Search

• PSTFS optimizes crop type classifications by fully mining crop phenology information. • PSTFS ranks features by importance and then iteratively prunes redundant features. • We used an SVM classifier to produce a high-quality corn map of Heilongjiang, China. • The optimal features identified by PSTFS are key crop phenological characteristics. • PSTFS can be extended to other land cover types and other dense remote sensing data. Accurate information on crop distribution and its changes is important for food security and environmental management. Although time series analysis is a widely-used and useful tool to characterize the seasonal dynamics of crops, the traditional image stacking approach misses important phenological events. This condition makes it difficult to identify the spectral and temporal features that are potentially important for crop identification, and therefore, makes it difficult to determine the optimal feature inputs for classifying crops with both high accuracy and low computation time. To address this gap, we developed a method to automatically select the spectro-temporal features by mining crop phenology information so as to improve the accuracy of crop classifications. This method of Phenology-based Spectral and Temporal Feature Selection (PSTFS) contains two major components: to identify the features with the highest separability between each pair of classes, and to prune redundant features to retain the best for classification. Using this optimal set of features and support vector machines (SVMs), we generated a high-quality corn cultivation map of China's Heilongjiang Province for 2011. The corn map had accuracies greater than 85% and agreed well with the corn census areas. We also demonstrate the goodness of this method for selecting features with high interpretability: it identified two phenological stages (three leaf and milky mature) that could best separate corn from other land use classes in the region. Our approach indicates the great potential for using the PSTFS method in conjunction with SVM classifiers to accurately map crop types based on satellite time series data. [ABSTRACT FROM AUTHOR]

The impact of climate change on the crop water footprint (WF) is highly uncertain, hindering effective agricultural water use in response to climate change. This study proposes a factor sensitivity analysis method, which can screen out statistically significant climatic factors and interactions on crop WF under various uncertainties and determine water resource management measures to mitigate drought for different crops. A special case study was conducted in Heilongjiang Province, China. The results showed that (1) the annual change in crop WF showed a downward trend from 1988 to 2018. The crop WF was dominated by green water footprint (WF green), and the average occupancy rate of WF green in crops was 58.7%–74.1 %; the spatial distribution of WF has latitude zonality. (2) Drought has different effects on WF of different crops, and WF of soybean is susceptible to drought. Wind speed, sunshine hours, and humidity have a greater impact on crop WF in most growth stages. (3) The effect of climatic factors on crop WF varies in different months. The rice WF is mainly affected by the climate in May, and there is an interaction between May humidity and May rain. The WF of maize and soybeans are affected primarily by the climate in July, especially sunshine hours. The proposed approach attempts to analyze that crop WF is affected by not only an individual climatic factor but also their interactions. Crop water management practices should be adjusted based on the results to mitigate the adverse impact of climatic conditions on crop WF during different growing months. [ABSTRACT FROM AUTHOR]

Reliable crop type maps are vital for agricultural monitoring, ensuring food security, and environmental sustainability assessments. Coarse-resolution imagery such as MODIS are widely used for crop type mapping due to their short revisit cycles, which is advantageous for detecting the seasonal dynamics of different crop types. However, the inherently low spatial resolution may restrict their utility for mapping crop types in regions with heterogeneous agricultural landscapes. Agricultural statistics, which provide crop acreage information at different spatial and temporal scales, have the potential to improve crop type mapping from remote sensing. Yet, previous studies have often used agricultural statistics as reference data to evaluate the accuracy of satellite-derived crop type maps but have rarely utilized them to improve crop type distribution mapping. The utility of integrating agricultural statistics with satellite images to produce high-accuracy crop type maps is rarely explored. This study presents a methodology for mapping sub-pixel crop type distributions via the integration of MODIS time series and agricultural statistics. We tested our approach in Heilongjiang Province, which has the highest agricultural production in China. First, we used an optimized random forest regression (RF-r) model with training samples derived from high spatial resolution images (i.e. SPOT and Landsat) to predict the sub-pixel crop type distributions from MODIS time series. To optimize the RF-r model, an 8-day MODIS time series of five vegetation indices in 2011 were used as the candidate independent variables, and a backward feature elimination strategy was implemented to select the best variables for model prediction. Second, we developed an Iterative Area Gap Spatial Allocation (IAGSA) method to spatially reconcile the discrepancies between the crop acreage estimated from MODIS-based maps and the agricultural statistics. We found that the MODIS-derived crop fractions agreed with those derived from the high-resolution images, with R2 > 0.75 for all crop types, yet there was a clear discrepancy between the crop acreage estimated from MODIS and agricultural statistics. The sub-pixel crop type maps adjusted by IAGSA were not only consistent with the agricultural statistics for crop acreage, but also retained the spatial distribution patterns of the original MODIS-derived crop fraction. Our results suggest the advantages of integrating coarse-resolution images and agricultural statistics to map sub-pixel crop type distributions, and to provide consistent estimation of crop acreage. The presented methodology has the potential to map large-scale crop type extent across regions in a cost-effective way. • Integrating MODIS images and agricultural statistics for crop type mapping. • Combing random forest with backward elimination strategy to select optimal features. • IAGSA reconciles the discrepancy between MODIS-based crop area and statistical data. • It is tested in Heilongjiang Province and generated high-quality crop fraction maps. • This method can be extended to other crop types and other coarse resolution images. [ABSTRACT FROM AUTHOR]