memo of Wiki.blogger

mem - 建立

魚狗 — Tue, 21 Feb 2012 21:47:38 -0700

《Numerical Ecology with R》Old version? p263

The PCNM method provides an elegant way of constructing sets of linearly independent spatial variables. Since its publication, it has gained a wide audience and has been applied in several research papers. But it is not the end of the story.

Dray et al. (2006) have greatly improved the mathematical formalism of PCNM analysis by showing that it is a particular case of a wider family of methods that they called Moran’s eigenvector maps (MEM). They demonstrated the link between the eigenvalues of the MEM eigenvectors and Moran’s spatial correlation index, I (7.3). They reasoned that the relationship among sites, which is the basis for any spatial eigenvector decomposition, actually has two components:

(1) a list of links among objects, represented by a connectivity matrix and (2) a matrix of weights to be applied to these links. In the simplest case, the weights are binary (i.e. either two objects are linked, or they are not). In more complex models, non-negative weights can be placed on the links; these weights represent the easiness of exchange (of organisms, energy, information, etc.) between the points connected by the links. For instance, link weights can be made to be inversely proportional to the squared Euclidean distance among sites.

Furthermore, Dray et al. (2006) showed that (1) by using similarities instead of distances among sites, (2) setting the relationship of the sites with themselves to null similarity and (3) avoiding a square-root standardization of the eigenvectors within the PCoA procedure, one obtains a family of flexible methods (MEM) that bear an immediate connexion with Moran’s I and can be modulated to optimize the construction of spatial variables. The MEM method produces n − 1 spatial variables with positive and negative eigenvalues, allowing the construction of a wide range of variables modelling positive and negative spatial correlation. The eigenvectors maximize Moran’s I index, the eigenvalues being equal to Moran’s I multiplied by a constant. Therefore, the spatial structures of the data are extracted in such a way that the axes first optimally display the positively autocorrelated structures in decreasing order of importance, and then the negatively autocorrelated structures in increasing order.

The MEM method consists in defining two matrices describing the relationships among the sites: A binary connectivity matrix B defining which pairs of sites are connected (1) and which are not (0) A weighting matrix A providing the intensity of the connexions

The final spatial weighting matrix W results from the Hadamard (i.e. term-by-term) product of these two matrices, B and A. The connectivity matrix B can be constructed on the basis of distances (by selecting a distance threshold and connecting all points that are within that distance) or by other connexion schemes, such as Delaunay triangulation, Gabriel graph or others (described by Legendre and Legendre 1998, Section 13.3). The connexion matrix can of course be customized to fit special needs – for instance, by only allowing connexions among sites along the littoral zone of a lake (not across water) or along the shoreline of an island.

Matrix A is not mandatory, but is often used to weight the connexions according to distance, e.g. by inverse distance or inverse squared distance, since it is ecologically realistic to assume that a process influences a community with an intensity decreasing with distance. The choice of both matrices is very important because it greatly affects the structure of the spatial variables obtained. These variables, in turn, condition the results of the spatial analysis, especially in the case of irregular sampling: “In the case of regular sampling (e.g. a regular grid), structures defined by eigenvectors are roughly similar for different definitions of W. For irregular distributions of sites, however, the number of positive/negative eigenvalues and the spatial structures described by their associated eigenvectors are greatly influenced by the spatial relationships defined in W” (Dray et al. 2006). These authors provide the following general recommendations:

The choice of the spatial weighting matrix W is the most critical step in spatial analysis. This matrix is a model of the spatial interactions recognized among the sites, all other interactions being excluded. In some cases, a theory-driven specification can be adopted, and the spatial weighting matrix can be constructed based upon biological considerations […]. In most situations, however, the choice of a particular matrix may become rather difficult and a data-driven specification could then be applied. Under this latter approach, the objective is to select a configuration of W that results in the optimal performance of the spatial model.

moran_s_i

魚狗 — Tue, 21 Feb 2012 18:44:52 -0700

spatial analysis,statistics in ecology

correlogram

魚狗 — Tue, 21 Feb 2012 18:21:37 -0700

sp.correlogram {spdep}

correlog {pgirmess}

Different Moran's I correlograms using correlog{ncf} and sp.correlogram{spdep} from [R-sig-Geo]

It's said….

“No, because correlog{ncf} bins the distances, but sp.correlogram{spdep} adds higher order lags - here the second lag of i (with first order neighbours j) is the set of all points k that are neighbours of points j and neither i nor in j. It depends on graph edge counts, not distance as such. They aren't comparable, really. If you want to do it by hand, check the distances reported by correlog() (they may not be what you think), take the bin boundaries, plug them into multiple calls to dnearneigh(), and go from there.

Roger ”

see also correlogram in 《Numerical Ecology with R》: Coefficients for the larger distance values (grey zones in correlograms) should not be considered in correlograms, nor interpreted, because they are based on a small number of pairs (test with low power) and only include the pairs of points bordering the series or surface.

autocorrelation

魚狗 — Tue, 21 Feb 2012 06:06:57 -0700

statistics in ecology see also spatial analysis, 空間迴歸模型

the autocorrelation of a series ryy(k) may be deﬁned as the ratio of its autocovariance syy(k) to its variance s2y = syy(0) 《Numerical Ecology with R》p655

stationary

魚狗 — Tue, 21 Feb 2012 06:06:47 -0700

statistics

At this stage of series analysis, it is assumed that the data series is stationary, either because it originally exhibited no trend or as the result of trend extraction(Section 12.2). It is also assumed that variability is large enough to emerge from random noise.

A general approach for analysing periodic variability is derived from the concepts of covariance and correlation, which were deﬁned in Chapter 4. The methods are called autocovariance and autocorrelation. The approach is to quantify the relationships between successive terms of the data series. These relationships reﬂect the pattern of periodic variability. 《Numerical Ecology with R》p653

青瞑蛇 - 建立

魚狗 — Sun, 19 Feb 2012 04:11:50 -0700

台南自然史

04‧【清代南瀛地區的海岸線變遷】/趙文榮

一、緒言

      清代南瀛地區的海岸線變遷，其變化的原因，除了西南部海岸的隆起之外，最為重要的關鍵，則是曾文溪與急水溪兩溪流沖刷搬運與堆積作用，直接造成臺江內海與倒風內海的陸化，從而改變清代臺南地區的海岸線，並進而影響了清代南瀛地區的產業經濟，也影響南瀛地區在全臺政經地位的變遷，因之，頗值一探，即嘗試由清代方志輿圖、港口數量等面向析論之。

二、清代方志輿圖上的海岸線變化

〈康熙輿圖〉中，可明顯看出南瀛內海的狀況，內海的港汊，由北而南分別是鐵線橋港、茅港尾港、麻豆港、蕭?、西港仔、洲仔尾、柴頭港、大橋港、小橋港；而港外的內海中，有一串如珍珠般、大小不一的沙洲，是內海的重要屏障，由北而南依序是北鯤身、南鯤身、網寮、北線尾、安平、一至七鯤身。

      清康熙56年（1717）〈諸羅縣輿圖〉中，蚊港是出入倒風內海的樞紐，往內海更裡面走，有一網狀的水路系統，汫水港、鹽水港都是海汊，比較使人側目的是由蚊港、汫水港、鹽水港以迄鐵線橋、茅港尾與麻豆港畫出一條陸線，陸線外是倒風水域，水域之西則是一隻長長宛若手臂的沙陸，範疇由南鯤身沙洲以南，往南經「內連椼」，經「佳里興」而後一直延伸至灣裡溪口。

‧〈康熙輿圖〉中，可顯明看出南瀛內海的狀況。（資料來源：國立臺灣博物館）

  清乾隆12年（1760）〈重修臺灣府志輿圖〉，則可以看到南瀛內海沙洲島嶼的變化，由北而南為「北門嶼」、「南鯤身」、「青峰闕」、「青鯤身」、「馬沙溝」;而臺江部分由北而南則有「海翁線」、「鹿耳門」、「隙仔」、「北線尾」、「安平」、「一鯤身」、「二鯤身」、「三鯤身」、「四鯤身」、「五鯤身」、「六鯤身」和「七鯤身」。

清乾隆25年（1760），〈續修臺灣府志輿圖〉可看出內海中的沙洲島「馬沙溝」，已與自佳里興方面延伸而來的手臂狀沙洲相連，此意謂著倒風內海陸浮的速度仍在加速進行中。

而〈乾隆輿圖〉中，鐵線橋港、茅港尾港、麻豆港，此時雖有港口之名，但其實已非臨海的海港，而已成為通往倒風內海的河港了；其次進入清乾隆時期，佳里興離海已有一定距離，倒風內海陸化可見，而臺江部分，則沙洲其由北而南依序是「北線尾」、「海翁線與隙仔線」、「鹿耳門」、「安平」、「一鯤身」、「二鯤身」、「三鯤身」、「四鯤身」、「五鯤身」、「六鯤身」、「七鯤身」，這些沙洲即是臺江西緣的屏藩，隔臺江與府城臺南相望，此時的輿圖，仍可看出維持其內海貿易鼎盛的榮況。

  由上之述，顯然清乾隆時期是倒風內海陸化的關鍵，一些原本是港汊的繁華港市，開始出現淤塞或距海漸遠之實，如茅港尾、鐵線橋、麻豆港，而鹽水港暫得以維持港貿；至於臺江，則須至清道光3年（1823）曾文溪大水之後，陸浮嚴重，府城各港、鹿耳門等街，遂失去其港貿功能，只有安平仍勉力維持，而臺江內海之經貿功能不再，南瀛經濟優勢亦不再。

三、港口數量的變化

港口數量變化的情況，也提供海岸線變遷的訊息，據林玉茹的研究，清康熙22年（1683）至清康熙49年（1710），全臺共出現64個港口，大多集中於南部，尤其八掌溪以南至二仁溪之間的港口最多，大約有23個，幾乎佔了全臺港口總數的一半，可見分佈極為密集，[1]林玉茹所說的港口地域範疇，則大部分皆位於南瀛地區，而且港口數目居全臺第一。

清康熙50年（1711）至清雍正8年（1730），臺江內海有19個港口，倒風內海約有20個港口，合計39個港口，[2] 約佔南部港口數的88%，全臺港口數的45%，亦居全臺港口數第一。

  清雍正9年（1731）至清乾隆48年（1783），八掌溪以南至二仁溪之間，港口依然密集，但是臺江與倒風內海的海岸線逐漸出現變化，開始出現有港口淤塞而消失的現象，港口數有逐漸下降的趨勢；[3]相較之下，此一時期的中北部，則因墾務與人口增加的發展，港口數有增加的趨勢。

‧倒風內海的海岸線出現變化，最後退縮為今日的北門潟湖。（黃文博/攝）

清乾隆49年（1784）至清道光10年（1830），此一時期是南瀛地區港口數與港口功能消退的關鍵時期，倒風內海逐漸消失，其內海沿岸港口有的完全失去港灣功能，有的由海港變為河港；而臺江內海則因清道光3年（1823）7月的一場大風雨，而逐漸浮淺，影響所及，港口功能性大受影響。[4]

清道光11年（1831）至清咸豐10年（1860），南部港口大量減少，尤以八掌溪以南至二仁溪之間為最，港口減少了12個；而此一時期的北部，港口數量卻逐漸增加，呈現出南消北長的趨勢。[5]這說明了清道光時期是南瀛海岸線另一個變化的高峰期。

四、結語－「青瞑龍」與「青瞑蛇」

清代南瀛海岸變化與曾文溪及急水溪難脫關係，而其氾濫成災則又與漢人拓墾有關，移民溯溪往上墾殖，火燒除草，伐木整地，上游之水土保持不易，加上臺南地區夏秋氣候多水災、風災，往往容易釀禍，因之，當談及清代南瀛海岸線的變遷，雨季時反覆無常而動輒成災的曾文溪及急水溪，也就成為臺南地區移民的夢魘。

無奈中，先民不免也自我解嘲一番，比如倒風內海地域內的人們，即暱稱急水溪為「青瞑龍」，意思就是說急水溪不長眼睛、亂流。而使臺江陸化的曾文溪，其「青瞑蛇」的稱號，也是不遑多讓，威名直使其流域內的居民膽寒，歷史上曾文溪河道變化之大之廣，[6]令人咋舌，每逢夏秋雨季，一翻身，往往氾濫成災，出海口時常變動，時或偏南，時或偏北，先民給他「活溪」之名，改道與成災之態勢驚人，清道光3年（1823） 7月臺灣大風雨，這尾「青瞑蛇」改道臺江內海，洪流挾帶內山崩陷的泥沙，也使臺江為之淤塞。

故而，今日談及清代的海岸線變遷，「青瞑龍」與「青瞑蛇」當是關鍵，他們是創造者也是終結者，他們使港汊河海運的功能消失，使繁榮的港汊變寒村，於是乎他們是清代南瀛內海繁榮的終結者；但另一方面，他們創造的則是新的海岸線、新的墾殖田地，因之，其功過留給歷史去評判吧！

【參考資料】

‧趙文榮，2007年，《南瀛內海誌》。新營：臺南縣文化局。

‧林玉茹，1996年，《清代臺灣港口的空間結構》。臺北：知書房。

1 林玉茹，《清代臺灣港口的空間結構》（臺北：知書房，1996年），頁39。

2 林玉茹，《清代臺灣港口的空間結構》（臺北：知書房，1996年），頁40。

3 林玉茹，《清代臺灣港口的空間結構》（臺北：知書房，1996年），頁45~47。

4 林玉茹，《清代臺灣港口的空間結構》（臺北：知書房，1996年），頁52。

5 林玉茹，《清代臺灣港口的空間結構》（臺北：知書房，1996年），頁52~55。

6.北到將軍鄉，南到臺南市安南區。

蟹工船 - 建立

魚狗 — Sat, 11 Feb 2012 08:30:59 -0700

人權, 勞工, 資本主義, canon camera

平成年間的蟹工船--沒有椅子的佳能電子

作者雷路許的小小讀書室 blog中寫：「《蟹工船》，是日本知名社會主義文學家小林多喜二在二戰前寫成的一部作品，故事內容是敘述一群勞工在捕蟹船上受到種種不人道的壓榨，最後起而反抗的故事。因為這個故事把勞工受壓迫的情狀寫得太生動了，所以從此以後，日本人都稱壓榨勞工的所謂「血汗工廠」為「蟹工船」。照理說，蟹工船的故事距離今日已有七八十年之久，社會對於勞工的保障也應該有所進步才對，

然而令人驚奇的是，在日本竟然有一間以壓榨員工自豪，被人怒稱為「平成蟹工船」的電子公司，

而且這家公司還是鼎鼎大名的相機製造商–佳能(KANON)的子公司?」…. 詳見原聯結

mixed_effects - 建立

魚狗 — Sun, 05 Feb 2012 08:00:17 -0700

statistics, linear regression modelling

see Some Notes on Mixed Models

“A mixed-effects model or simply a mixed model is a model that includes a mixture of fixed and random factors. Recall that each factor in an experiment has levels. The effects associated with a factor are the effects that the levels of the factor have on the response variable of interest.

Fixed vs. Random

Generally speaking a factor is fixed if the levels of the factor were selected by the investigator with the purpose of comparing the effects of the levels to one another. One of the major goals of the analysis is to test for differences among the effects associated with the specifically chosen levels of the factor and to describe the specific differences that exist.

A factor is random if the effects associated with the levels of the factor can be viewed as being like a random sample from a population of effects. For random effects, we can make statements about variation in the population of random effects from which the effects at hand are considered to be like a random sample. Furthermore, we can generalize our conclusions about fixed factors to the populations associated with random factors. We are usually not interested in comparisons among the levels of random effects. Rather, we are interested in studying variation in the population from which the random effects are like a random sample or in controlling for that variation so that proper conclusions about fixed effects can be drawn.

An interaction between or among factors is considered to be random if any one of the factors involved in the interaction is random.”

spatial_analysis

魚狗 — Sat, 04 Feb 2012 17:17:38 -0700

spatial analysis

Spatial Analysis Introduction

CH2 空間統計介紹.pdf

CH3 群集模型.pdf

CH4 空間迴歸模型.pdf(Spatial Regression)

CH4 中有一段(by 余清祥)： “在配適空間迴歸模型之過程中，我們可先以傳統迴歸之模型建立方法，找出影響大尺度變化之重要變數，再以空間相關檢定（Moran’s I 或 Geary’s C），檢定其殘差是否存在空間相關性，若檢定結果顯示不具空間相關性，且其他部分皆符合傳統迴歸之基本假設，則不需再配適空間迴歸模型；若檢定結果具有空間相關性，則表示模型明顯不符合傳統迴歸中獨立之基本假設，可再依傳統迴歸模型所挑選出重要解釋變數之部分，再進一步配適空間迴歸模型，將殘差中之空間相關性納入最終模型中，使殘差符合獨立之基本假設，完成空間迴歸模型之建立。” (see also GLS)

空間迴歸模型

魚狗 — Sat, 04 Feb 2012 15:44:45 -0700

literature to read about spatial autocorrelation(SAC) Methods to account for spatial autocorrelation in the analysis of species distributional data: a review Dormann et al., 2007

Moran's I correlogram by Legendre and Legendre 1998

Geary's c correlograms and semi-variograms

(in the book «Mixed effects models and extensions in ecology with R», p491) …” The use of GLMM was an effective way of dealing with spatial autocorrelation in the data, but this may not always be the case, such as if spatial autocorrelation existed between sites. However, other approaches, such as autoregressive models, do exist that could be used to deal with between-site auto-correlation (e.g., Miller et al., 2007). ”

to be conti.

zero-inflated

魚狗 — Fri, 03 Feb 2012 19:36:34 -0700

statistics, literature

A nice overview and comparison of Poisson, NB,and zero-inﬂated models in R is given in Zeileis et al. (2008). This paper also gives a couple of useful references to publications using mixture and two-part models.

poisson

魚狗 — Fri, 03 Feb 2012 19:32:21 -0700

statistics, (in the book « Mixed effects models and extensions in ecology with R»)

the Poisson probability distribution (in Fig. 8.2D) looks like a normal distribution, it is not equal to a Normal distribution; a Normal distribution has two parameters (the mean μ and the variance σ 2 ),whereas a Poisson distribution only uses one parameter μ (which is the mean and the variance).

The Poisson distribution is typically used for count data, and its main advantages are that the probability for negative values is 0 and that the mean variance relationship allows for heterogeneity. However, in ecology, it is quite common to have data for which the variance is even larger than the mean, and this is called overdispersion. Depending how much larger the variance is compared to the mean, one option is to use the correction for overdispersion within the Poisson GLM(, and this is discussed in Chapter 9). Alternatively, we may have to choose a different distribution, e.g. the negative binomial distribution.

statistics

魚狗 — Fri, 03 Feb 2012 17:48:31 -0700

古典統計專注在Normal distribution X~N(u,var),u: mean = mode = median in Normal distrib.: location parameter; var = sigma square: scale parameter。古典統計問： 1. u, var = ? 2. 如何用sample 推論 population

針對每一個Q (theta, general parameter如u,var)找到一個estimator ~Q(hat)來estimate Q ~Q = f(x1,x2,…xn), function of data = statistics ps : 3. What is good estimator?

… 待續

normal_distribution - 建立

魚狗 — Fri, 03 Feb 2012 17:41:23 -0700

statistics

… <待續>

gls

魚狗 — Fri, 03 Feb 2012 16:47:55 -0700

regression model in statistics, mixed effects modelling

in the book «Mixed effects models and extensions in ecology with R»

In all these chapters, the model consists of a ﬁxed term and a random term. The ﬁxed term describes the response variable Y as a function of the explanatory variables via α + β 1 × X 1 + . . . + β q × X q in linear regression or α + f 1 (X 1 )+ . . . + f q (X q ) in additive modelling. This part of the model is described in Appendix A and Chapter 3.

The random part contains components that allow for heterogeneity, nested data (random effects), temporal correlation, spatial correlation, and a real random term. It is also possible to have a combination of these components.

If the random part only contains the real random term, we are back to linear regression or additive modelling. If it allows for nested data, the resulting model is called a mixed effects model. If it only allows for heterogeneity, we call it a generalised least squares (GLS) model. This is essentially a weighted linear regression.

….

In linear regression and additive modelling, we use the Normal (or: Gaussian) distribution. It is important to realise that this distribution applies for the response variable. GLM and GAM are extensions of linear and additive modelling in the sense that a non-Gaussian distribution for the response variable is used and the relationship (or link) between the response variable and the explanatory variables may be different.