@article{oai:mie-u.repo.nii.ac.jp:00015058, author = {葛葉, 泰久 and KUZUHA, Yasuhisa and 水木, 千春 and MIZUKI, Chiharu}, issue = {第5号}, journal = {水文・水資源学会誌, Japan Society of Hydrology and Water Resources}, month = {Sep}, note = {application/pdf, 国土交通省をはじめ,多くの行政機関が河川計画策定時に「中小河川計画の手引き(案)」という資料を用いている.しかし,ここ数年著者らが指摘しているように,この手引きのT年確率水文量算定の手法を表すフローチャートには重大な誤りがある.特に,本邦で長く使われてきたSLSCについて,標本数・確率分布に関してフェアでないことが問題となろう.そこで本稿では,従来の方法からそれほど大きく変わらない手続きを提案する.つまり, 1) 何らかの手法でいくつかの分布関数の母数推定を行う. 2) それぞれの分布関数に関してSLSCを求める. 3) モンテカルロ・シミュレーションによって生成させた乱数を用いてSLSCを多数発生させ,それぞれの確率分布について,SLSCの分布関数を求める. 4) 最初に求めたそれぞれの分布のSLSCの非超過確率を求め,それが小さい(つまり「より有意」,すなわち確率密度関数の「より左の裾」にある)ものを「優秀な分布」と考える. というようなものである. この手法でd4PDF過去実験データの年最大1時間降水量の,最適な確率分布を選定したところ,SLSCそのものを用いる場合と,本稿で提案する手法を用いる場合では,結果が若干異なることが分かった.著者らはこの手法により,よりフェアな適合度評価ができると考える., We propose a method for selecting an optimal stochastic distribution that can be used along with hydrologically extreme data. In Japan, many civil engineering departments of governmental organizations refer to the “Guide for River Plan Design for Small and Medium-sized Rivers.” However, the flow chart for estimating T-year hydrological events included in guide includes important defects. For instance, this guide recommends the standard least squares criterion (SLSC) method for estimating the goodness of fit of each distribution. Some researchers have pointed out that SLSC is not a fair criterion. Our proposed method uses not SLSC itself, but the degree of significance: specifically the probability of non-exceedance of SLSC. We propose the following procedures. 1) Estimating parameters of population for various stochastic distributions 2) Estimating SLSC (referring to the “original SLSC”) of each distribution 3) Running a Monte Carlo simulation, which generates various random numbers with estimated distributions and parameters 4) Estimating various SLSCs with generated random numbers and estimating SLSC distributions for each distribution 5) Evaluating the probability of non-exceedance of the “original SLSC” by comparison to the SLSC distribution 6) Comparing the probability of non-exceedance of SLSC for each distribution and selecting the most appropriate distribution for which the probability is smallest Results indicate that the optimal distribution selected using our new method is sometimes different from the distribution selected when using SLSC. Data used for this study were one-hour precipitation data calculated by the d4PDF project. Results suggest that the modified method is superior to the conventional method.}, pages = {283--302}, title = {確率水文量算定手法の改良と従来からの手法の問題点指摘-修正SLSC 法を含む手法-}, volume = {第34巻}, year = {2021}, yomi = {クズハ, ヤスヒサ and ミズキ, チハル} }