基于CASA模型的呼伦贝尔牧草产量估算研究

Research on the Estimation of Pasture Yield in Hulunbuir Based on Casa Modeling

摘要: 呼伦贝尔草原是著名的天然牧场，畜牧业也是呼伦贝尔草原地区的基础性产业，充分发挥畜牧业优势对于提高草原地区牧民收入有着重大意义。为获得长时序牧草产量数据，本文利用地理信息和遥感技术，基于CASA模型估算出呼伦贝尔市牧业五旗2001~2020年牧草生长季(5~8月)的牧草产量，并分别用各牧业旗生态观测站的实测产量数据对模拟产量数据进行精度验证。本研究首先阐释CASA模型的基本原理及模型框架，介绍CASA模型中估算牧草产量的两个关键参量：植被吸收的光合有效辐射和实际光能利用率。通过太阳总辐射和植物光合有效辐射吸收比例确定植被吸收的光合有效辐射，通过温度胁迫因子、水分胁迫因子、最大光能利用率确定实际光能利用率，最终计算出2001~2020年呼伦贝尔市牧业五旗牧草生长季的NPP。根据2001~2020年呼伦贝尔地区逐年的土地覆盖数据提取草原用地，经空间统计分析提取出牧业五旗的牧草生长季(5~8月)逐年平均NPP，进而转换成牧草产量。对比分析牧草估测产量与实测产量之间的相关性系数、均方根误差、平均绝对误差三种精度评价指标，验证CASA模型估产精度。实测产量值与估测产量值的相关性系数分布区间为0.76~0.82、均方根误差区间为20.42~24.21 g/m²、平均绝对误差区间为17.94~23.19 g/m²，三项误差指标均在合理误差范围内。结果表明CASA模型模拟出的牧草产量数据精度较高，可为呼伦贝尔畜牧业高质量发展提供技术支撑。

Abstract: Hulunbuir grassland is a famous natural pasture, and animal husbandry is also a basic industry in Hulunbuir grassland area, giving full play to the advantages of animal husbandry is of great significance to improve the income of herdsmen in the grassland area. In order to obtain the long time series pasture yield data, this paper estimates the pasture yield of the pasture growing season (May-August) of the five flags of Hulunbuir City pastoralism from 2001 to 2020 based on CASA model using geographic information and remote sensing technology, and verifies the accuracy of the simulated yield data with the measured yield data from the ecological observatory of each flag of pastoralism, respectively. In this study, the basic principles and model framework of CASA model were firstly explained, and two key parameters for estimating pasture yield in CASA model were introduced: photosynthetically active radiation absorbed by vegetation and actual light energy utilization rate. The photosynthetically active radiation absorbed by vegetation was determined by the ratio of total solar radiation and plant photosynthetically active radiation absorption, and the actual light energy utilization was determined by the temperature stress factor, moisture stress factor, and maximum light energy utilization, and the NPP of the growing season of forage grass in the five flags of pastoral industry in Hulunbuir City from 2001 to 2020 was finally calculated. Based on the year-by-year land-cover data of Hulunbuir area from 2001 to 2020, the NPP of the growing season of pasture grass was extracted. Grassland land was extracted, and the yearly average NPP of the pasture growing season (May-August) in the five banners of pastoralism was extracted by spatial statistical analysis, and then converted into pasture yield. The correlation coefficient, root-mean-square error, and average absolute error of three accuracy evaluation indexes between the estimated and measured pasture yield were comparatively analyzed to verify the accuracy of CASA model estimation. The distribution range of correlation coefficient between measured and estimated yield values was 0.76~0.82, the root mean square error range was 20.42~24.21 g/m², and the average absolute error range was 17.94~23.19 g/m², and the three kinds of error indexes were all within the reasonable error range. The results showed that the CASA model simulated pasture yield data with high accuracy, which can provide technical support for the high-quality development of Hulunbuir’s livestock industry.