#pragma GCC optimize ("O3")
#pragma GCC optimize ("unroll-loops")
//#pragma GCC target("sse,sse2,sse3,ssse3,sse4,popcnt,abm,mmx,avx,tune=native")
#include <bits/stdc++.h>
typedef long long ll;
typedef long double ld;
using namespace std;
const int N = 50 + 5;
const int M = 6e3 + 5;
int n, m, r, s, mn = 1e9;
ld d[N][N][M];
char a[N][N];
int sqr(int x) {
	return x * x;
}
ld dij(int i, int j) {
	priority_queue<pair<ld, array<int, 3>>> q;
	for(int i = 0;i < n;i++) {
		for(int j = 0;j < m;j++) {
			for(int k = 0;k <= s;k++)
				d[i][j][k] = 1e9;
		}
	}
	int c = a[i][j] - '0';
	d[i][j][c] = 0;
	q.push({0, {c, i, j}});
	while(!q.empty()) {
		ld dis = -q.top().first;
		int curi = q.top().second[1];
		int curj = q.top().second[2];
		int curs = q.top().second[0];
		q.pop();
		if(d[curi][curj][curs] < dis)
			continue;
		if(curi == mn)
			return dis;
		for(int k = curi;k >= 0;k--) {
			for(int p = 0;p < m;p++) {
				if(a[k][p] == '.')
					continue;
				int ndis = sqr(k - curi) + sqr(p - curj);
				if(ndis > sqr(r))
					continue;
				int c = a[k][p] - '0';
				if(c + curs > s)
					continue;
				ld cdis = dis + sqrt(ndis);
				int cs = c + curs;
				if(cdis < d[k][p][cs]) {
					d[k][p][cs] = cdis;
					q.push({-d[k][p][cs], {cs, k, p}});
				}
			}
		}
	}
	return 1e15;
}
int main() {
    ios_base::sync_with_stdio(0);cin.tie(0);cout.tie(0);
    cin >> n >> m >> r >> s;
    for(int i = 0;i < n;i++) {
		for(int j = 0;j < m;j++) {
			cin >> a[i][j];
			if(a[i][j] != '.')
				mn = min(mn, i);
		}
	}
	for(int i = n - 1;i >= 0;i--) {
		for(int j = 0;j < m;j++) {
			if(a[i][j] != '.') {
				ld ans = dij(i, j);
				if(ans == 1e15)
					cout << "impossible";
				else
					cout << fixed << setprecision(9) << ans << endl;
				return 0;
			}
		}
	}
    return 0;
}