随机森林算法在全球干旱评估中的应用
方秀琴(1978— ),女,安徽池州人,博士,副教授,主要从事地表参数遥感反演、水文模型集成及水旱灾害防治等研究。E-mail: kinkinfang@hhu.edu.cn |
收稿日期: 2020-08-19
要求修回日期: 2020-12-23
网络出版日期: 2021-08-25
基金资助
国家重点研发计划项目(2016YFA0601500)
国家自然科学基金项目(42071040)
中央高校基本科研业务费专项(2019B04714)
版权
Application of Random Forest Algorithm in Global Drought Assessment
Received date: 2020-08-19
Request revised date: 2020-12-23
Online published: 2021-08-25
Supported by
National Key Research and Development Program of China(2016YFA0601500)
National Natural Science Foundation of China(42071040)
Fundamental Research Funds for the Central Universities(2019B04714)
Copyright
干旱是发生频率最高,造成社会、经济损失和生态破坏最严重、最广泛的自然灾害之一,因此对干旱进行可靠、有效的评估十分重要。本文以月平均降水、月平均温度、月最高温度、月最低温度、土壤湿度、蒸散发、NDVI、叶绿素荧光等作为解释变量,以基于SPI的干旱等级作为目标变量,采用随机森林算法,以2007—2012年的数据作为训练数据,以2013—2014年的数据作为预测数据,对全球11个气候区分别建立干旱等级评估模型。研究结论如下:SPI的时间尺度影响模型精度,在基于SPI1、SPI3、SPI6和SPI12划分的干旱等级的评估模型中,以基于SPI1的干旱等级为目标变量的模型的预测精度(60%~75%)较高,且模型能够捕捉到EM-DAT旱灾记录次数的90.91%、月份的78.47%,表明该模型对实际干旱事件具有良好的评估性能;干旱等级划分标准对模型的预测性能影响较小,可根据需求选择标准I(干旱/非干旱)或标准Ⅱ(重旱/非重旱)进行干旱评估;解释变量的相对重要性与SPI的时间尺度和气候差异等因素有关。降水对基于SPI1的干旱等级的重要性最大,随着SPI时间尺度的增加,降水的重要性逐渐减小,温度、土壤湿度、NDVI和ET的重要性逐渐增大。降水以外的其他变量在不同气候区的重要性不同。在热带气候区、亚寒带气候区和苔原气候区,温度或蒸散发的影响较大;在干燥气候区,土壤湿度的影响较大;在温带气候区,仍以降水的相对重要性最大;在湿润大陆性气候区,植被对干旱的影响较大。
方秀琴 , 郭晓萌 , 袁玲 , 杨露露 , 任立良 , 朱求安 . 随机森林算法在全球干旱评估中的应用[J]. 地球信息科学学报, 2021 , 23(6) : 1040 -1049 . DOI: 10.12082/dqxxkx.2021.200474
Drought is one of the most frequent and widespread climate extremes, causing devasting social, economic and ecological damages. It is of key importance to evaluate drought reliably and effectively. In this study, in order to assess global drought grade, the Random Forest (RF) algorithm was used to establish the drought grade assessment models for the 11 climate zones in the world. We chose monthly mean precipitation, mean temperature, maximum temperature, minimum temperature, soil moisture, evapotranspiration (ET), Normalized Difference Vegetation Index (NDVI), and Sun/Solar-induced Chlorophyll Fluorescence (SIF) as explanatory variables and drought grades based on Standardized Precipitation Index (SPI) as target variable. The SPI on different timescales of 1 month, 3 months, 6 months and 12 months were labeled as SPI1, SPI3, SPI6 and SPI12, respectively. The data from 2007 to 2012 were used as training data of the assessment models while those from 2013 to 2014 were used as prediction data. The results showed that: (1) The temporal scale of SPI influenced the model accuracy. Among the models with drought grade based on SPI1、SPI3、SPI6 and SPI12, the one with drought grade based on SPI1 had the highest accuracy (60%~75%) and prediction performance. The model with drought grade based on SPI1 was able to capture 90.91% of the drought records in the global emergency events database (EM-DAT). It could capture 78.47% of the drought duration month in the EM-DAT. The agreements with records and drought duration month in the EM-DAT indicated the good performance of the drought grade assessment model based on 1-month SPI and RF algorithm. (2) The drought grading criterion had little impact on the model performance. Users could select criterion I (drought/not drought) or criterion II (severe/not severe) depending on the real needs. (3) The relative importance of each explanatory variable depended on both the temporal scale of SPI and climatic differences. Precipitation was the most important factor for the drought grade based on SPI1. The importance of precipitation decreased and the ones of other explanatory variables such as temperature, soil moisture, NDVI, and ET increased as the timescale of SPI increased. The importance of variables except precipitation showed differences in different climate zones. Among the tropical, subfrigid, and tundra climate zones, temperature or ET is relatively important for drought. Soil moisture is relatively important in dry climate zone and precipitation is the most important in mild temperate climate zone, while vegetation is relatively important in the humid continental climate zone.
Key words: drought evaluation; drought grade; SPI; random forest; climate zone; precipitation; temperature; soil moisture
表1 气候区划分条件及文中缩写[11]Tab. 1 Climatic condition and acronym for each climate zone |
气候区 | 归类条件 | 文中缩写 |
---|---|---|
热带雨林气候 | Tmin≥18 °C,Pmin≥60 mm | Ay |
热带季风气候 | Tmin≥18 °C,Pann≥(100-Pmin)×25 | Aj |
热带稀树草原气候 | Tmin≥18 °C,Pann<(100-Pmin)×25 | Ax |
半干旱气候 | 5 Pth<Pann<10 Pth | Bb |
沙漠气候 | Pann≤5 Pth | Bs |
地中海气候 | -3°C<Tmin<18 °C,Psmin<Pwmin,Pwmax>3Psmin,Psmin<40 mm | Cd |
湿润亚热带气候 | -3°C<Tmin<18 °C,Tmax≥22 °C | Cs |
海洋性气候 | -3°C<Tmin<18 °C,Tmax<22 °C | Ch |
湿润大陆性气候 | Tmin≤-3 °C,至少有4个月的Tmon≥10 °C | Ds |
亚寒带气候 | Tmin≤-3 °C,Tmon≥10 °C少于4个月 | Dy |
冰原气候 | Tmax<0 °C | Eb |
苔原气候 | 0 °C≤Tmax<10 °C | Et |
表2 研究数据基本信息Tab. 2 Information of the data |
数据 | 来源 | 时间分辨率 | 空间分辨率 | 时间序列 |
---|---|---|---|---|
SPI | http://amir.eng.uci.edu/ | 月 | 0.625° | 1980.06—2016.12 |
SM | http://cci.esa.int/ | 月 | 0.25° | 1978.11—2016.12 |
ET | http://www.ntsg.umt.edu/ | 月 | 0.05° | 2000.01—2014.12 |
PRE | https://crudata.uea.ac.uk/ | 月 | 0.5° | 1901.01—2017.12 |
TMN | https://crudata.uea.ac.uk/ | 月 | 0.5° | 1901.01—2017.12 |
TMP | https://crudata.uea.ac.uk/ | 月 | 0.5° | 1901.01—2017.12 |
TMX | https://crudata.uea.ac.uk/ | 月 | 0.5° | 1901.01—2017.12 |
NDVI | https://modis.gsfc.nasa.gov/ | 月 | 1 km | 2000.02至今 |
SIF | https://avdc.gsfc.nasa.gov/ | 月 | 0.5° | 2007.01—2018.10 |
表3 不同时间尺度SPI反映的现象及其相关应用Tab. 3 Phenomena reflected by SPI on different timescales and their applications |
SPI时间尺度 | 反映现象 | 应用 |
---|---|---|
1个月 | 短期湿度条件 | 短期土壤水分和作物胁迫(特别是在生长季节) |
3个月 | 短期和中期湿度条件 | 对降水量的季节估计 |
6个月 | 中期降水趋势 | 表征不同季节降水的潜力 |
12个月 | 长期降水模式 | 与水流、水库水位和地下水水位有关 |
表4 标准化降水指数干旱等级划分表Tab. 4 Classification of the drought grades indicated by SPI |
分类标准 | 等级 | SPI |
---|---|---|
标准Ⅰ | 非干旱 | SPI >-1.0 |
干旱 | SPI≤-1.0 | |
标准Ⅱ | 非重旱 | SPI >-2.0 |
重旱 | SPI≤-2.0 |
表5 各气候区8个解释变量对基于SPI1、SPI3、SPI6、SPI12的干旱等级的相对重要性排序Tab. 5 Rank of the relative importance of the 8 explanatory variables for drought grades based on different SPI (SPI1, SPI3, SPI6, SPI12) in each climate zone |
重要性排序 | 热带雨林气候 | 热带季风气候 | 热带稀树草原气候 | 半干旱气候 | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | |
1 | PRE | PRE | TMX | TMX | PRE | TMX | TMX | TMX | PRE | ET | ET | ET | PRE | PRE | SM | SM |
2 | TMX | TMP | SM | SM | TMP | TMP | TMP | TMP | ET | PRE | PRE | TMX | SM | SM | PRE | PRE |
3 | TMP | TMX | TMP | TMN | TMN | PRE | SM | SM | TMN | TMX | TMN | PRE | ET | ET | TMX | TMN |
4 | SM | SM | PRE | ET | TMX | SM | TMN | TMN | TMX | TMN | SM | TMN | TMX | TMN | ET | TMX |
5 | TMN | ET | TMN | TMP | SM | TMN | PRE | NDVI | TMP | TMP | TMX | SM | TMP | TMX | TMN | NDVI |
6 | ET | TMN | ET | PRE | ET | ET | NDVI | PRE | SM | SM | TMP | TMP | NDVI | TMP | NDVI | ET |
7 | SIF | SIF | NDVI | NDVI | SIF | NDVI | ET | ET | NDVI | NDVI | NDVI | NDVI | TMN | NDVI | TMP | TMP |
8 | NDVI | NDVI | SIF | SIF | NDVI | SIF | SIF | SIF | SIF | SIF | SIF | SIF | SIF | SIF | SIF | SIF |
重要性排序 | 沙漠气候 | 地中海气候 | 湿润亚热带气候 | 海洋性气候 | ||||||||||||
SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | |
1 | PRE | PRE | PRE | TMN | PRE | PRE | PRE | PRE | PRE | PRE | PRE | PRE | PRE | PRE | PRE | PRE |
2 | SM | SM | SM | SM | TMX | SM | TMX | TMN | TMX | TMX | TMX | TMX | TMX | ET | ET | TMX |
3 | ET | ET | ET | NDVI | TMN | TMN | SM | SM | SM | ET | ET | TMN | ET | TMX | TMX | ET |
4 | TMN | TMN | TMX | PRE | TMP | TMP | TMN | ET | NDVI | TMN | TMN | ET | NDVI | TMN | TMN | TMN |
5 | TMX | TMX | TMN | TMX | ET | ET | TMP | TMX | TMP | SM | SM | TMP | SIF | SM | SM | SM |
6 | TMP | TMP | TMP | ET | NDVI | TMX | ET | TMP | TMN | SIF | TMP | SM | TMN | TMP | NDVI | NDVI |
7 | NDVI | NDVI | NDVI | TMP | SM | NDVI | NDVI | NDVI | ET | NDVI | SIF | NDVI | TMP | SIF | TMP | TMP |
8 | SIF | SIF | SIF | SIF | SIF | SIF | SIF | SIF | SIF | TMP | NDVI | SIF | SM | NDVI | SIF | SIF |
重要性排序 | 湿润大陆性气候 | 亚寒带气候 | 苔原气候 | |||||||||||||
SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | SPI1 | SPI3 | SPI6 | SPI12 | |||||
1 | PRE | PRE | PRE | SM | PRE | PRE | PRE | PRE | PRE | PRE | PRE | PRE | ||||
2 | NDVI | SM | TMN | PRE | ET | ET | ET | ET | TMX | TMN | TMN | TMN | ||||
3 | SIF | TMX | SM | TMN | TMX | TMX | TMN | TMN | ET | TMX | TMX | TMX | ||||
4 | TMX | TMN | ET | ET | NDVI | TMN | TMX | TMX | TMN | ET | ET | ET | ||||
5 | ET | NDVI | TMX | TMX | TMN | SM | SM | SM | TMP | TMP | TMP | TMP | ||||
6 | TMP | ET | NDVI | NDVI | SM | NDVI | TMP | NDVI | SM | SM | SM | NDVI | ||||
7 | TMN | TMP | TMP | SIF | TMP | TMP | NDVI | TMP | NDVI | NDVI | NDVI | SM |
[1] |
|
[2] |
|
[3] |
United nations: Convention to combat desertification in those countries experiencing serious drought and/or desertification, particularly in africa[J]. International Legal Materials, 1994,33(5):1328-1382.
|
[4] |
|
[5] |
|
[6] |
王奕森, 夏树涛. 集成学习之随机森林算法综述[J]. 信息通信技术, 2018,12(1):49-55.
[
|
[7] |
|
[8] |
沈润平, 郭佳, 张婧娴, 等. 基于随机森林的遥感干旱监测模型的构建[J]. 地球信息科学学报, 2017,19(1):125-133.
[
|
[9] |
吴晶, 陈元芳, 余胜男. 基于随机森林模型的干旱预测研究[J]. 中国农村水利水电, 2016(11):17-22.
[
|
[10] |
|
[11] |
吴陶樱, 方秀琴, 吴小君, 等. 基于气候区的全球干旱形势分析[J]. 水土保持研究, 2019,26(3):166-173.
[
|
[12] |
|
[13] |
|
[14] |
马柱国, 符淙斌, 谢力, 等. 土壤湿度和气候变化关系研究中的某些问题[J]. 地球科学进展, 2001(4):563-568.
[
|
[15] |
|
[16] |
|
[17] |
张荣华, 杜君平, 孙睿. 区域蒸散发遥感估算方法及验证综述[J]. 地球科学进展, 2012,27(12):1295-1307.
[
|
[18] |
|
[19] |
孟晗, 黄远程, 史晓亮. 黄土高原地区2001-2015年植被覆盖变化及气候影响因子[J]. 西北林学院学报, 2019,34(1):211-217.
[
|
[20] |
|
[21] |
GB/T 20481-2017 气象干旱等级. 2017.
[ GB/T 20481-2017 Grades of meteorological drought. 2017.]
|
[22] |
|
[23] |
赖成光, 陈晓宏, 赵仕威, 等. 基于随机森林的洪灾风险评价模型及其应用[J]. 水利学报, 2015,46(1):58-66.
[
|
/
〈 | 〉 |