iowa almost reached 100% on July 31, 2012 (corresponding to the 212th day) within the reproductive period (blooming to setting pods) of soybean, the influence of drought may be one of the main reasons resulting in a decline soybean yields.

Fig. 4. the U.S. soybean area experiencing drought monitoring map in July

31, 2012. (data source: the National Drought Mitigation Center)

C. Relationships Between Drought Indices and Soybean Yield During the Growing Season The NDVI value every day in the soybean growth progress

were obtained by cubic spline interpolation method based on NDVI 16-day synthetic products,and then according to (4), the NDVI drought accumulated magnitude is calculated, respectively for the four soybean growth stage, including from planted to emerged, from emerged to blooming (vegetation period), from blooming to setting pods (reproductive period), and from emerged to dropping leaves (growing season) (see Fig. 5). We try to find corresponding relationship between accumulated drought inflection points and the soybean yield during the soybean different growth stages.

Fig. 5. NDVI accumulated value in the soybean four growth stages of

planted-emerged, emerged-blooming (vegetation period), blooming-setting pods (reproductive period) and emerged-dropping leaves (growing season)

Correlation and regression analysis were used to estimate the relationship between each drought index and soybean crops yield in the above four growth stages. The correlation coefficient is generated by statistically significant (P). In this

study, due to the vegetation period and the reproductive period are in the good vegetation cover growth stage, to some extent, NDVI can reflect the soybean crop growth status. NDVI accumulated value in the soybean primal planted and emerged stage that has correlation coefficient P > 0.05. It might not be correlation because of the planted period low vegetation coverage, in spite of vegetation index is more suitable for the high vegetation coverage area, and show up low correlation with soybean yield from planted to emerged stage. While the reproductive period (from blooming to setting pods) and the growth season (from emerged to dropping leaves) of soybean showed a close relationship with the correlation coefficient P < 0.05, and reproductive period of the relationship show a slight higher due to the numerical value (P) is smaller (Table ІІ).

TABLE II. THE CORRELATION ANALYSIS BETWEEN NDVI AND SOYBEAN YIELD IN THE FOUR SOYBEAN GROWTH STAGES

No. Soybean Growth Stage a R R Square P (Sig. F)

1 2 3 4

Planted-Emerged Emerged-Blooming Blooming-Setting Pods Emerged-Dropping Leaves

0.025 0.318 0.750 0.734

0.001 0.101 0.562 0.539

0.946 0.370 0.012 0.016

In order to compare three kinds of agricultural drought index derived from remote sensing effect on soybean yield in the different growth stage. We have experimented to observe the R-square values during the soybean growth stage (from emerged to dropping leaves), respectively using morphological indices (NDVI and VCI), physiological indices (TCI), and comprehensive index (VSWI). The results have shown that the VSWI drought index combination of NDVI and LST have correlation with soybean yield (R-square is 0.642), compared to NDVI (R-square is 0.539), VCI (R-square is 0.333), and TCI (R-square is 0.557) was more useful to assess the effect of drought on soybean yield. Although VSWI tends to exaggerate the role of vegetation when the early stage of the crop growth with the poor vegetation cover, and do not make a clear distinction between background and other environmental impacts on the agricultural drought.

VI. CONCLUSION This study proves that drought indices accumulated in

reproductive period (from blooming to setting pods) are slightly superior to other period accumulated for impacting on soybean yield, while over the growth season (from emerged to dropping leaves) is also highly correlated with the total yield assessment. According the above significant correlations analysis suggests that reproductive period of soybean would be potential to obtain the preliminary assessment of agricultural drought effect on soybean yield. Furthermore, this research imply that physiological indices and comprehensive index were more correlated to assess the effect of drought on soybean yield by comparing the above three category of agricultural drought indicator derived from remote sensing. This result can aid on evaluating the effects of drought on soybean yield in different growth stages, and it could be very useful to providing auxiliary decision-making information for drought relief, agricultural manager and grain merchant to plan and prepare for potential drought in advance.

Page 6: [IEEE 2014 Third International Conference on Agro-Geoinformatics - Beijing, China (2014.8.11-2014.8.14)] 2014 The Third International Conference on Agro-Geoinformatics - Assessment

ACKNOWLEDGMENT The authors would like to thank to all the teammate in our

lab and Dr. Shen of China University of Geosciences that gave me valuable comments about this paper. Moreover, we are grateful for USDA’s NASS and NASA’s Goddard Space Flight Center for providing remote sensing and statistics data.

REFERENCES [1] Bannayan Mohammad, Sanjani Sarah, Alizadeh Amin, S. Sadeghi

Lotfabadi, and Azadeh Mohamadian, “Association between climate indices, aridity index, and rainfed crop yield in northeast of Iran,” Field crops research, vol. 118, no. 2, pp. 105-14, April 2010.

[2] Heim and Richard R, “A review of twentieth-century drought indices used in the United States,” Bulletin of the American Meteorological Society, vol. 83, no. 8, August 2002.

[3] Jackson R D, Idso S B, Reginato R J, and Pinter, PJ, “Canopy temperature as a crop water stress indicator,” Water resources research, vol. 17, no 4, pp. 1133-1138, August 1981.

[4] M. Mkhabela, P. Bullock, M. Gervais, G. Finlay, and H. Sapirstein, “Assessing indicators of agricultural drought impacts on spring wheat yield and quality on the Canadian prairies,” Agricultural and forest meteorology, vol. 150, no. 3, pp. 399-410, January 2010.

[5] National Agricultural Statistics Service, USDA-NASS, Washington, DC, 2014. [Online]. Available: http://quickstats.nass.usda.gov

[6] William J. Sacks and Christopher J. Kucharik, “Crop management and phenology trends in the U.S. Corn Belt: Impacts on yields,

evapotranspiration and energy balance,” Agr. Forest Meteorol., vol. 151, no. 7, pp. 882-894, February 2011.

[7] Donald A. Wilhite, “Drought as a natural hazard: concepts and definitions,” Drought, a global assessment, vol. 1, pp. 3-18, 2000.

[8] Tim R. McVicar and David L. B. Jupp, “The current and potential operational uses of remote sensing to aid decisions on drought exceptional circumstances in Australia: a review,” Agricultural systems, vol. 57, no. 3, pp. 399-468, February 1998.

[9] F. N. Kogan, “Remote sensing of weather impacts on vegetation in non-homogeneous areas,” International Journal of Remote Sensing, vol. 11, no. 8, pp. 1405-1419, 1990.

[10] F. N. Kogan, “Application of vegetation index and brightness temperature for drought detection,” Advances in Space Research, vol. 15, no. 11, pp. 91-100, 1995.

[11] Toby N. Carlson, Robert R. Gillies, and Eileen M. Perry, “A method to make use of thermal infrared temperature and NDVI measurements to infer surface soil water content and fractional vegetation cover,” Remote Sensing Reviews, vol. 9, no. 1-2, pp. 161-173, 1994.

[12] Chunming Peng, Liping Di, Meixia Deng, Weiguo Han, and Ali Yagci, “A comprehensive agricultural drought stress monitoring method integrating MODIS and weather data (A case study of Iowa),” Agro-Geoinformatics, 2013 Second International Conference on, pp. 147-52, Aug. 2013.

[13] Jinlong Fan, Mingwei Zhang, Wenbo Xu, Wenzhi Zhang, Chuanshuang Wu, and Jianjun Wu, et al., “A comprehensive crop drought index based on multiple crop condition indicators,” Agro-Geoinformatics, 2013 Second International Conference on, pp. 383-387, Aug. 2013.