| Since the beginning of this century, the seed-maize was launched to plant at "Ganzhou-Linze-Gaotai" region of the Heihe River Basin, which gradually became a national maize seed production base. However, the maize acreage in this region has been appearing fluctuations affected by the production and market problem. It is great significant to forecast the maize production timely and accurately for the local government to develop a reasonable decision on maize acreage adjustment.Based on the analysis of the planting structure situation of the study area, using of HJ-1-A/B images, the ones corresponding to critical growth phase of the major crops (maize and wheat) were selected to estimate a variety of vegetation indices characterizing the vegetation growth. Combined with main crop phonological characteristics of the study area, the information of maize distribution was extracted by the time-series NDVI. The regression models for measured LAI with vegetation indices and vegetation coverage were established, and the distribution information for the LAI was acquired with the empirical relationship. The distribution of temperature and solar radiation was simulated respectively through the establishment of statistical regression models with latitude, longitude and elevation as independent variables and meteorological data as dependent variables. The moisture index (WI) corresponding to each growing season of maize was estimated by Normalized Water Index (NDWI). The LAI-Environmental Factors Model was constructed by LAI in the best phase for crop yield estimation and the major environmental factors; the LAI-Biomass Model was established with LAI and the stress factors that were in each growth season of maize. Ultimately, the maize production of the study area in2012was estimated respectively by these two models. The conclusions are as follows:(1) The LAI-environment model is one of the potentially-stress yield estimation modes. The yield conformation is divided into two independent processes with certain accuracy. However, for it is based on statistical regression analysis, the coefficients are only applicable in the study area. It is more theoretical and could reach certain accuracy for the LAI-Biomass Model. In this model, the crop yield is obtained by harvest index and biomass above-ground, with systemic combination of remote sensing information and stress factors.(2) The coefficient determination R2of LAI-Environmental Factors Model was0.725and the P-value was less than0.05. The maize yield estimated by the model was mainly range from 6500kg’ha-1to7500kg’ha-1.The maximum and minimum maize yield was8526.77kg’ha-1and4002.18kg’ha-1relatively, with an average value of6793.65kg’ha-1, which was slightly larger than6750kg’ha-1, the mean level of maize production provided by the Zhangye Seed Authority.(3) The result of maize yield estimated by LAI-Biomass Model was evaluated by contrast with actual field survey, and the average relative error was about9.63%. The maize yield estimated by the model was also mainly range from5500kg’ha-1to7500kg’ha-1.The maximum and minimum maize yield was7469.56kg’ha-1and3174.71kg’ha-1relatively, with an average value of6180.51k’ha-1, which was less than the mean maize production level of the study area.(4) Overall, the maize yield of study area was mainly ranged from5500kg’ha-1to7500’ha-1, which was almost in line with the overall level provided by the Zhangye Seed Authority. The results indicated that it could be relatively accurate to estimate crop yield based on remote sensing technology, with the impact of environmental factors being considered. |
PDF Full Text Request