Remote Sensing For Crop Phenology Detection

Agricultural production is the basis for the survival of human society, a wide range of reliable agricultural information is essential to the development of the grain market, related policies, the basis for the protection of regional and global food security. Modern remote sensing technology with low-cost, high time resolution and high spatial resolution can achieve large-area environmental monitoring, crop growth, crop yield, etc. It is one of the most important technique for the modernization of agricultural production. Remote sensing overwhelm the traditional method with large area observation and high accuracy. In recent years, it is a trend that remote sensing of agriculture uses multi-scale, multiple sensors data and is developed towards precision agriculture. Accurate phenology information extraction are the basises for the other applications of remote-sensed based agricultural, such as crop fertilizer and irrigation schedule, crop yield estimation, etc. So far, there have been a lot of phenology estimation models based on the remote sensing derived time-series vegetation index curve, but little attention was paid on the crop varieties, microscopic mechanism of crops and the environmental factors, most of the methods and models for crop phenology extraction basically based on the local feature point from the remote sensing derived time-series vegetation index curve. The main purpose of this paper is to propose a new phenology detection model combined with physical mechanism based crop models and remote sensing based crop phenology detection method (e.g. corn and soybeans), and conduct quantitative validation in multi-regions on both field scale and regional scale.Here presents three closely related topics in this work as following:(1) Air temperature (Ta) is a key input in a wide range of agroclimatic applications. In this sdudy, air temperature is an important input for the new proposed phenology detection model. Moderate Resolution Imaging Spectroradiometer (MODIS) Ts (Land Surface Temperature (LST)) products are widely used to estimate daily Ta. However, only daytime LST (Ts-day) or nighttime LST (Ts-night) data have been used to estimate Tmax/Tmin (daily maximum or minimum air temperature), respectively. The relationship between Tmax and Ts-night, and the one between Tmin and Ts-day has not been studied. In this study, both the ability of Ts-night data to estimate Tmax and the ability of Ts-day data to estimate Tmin were tested and studied in the Corn Belt during the growing season (May-September) from 2008 to 2012, using MODIS daily LST products from both Terra and Aqua. In addition, according to the surronding land cover types, the stations was classified into three categories:crops, forest and developed areas to analysis the influence of different land cover types to the estimation of air temperature using land surface temprature. The results show that using Ts-night for estimating Tmax could result in a higher accuracy than using Ts-day for a similar estimate. Combining Ts-day and Ts-night, the estimation of Tmax was improved by 0.19-1.85,0.37-1.12 and 0.26-0.93 ? for crops, deciduous forest and developed areas, respectively, when compared with using only Ts-day or Ts-night data. The main factors influencing the Ta estimation errors spatially and temporally were analyzed and discussed, such as satellite overpassing time, air masses, irrigation, etc.(2) Monitoring crop phenology provides essential information for irrigation scheduling and fertilizer management, as well as for understanding regional to global scale vegetation dynamics and estimating crop productivity. In this study, a new phenology detection method is presented that incorporates the "shape-model fitting" concept of the Two-Step Filtering method and a simulation concept of the crop models to detect all critical vegetative stages and reproductive stages of corn (Zea mays L.) and soybeans (Glycine max L.) from MODIS 250-m Wide Dynamic Range Vegetation Index (WDRVI) time-series data. At the field scale, the method was tested over a ten-year period (2003 to 2012) for three experimental study sites in eastern Nebraska, where the estimated phenology dates were compared to the ground-based phenology observations for both crops over the three field sites. The root mean square error (RMSE) of phenology stage estimation across all stages ranged from 1.99 to 4.30 days for corn and from 1.93 to 4.88 days for soybeans. The approach was also tested at a regional scale over eastern Nebraska and the state of Iowa to evaluate its ability to characterize the spatial-temporal variation of the targeted corn and soybean phenology stage dates over a larger area. Quantitative assessments at the regional scale were conducted by comparing the estimated crop stage dates with crop developmental stage statistics reported by the USDA NASS Crop Progress Reports (NASS-CPR) for both eastern Nebraska and Iowa. The accuracy of the regional-scale phenology estimation in Iowa (RMSE ranged from 2.63 to 3.91 for corn and from 3.24 to 3.87 for soybeans) was slightly lower than in eastern Nebraska (RMSE ranged from 1.79 to 2.87 for corn and from 1.75 to 2.94 for soybeans), However, the estimation accuracy in Iowa is still reasonable with RMSE of phenology date estimates being within 4 days or less of the observed date for both corn and soybeans.(3) Northeast China is one of the bread baskets in China, the foregoing method in the US Corn Belt shows good results, but the agriculture in the US has its own unique character, such as the large size of single piece of cropland which facilitate remote sensing observation, especially for the moderate spatial resolution sensors, high level of agricultural mechanization level wich make the planting area more consistent. Therefore, to validate the applicability and dependability the foregoing models and MODIS data in Northeast China could help to improve the models, as well as promote the high-precision remote sensing based crop monitoring in Northeast China. In this study, both air temperature estimation model and phenology estimation model developed in US corn belt were applies to China's northeast Jilin and Liaoning Provinces. The experimental results show that using Ts-night for estimating Tmax in northeast China could also result in a higher accuracy than using Ts-day for a similar estimate. Compared to the field phenlogy observation data from the agriculture meteorological stations, the corn phenology estimation accuracy in Northeast China using only MODIS data is much lower than that in the US field sites (RMSE of corn phenology estimate in Northeast China from 3.78 to 8.41 days). The main reason might be the small size of single piece of cropland in northeastern China which led to the mixed MODIS pixel. Accordingly, Landsat and MODIS data fusion method was applied to get time-series images with Landsat-like spatial resolution and MODIS-like temporal resolution, quantitative and qualitative validation was conducted to evaluate and verify the accuracy of the data fusion. The phenology information was extracted from the time-series vegetation index curves which were derived from the time-series high spatial and temporal resolution images. The results show that data fusion of Landsat and MODIS ensured the temporal resolution of time-series imges, to some extend improved the spatial resolution and decreased the influence of mixed pixel.Due to the limited time and knowledge of the auther, the research in this paper is to be further studied. It mainly includes:1) the factors that effect the estimation of air temperature based on LST included in this paper is still limited, more factors should be considered to improve the estimation accuracy, e.g. the diurnal temperature change, vegetation index, water stress, etc.2) only two important factors:temperature and photoperiod, which effct crop phonological development were included in the phonology detection model of this paper, further research would be done to improve this model, such as considering the influence of water stress. In addition, the idea of the phenology detection model in this paper could be extended to more types of crops, such as wheat, cotton, rice etc.3) to solve the problem of low resolusion of MODIS data, data fusion method was conducted in this area. However the sparse agricultural meteorological stations, and the lack part of ground observation data. The avaliable data is not enough for the validation of phenology detection model proposed in this paper based on the result of data fusion. For the further work, more data would be collected to verify the accuracy of phenology dectection based on the data fusion result.