Home Articles Assessment crop yield estimation methods by using satellite imagery

Assessment crop yield estimation methods by using satellite imagery


Sanaz Shafian
Student of Remote sensing
K.N.T.University
[email protected]

Dr.M Valadanzouj
Professor associate of K.N.T.University
[email protected]

 

 

Abstract
Because of the population increment there is a growing need for micro-level planning and particularly the demand for crop insurance (Anup, 2005), which increases the need for field level yield statistics. Therefore crop yield estimation is very important in national and regional scale.

Crop yield is necessary, particularly in countries that depend on agriculture as their main source of economy. Such predictions warn the decision makers about potential reduction in crop yields and allow timely import and export decision. Therefore policy of agricultural- economic and yield price are influence by the accuracy and speed crop yield estimation.

There are two methods for yield estimation: Conventional methods and Remote sensing methods.
Conventional methods are often complicated, costly, time consuming and they can not be run in large scale. Therefore it is necessary to use cheaper/faster methods for crop yield estimation.

Remote sensing data has the potential and the capacity to provide spatial information at global scale; of features and phenomena on earth on an almost real-time basis.

They have the potential not only in identifying crop classes but also of estimating crop yield.
In this paper first, conventional methods and their disadvantages are discussed. Then crop yield estimation methods based on remote sensing data, their advantages and disadvantages are discussed.

Introduction:
Crop yield estimation is very important in national and regional scale (Anup 2005).
Because of the population increment there is a growing need for micro-level planning and particularly the demand for crop insurance (Anup 2005), which increases the need for field level yield statistics.
Crop yield estimation has an important role on economy development (Hayest and Decker, 1996).These predictions warn the decision makers about potential reduction in crop yields and allow timely import and export decision.

Crop yield estimation in many countries are based on conventional techniques of data collection for crop and yield estimation based on ground-based field reports (Reynolds et al.2000).These methods are costly, time consuming and are prone to large errors due to incomplete ground observations, leading to poor crop yield assessment and crop area estimations. In most countries the data become available too late for appropriate actions to be taken to avert food shortage.

Objective, standardized and possibly cheaper/faster methods that can be used for crop growth monitoring and early crop yield estimation are imperative.

Remote sensing data has the potential and the capacity to provide spatial information at global scale; of features and phenomena on earth on an almost real-time basis.

They have the potential not only in identifying crop classes but also of estimating crop yield.
Most studies have shown that there is high correlation between vegetation spectral index extracted from satellite images and the green biomass and yield. Therefore, combining vegetation spectral index and the green biomass and yield can be used to estimate yield before harvesting (Groten, 1993).

There are many ways to estimate crop yield by satellite image data.

Agricultural production is a result of complex environmental such as solar radiation, water consumption and etc. Objective is a crop yield estimation method that can estimate crop yield as a function of these factors by minimum time and cost and maximum accurate.

2. Conventional crop yield estimation methods
For many years, crop yield estimation has been very important for government.
There are many conventional methods to estimate crop yield. These methods are based on field reports.
There are two conventional methods to estimate crop yield: empirical-statistical models and crop growth models (Jorgensen, 1994).
In a definite region, empirical -statistical models consider crop yield for many years and effective factor on crop yield are found. Then crop yield is related to effective parameter by an empirical equation and the coefficient of each factor is found. Now by these coefficients, crop yield is estimated. Every set of empirical models relate crop yields to one set factors .In the most relations, effective factors are environmental.

Crop growth models estimate crop yield as function of complex interaction of different physiological processes with environment. These models estimate biomass production potential by daily crop growth simulator. Running of these methods has too many difficult such as: require too many ground factors, lake of data in a correct form and much cost. Conventional methods have some disadvantages:

  1. They can not consider over field
  2. Costly and time consuming
  3. They are not real time

In these days precision farming is running in most countries. Conventional methods can not consider all crops simultaneously.
Therefore purpose is achieved to method that removes these disadvantages.
Remote sensing methods remove above disadvantages simultaneously (X.Mo et al, 2004).
3. Crop yield estimation methods by using satellite images and ground observation use of satellite images in agriculture were started from 1970.

In those days uses of these methods were not tradition because satellite images were expensive and their spatial resolution was low.
From 1990, above disadvantages were removed and high spatial and spectral resolution images were produced and most researches used these methods.

Now, most country uses these methods because there is a big archive of these images and we can use them easily.
Now we have images that are produced with electromagnetic wave reflection and different vegetation indices are calculate from these images. These indices are commonly used for real time evaluation of vegetation health and productivity because green mass and content of water, protein…have effect on wave’s reflection (Anup, 2005).

These methods are divided into 3 categories: 1-Remote sensing methods based on Empirical -statistical models.2- Remote sensing methods based on water consuming balance model.3- Remote sensing methods based on biomass estimation models.
The recent method based on biomass estimation methods are divided to two groups: crop growth models (MASS, 1988) and Monteith model (Monteith, 1972)

3.1. Remote sensing methods based on Empirical -statistical models
These methods are based on conventional methods, but here spectral indices are calculated from satellite images and not from ground measurement. Indies such as NDVI1 which calculation them by conventional methods are timely and consuming.

3.2. Remote sensing methods based on water consuming balance model
These methods estimate crop yield as a function of evaporation fraction during crop growth stages and use water consuming balance model to estimate evaporation fraction. At first whole growth period is divided to ten days sets and then evaporation fraction is calculated in these sets. If the growth period is divided to 30 days sets, some changes in crop water will be ignored also if model runs in daily format; it will be costly and time consuming.

3.3.1. Crop growth model
Crop growth models focus on complex interaction of different physiological processes with environment. In fact these models describe growth stages. There are many ways to combine crop growth models and spectral observation from satellite data were initially described by MASS(1988)and their classification was revisited by Delecolle (1992).Three methods of data integration have been identified:

  1. Direct use of a driving variable estimated from remote sensing information in the model;
  2. The updating of state variable of the model (for example LAI) derived from remote sensing data;
  3. The calibration of model variables by using satellite images (assimilation method)

The general strategy of the model/observations coupling consists of driving variables or parameters which directly occur in the modeling procedure from radiometric observations.

The direct use of remote sensing data to derive a variable assumes that remote sensing data are available at an adequate time step (from daily to weekly) .Due to cloud contamination and intrinsic properties of sensors, this is rarely case .Therefore other approaches should be used.

Gaps between dates must therefore be filled by some interpolation procedure.
Substitution of a simulated data value by an observed one (actually derived from the observed reflectance) suggests that simulated of data is flawed, and therefore the biophysical processes are not well described by the model. But a good description of those processes is required to obtain a consistent estimation of variables such as crop Biomass, which can not be monitored directly by remote sensing.

Assimilation method consists of minimizing the difference between a derived state variable radiometric signal and it’s simulated. Difference between satellite observed and simulated value is minimized model parameter calibration.

3.3.2. Monteith model
A simple and useful paradigm for modeling crop yield with remote sensing is derived from Monteith (1972). This model use Biomass to estimate crop yield.

(1) Biomass=APAR*e (Monteit, 1972) (2) Crop yield=APAR*e*HI
Where:
e: the light-use efficiency in units of g biomass MJ-1
HI: harvest index
APAR: absorbed photo synthetically active radiation

Variability in e can result from a variety of nutrient, water. Numerous studies have demonstrated that if not water short, and temperature is optimal, e is a relatively constant property of plants.

In some calculation the effect of temperature and soil moisture is considered for accuracy increment.
Like e, HI is a relatively constant. Values of this factor are experimentally determined and described in the international literature. It can be calculated from crop information in the last years.

Variability in e and HI can result from a variety of nutrient, water, and temperature stresses (Russell et al., 1989).
APAR is a fraction of PAR2 that absorbed by canopy. Richards and Townley -Smith (1987) indicated that the proportion of water used after synthesis affects the harvest index.

4. Result and discussion
In this paper crop yield estimation methods were discussed. Conventional methods were based on ground reports. Those were timely, consuming and could not consider over field therefore were prone to large errors due to incomplete ground observations, leading to poor crop yield assessment and crop area estimations (Reynolds et al.2000).

Remote sensing methods removed above disadvantages simultaneously (X.Mo et al, 2004).
Remote sensing methods based on Empirical -statistical models should be calibrated in other regional because factor weights were different in each region and those could not be run in large scale also ignored the effect of another factors.

Remote sensing methods based on water consuming balance model could be run in large scale but those ignored the effect of many parameters such as: solar radiation, photosynthesis magnitude…
Crop growth models were complete models that considered the most effect parameters. They could be run in large scale but entered to much agriculture science details and had lots of parameters. Therefore their runnings were costly and time consuming.

Monteith model considered effect of solar radiation and photosynthesis by APAR calculation and effect of temperature and soil moisture on crop yield. That model had few parameters that could be calculated from satellite images .That model could be run in large scale. Therefore, Monteith model estimated crop yield by maximum accuracy.

References:

  • Anup K. Prasad a, Lim Chai b, Ramesh P. Singh a, b,*, Menas Kafatos b (2006). Crop yield estimation model for Iowa using remote sensing and surface parameters. International Journal of Applied Earth Observation and Geoinformation 8 (2006) 26-33
  • Groten, S. M. E. (1993). NDVI – crop monitoring and early yield assessment of Burkina Faso. International Journal of Remote Sensing, 14(8), 1495-1515.
  • Hayes, M.J., Decker, W.L., 1996. Using NOAA AVHRR data to estimate maize production in the United States Corn Belt. Int. J.Remote Sense. 17, 3189-3200.
  • Jorgensen, S.E., 1994. Models as instruments for combination of ecological theory and environmental practice. Ecol. Model. 75-76,5-20.
  • Mass, S. J., 1988. Use of remotely-sensed information in agricultural crop growth models. Ecological Modeling. 41,247-268
  • Moa,*, S. Liua, Z. Lina, Y. Xub, Y. Xianga, T.R. McVicarc (2005). Prediction of crop yield, water consumption and water use efficiency with a SVAT-crop growth model using remotely sensed data on the North China Plain. Ecological Modelling 183 301-322
  • Monteith, J.L., 1972.Solar radiation and productivity in tropical ecosystems. J.Appl. Ecol. 9,747-766.
  • Monteith, J.L., 1977.Climate and the efficiency of crop production Britain. Phil. Trans.Roy.Soc.Lond. B 281,277-294.
  • Reynolds, M. Yittayew, D. C(2000).Slack, Estimation crop yields and production by integrating the FAO Crop Speci. C Water Balance model with real-time satellite data and ground-based ancillary data. Int.g. remote sensing, 2000, vol,21, no.18,3487-3508
  • Richards, R.A., Townley-Smith, T.F., 1987. Variation in leaf area development and its effect on water use, yield and harvest index of drought wheat. Aust. J. Agric. Res. 38, 983-992.
  • Russell, G., Jarvis, P.G., Monteith, J.L., 1989. Absorption of radiation by canopies and stand growth. In: Russell, G., Marshall, B., Jarvis, P.G. (Eds.), Plant Canopies: Their Growth,Form and Function.