Change detection analysis of Cropland using Geospatial technique -A case Study of Narsinghpur District

Access to accurate and up-to-date information on the extent and distribution of individual crop types, associated with land use changes and practices, has significant value in intensively agricultural regions. Explicit information of croplands can be useful for sustainable water resources, land and agriculture planning and management. Remote sensing, has been proven to be a more cost-effective alternative to the traditional statistically-based ground surveys for crop coverage areas that are costly and provide insufficient information. Satellite images along with ground surveys can provide the necessary information of spatial coverage and spectral responses of croplands for sustainable agricultural management. This study strives to differentiate different crop types and agricultural practices to achieve a higher detailed crop map of the Narsinghpur district.


I. INTRODUCTION
Observation and assessment of crop status and development is a crucial topic for agronomic planning and management and for mitigating the effects induced by climate change and extreme events. In order to meet the need of increasing population demand, the timely and precise information on the area covered by different crops is quite necessary (Foerster et al., 2012; Conrad et al., 2014). In particular, in-season crop type maps produced during early growth stages are the key information source for operational agricultural monitoring by both public authorities and private sectors. Early information on crop type and acreage is necessary to forecast agricultural water demand during the summer season (Mo et al., 2005; Reichstein et al., 2007). Despite the need for information delivered in near-real time during the crop season, official figure and statistics are usually provided after the end of the growing season, since data have to be collected, verified and compiled into a database.
Satellite remote sensing is a unique source of data for the identification of crop types over large areas, as described in the last two decades in scientific literature (e.g. Ortiz et al., 1997;Pohl and Van Genderen, 1998 Yang et al. (2011) demonstrated that spatial resolution in the range 10-140 m could be considered an optimal choice for a wide range of agricultural landscapes.

II. MATERIALS AND METHODS Study area
The Study area lies between 23o16' to 24o36' N coordinates and 78o27' to 79o40' E coordinates with respect to the projection of zone no. 43N UTM on WGS 84 datum. It is covering approximately 5133 Sq. km. area. Its elevation range between 286.59 and 882.2 above MSL. The normal annual rainfall of Narsinghpur district is 1192.1mm. There are four tehsils fall under this district namely Narsinghpur, Gotegaon, Gadarwara & Kareli and the district further divided into six administrative blocks namely Saikhera, Babai Chichali, Chawarpatha , Kareli, Narsinghpur & Gotegaon. The study area is illustrated in Fig-1.  (Table-1).

Methodology
The methodology flowchart is illustrated in Fig-2. Different registrations for a specific sensor, after preprocessing, form a "multivariate image set". These data contain two types of information: spectral-temporal and spatial-contextual. The spectral-temporal information can be extracted with a supervised Maximum Likelihood algorithm (Duda et al.2000), resulting in a per-pixel classification. The spatial information can be derived by means of a signature data set collection on the basin of histogram group. On the ideal situation, these group correspond with parcels on the grounds. In this study the segmentation was applied with supervised classification tool which is present in ERADAS Imagine software. Classification and the segmentation are combined using signature dataset collection of different crop type reflectance. This procedure determines for each parcel the pre-dominant class and assign all Pixels of the parcel to this modal class. By reducing speckle and errors in the vicinity of field boundaries, this application enhances the accuracy and legibility of the final map. The ground truth data needed for the calibration of supervised classifications were collected in two field surveys in before classification and after classification. The garmin GPS handset and the associated software were used for the surveys. During the field surveys, large parcels or plots corresponding different crops were selected and identified and demarcated in FCC to form a ground truth vector GIS. Two mappings as per year 2006 and 2015 to estimate the areas changes of crops.

Classification Procedure
The digital image processing of satellite data has been carried out using ERDAS IMAGINE 9.   (Table-2).

Crop cover changes
Fertile alluvial soils and the availability of water through canals as well as ground water sources support intensive agricultural activity in the region. The population pressure, economic potential of the region and the entrepreneurial attitude of a majority of the farmers in the region have significantly transformed the crop cover pattern of the area, especially during the recent years. A study has been conducted by Rao et al. (1991) which was based on the interpretation of IRS-1A LISS-I image on 1:250,000 scale. They observed that due to changes in cropping pattern from seasonal crops to long term crops, there has been an increase in the cropping intensity by about 160% in the deltaic region during 1980s when compared to the earlier period. In the present study also the Landsat imagery of 2015 are found to be useful, especially for identifying broad categories of crop cover required for the purpose of this study. Large-scale exploitation of groundwater resources during the recent years has increased the area of more water requirement crops possible instead of less water requirement crops earlier. Further, the area under

International Journal of Environment, Agriculture and Biotechnology (IJEAB)
Vol -2, Issue-4, July-Aug-2017  http://dx.doi.org/10.22161/ijeab/2.4.33  ISSN: 2456-1878 www.ijeab.com Page | 1728 wheat and sugarcane has increased during the last ten years: This is evident from the observation of the satellite imagery. While 2006 image (Fig-3) shows maximum area under seasonal gram crop, the 2015 satellite image (Fig-4) shows wheat cover in many parts of the study area. The progressive increase in the wheat crop as well as sugarcane crop extent in the district is revealed by the area statistics obtained from the G1S analysis of the two datasets ( IV. CONCLUSION Based on, a supervised classification method is proposed for identifying crops at the level in agricultural land by using multispectral data. It is concluded that the Agricultural practices in the study area have altered significantly in 10 years. The crop area overlapped in the fallow/barren area was evident by the development in of the canal in the area and augmentation of area covered by of wheat (109.25%) and sugarcane (115.21%) crop while decline in gram (-35.92%) and other cropped area ( -24.57%). The Change Detection analysis is an efficient way of describing the changes observed in each category. Study implies that in the year 2006 agricultural area were found 231889.47 ha, while in the year 2015 area were 264284.8 ha, which is easily identified through classification that open /fallow land converted in to Agriculture/other vegetation. The supervised method gives quite satisfactory results for vegetation varying in densities and also for scattered vegetation from a multispectral remote sensing image. Our findings demonstrate that near real-time, in season crop mapping is feasible using satellite data with suitable spatial and temporal resolution in a simple, operational and inexpensive way. This early in-season crop map could be useful to support agricultural practices and management, especially for supporting the analysis of water demand for major crops during dry summer months.

ACKNOWLEDGEMENT
Authors are highly thankful to data provider national and international agencies. Also acknowledge the RS and GIS Lab of Department of Soil and Water Engineering, College of Agricultural Engineering, JNKVV University-482004, providing complete lab facility for study and research work.